igneum/site/miners.html
igneum-labs dda5dedf44 Site reset to Igneum 2.0, first pass: /bench and /journey gone, no testnet, no 0.x history, the Igneum 2.0 devnet, the positioning line
- build.mjs no longer writes bench.html, journey.html or journey.json; the three files are deleted; /bench -> /evidence and
  /journey -> /live redirects (permanent: false); nav, footer, sitemap, capture routes and the site-nav-check groups drop them;
  link-check resolves literal vercel.json redirects and lists every link that leans on one (evidence.html: /bench, another hand)
- every testnet card, column, row, notice and link removed from the pages this lane owns; release manifest: the testnet entry
  gone, network igneum-devnet-4 "Igneum 2.0 devnet", the Devnet 3 genesis, digest, vote domain and floor lines dropped
- "Devnet 3" leaves every owned page; the devnet state is one constant in build.mjs (DEVNET_STATE, 'starting' now: the nav
  pill and the homepage badge read "Igneum 2.0 devnet starting"; 'live' flips both and re-enables the nav's live read)
- no 0.x release history: screenshot alt texts say the current build, the miner page's "Current build:" keeps the data-rm
  spans without the node version; miner-bench.json keeps the current-program rows only, no Counter ASIC names, no versions;
  /miners drops the earlier-classes table, the history note and the miner-version column
- /income keeps the calculator; the testnet-era tier tables (docs/analysis/income-tiers.md) are not served
- the positioning line is the first line of the footer partial (every page), the homepage meta description's first sentence
  and the hero's sub-line; share metas (site/og/pages.mjs) carry no Devnet 3 or zkEVM
- ledger-text-check is red on two index.html pins (X2, X31: the testnet sentences), owned by the ledger hand

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-08 15:40:33 +00:00

402 lines
No EOL
153 KiB
HTML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, viewport-fit=cover">
<title>Igneum GPU bench table</title>
<meta name="description" content="Measured Igneum hash rates per GPU on the current program: card, best MH/s, MH per wall watt where measured, date and the record each number came from.">
<link rel="canonical" href="https://igneum.network/miners">
<meta name="theme-color" content="#0C0C0E">
<!-- share:start -->
<meta property="og:type" content="website">
<meta property="og:site_name" content="Igneum">
<meta property="og:title" content="Igneum GPU bench table">
<meta property="og:description" content="Measured hash rates per card, MH per watt where measured, and the record behind each number.">
<meta property="og:url" content="https://igneum.network/miners">
<meta property="og:image" content="https://igneum.network/og/home.png?v=4">
<meta property="og:image:type" content="image/png">
<meta property="og:image:width" content="1200">
<meta property="og:image:height" content="630">
<meta property="og:image:alt" content="The GPU-mined chain that proves every block. igneum.network">
<meta property="og:image" content="https://igneum.network/og/home-square.png?v=4">
<meta property="og:image:type" content="image/png">
<meta property="og:image:width" content="1200">
<meta property="og:image:height" content="1200">
<meta property="og:image:alt" content="The GPU-mined chain that proves every block. igneum.network">
<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:title" content="Igneum GPU bench table">
<meta name="twitter:description" content="Measured hash rates per card, MH per watt where measured, and the record behind each number.">
<meta name="twitter:image" content="https://igneum.network/og/home.png?v=4">
<meta name="twitter:image:alt" content="The GPU-mined chain that proves every block. igneum.network">
<!-- share:end -->
<link rel="icon" href="data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 1024 1024'%3E%3Crect width='1024' height='1024' fill='%230C0C0E'/%3E%3Cg transform='translate(166.95 166.95) scale(6.901)'%3E%3Cpolygon points='50,4 74,34 67,58 80,54 61,96 39,96 20,54 33,58 26,34' fill='%23F2541B'/%3E%3Cpolygon points='50,42 59,58 50,82 41,58' fill='%230C0C0E'/%3E%3C/g%3E%3C/svg%3E" type="image/svg+xml">
<link rel="icon" href="/favicon.ico" sizes="48x48">
<link rel="icon" href="/favicon-32.png" type="image/png" sizes="32x32">
<link rel="icon" href="/icon-192.png" type="image/png" sizes="192x192">
<link rel="icon" href="/icon-512.png" type="image/png" sizes="512x512">
<link rel="apple-touch-icon" href="/apple-touch-icon.png" sizes="180x180">
<link rel="manifest" href="/site.webmanifest">
<!-- head:start -->
<link rel="preload" href="/fonts/unbounded-900.woff2" as="font" type="font/woff2" crossorigin>
<link rel="preload" href="/fonts/plex-sans-400.woff2" as="font" type="font/woff2" crossorigin>
<meta name="color-scheme" content="dark light">
<link rel="stylesheet" href="/site.css">
<style>
/* Shared by every page: the head partial, injected by the build between the head markers. Edit the partial, not the page. */
/* Fonts, self-hosted (latin subsets, OFL): Unbounded 500/700/900, IBM Plex Sans 400/500/600, IBM Plex Mono 400/500. Fallbacks carry size and ascent overrides so the swap does not move the layout. The tokens, type and components are in /site.css (docs/plans/site-ui-3.md). */
@font-face{font-family:'Unbounded';font-style:normal;font-weight:500;font-display:swap;src:url(/fonts/unbounded-500.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'Unbounded';font-style:normal;font-weight:700;font-display:swap;src:url(/fonts/unbounded-700.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'Unbounded';font-style:normal;font-weight:900;font-display:swap;src:url(/fonts/unbounded-900.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'IBM Plex Sans';font-style:normal;font-weight:400;font-display:swap;src:url(/fonts/plex-sans-400.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'IBM Plex Sans';font-style:normal;font-weight:500;font-display:swap;src:url(/fonts/plex-sans-500.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'IBM Plex Sans';font-style:normal;font-weight:600;font-display:swap;src:url(/fonts/plex-sans-600.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'IBM Plex Mono';font-style:normal;font-weight:400;font-display:swap;src:url(/fonts/plex-mono-400.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'IBM Plex Mono';font-style:normal;font-weight:500;font-display:swap;src:url(/fonts/plex-mono-500.woff2) format('woff2');unicode-range:U+0000-00FF,U+0131,U+0152-0153,U+02BB-02BC,U+02C6,U+02DA,U+02DC,U+0304,U+0308,U+0329,U+2000-206F,U+20AC,U+2122,U+2191,U+2193,U+2212,U+2215,U+FEFF,U+FFFD}
@font-face{font-family:'Unbounded Fallback';src:local('Arial Black'),local('Arial-Black'),local('Impact');size-adjust:114%;ascent-override:87.5%;descent-override:21.5%;line-gap-override:0%}
@font-face{font-family:'Plex Sans Fallback';src:local('Arial'),local('Helvetica Neue'),local('Helvetica');size-adjust:100.5%;ascent-override:102%;descent-override:27.4%;line-gap-override:0%}
@font-face{font-family:'Plex Mono Fallback';src:local('Courier New'),local('Menlo');size-adjust:100%;ascent-override:102.5%;descent-override:27.5%;line-gap-override:0%}
</style>
<script>
/* theme before first paint: ?theme=light|dark for one view (the capture tool), else the stored choice under igneum-theme, else the system setting; ?motion=off marks the page data-motion=off so every .reveal is drawn at once (a headless render never scrolls) */
(function(){var t=null;try{t=new URLSearchParams(location.search).get("theme");}catch(e){}if(t!=="light"&&t!=="dark"){try{t=localStorage.getItem("igneum-theme");}catch(e){t=null;}}if(t!=="light"&&t!=="dark")t=window.matchMedia&&window.matchMedia("(prefers-color-scheme: light)").matches?"light":"dark";document.documentElement.setAttribute("data-theme",t);try{if(new URLSearchParams(location.search).get("motion")==="off")document.documentElement.setAttribute("data-motion","off");}catch(e){}})();
</script>
<!-- head:end -->
<style>
/* generated-page block (the engineering log, the bench table, the provenance table): the page hero, the contents rail, the sections; tokens, type and components are in /site.css */
[id]{scroll-margin-top:120px}
article h1{display:none}
article{color:var(--ink-2)}
article h2{font-size:var(--t-h3);margin:0 0 12px}
article h3{font-size:16px;margin:26px 0 8px}
article h4{font:600 15px/1.4 var(--f-sans);margin:20px 0 6px;color:var(--bone)}
article p{margin:0 0 14px}article ul,article ol{margin:0 0 14px;padding-left:22px}article li{margin-bottom:6px}
article p,article li,article h2,article h3{overflow-wrap:anywhere}
article pre{margin:0 0 var(--s-4)}
.doc-section:last-child{border-bottom:0}
.doc-section[hidden]{display:none}
.tbl{margin:var(--s-4) 0 var(--s-5)}
table{min-width:560px}
.gen{color:var(--ash);font-size:13px;margin:40px 0 0;max-width:72ch}
.doc-tools{margin:0 0 var(--s-5)}
.doc-tools .searchbar{max-width:560px}
</style>
</head>
<body>
<!-- nav:start -->
<a class="skip" href="#main">Skip to content</a>
<nav class="nav" aria-label="Main" data-nav-version="5">
<div class="container nav-row">
<a href="/" class="logo" aria-label="Igneum home">
<svg class="brand-mark" viewBox="0 0 1024 1024" aria-hidden="true"><rect width="1024" height="1024" rx="160" fill="#0C0C0E"></rect><g transform="translate(166.95 166.95) scale(6.901)"><polygon points="50,4 74,34 67,58 80,54 61,96 39,96 20,54 33,58 26,34" fill="#F2541B"></polygon><polygon points="50,42 59,58 50,82 41,58" fill="#0C0C0E"></polygon></g></svg>
<span class="word">IGNEUM</span>
</a>
<span class="devnet" id="nav-devnet" title="igneum-devnet-4, the Igneum 2.0 devnet, chain id 4464. Coins have no value. The chain may reset."><span class="dot off" id="foot-dot"></span><span id="foot-state" data-devnet-state="starting">Igneum 2.0 devnet starting</span></span>
<div class="nav-groups" id="nav-groups" role="list">
<div class="nav-group" role="listitem"><button type="button" class="group-btn" id="nav-btn-mine" data-group="mine" data-active aria-expanded="false" aria-controls="nav-panel-mine" aria-haspopup="true">Mine<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="m7 10 5 5 5-5"/></svg></button></div>
<div class="nav-group" role="listitem"><button type="button" class="group-btn" id="nav-btn-network" data-group="network" aria-expanded="false" aria-controls="nav-panel-network" aria-haspopup="true">Network<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="m7 10 5 5 5-5"/></svg></button></div>
<div class="nav-group" role="listitem"><button type="button" class="group-btn" id="nav-btn-learn" data-group="learn" aria-expanded="false" aria-controls="nav-panel-learn" aria-haspopup="true">Learn<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="m7 10 5 5 5-5"/></svg></button></div>
<div class="nav-group" role="listitem"><button type="button" class="group-btn" id="nav-btn-build" data-group="build" aria-expanded="false" aria-controls="nav-panel-build" aria-haspopup="true">Build<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="m7 10 5 5 5-5"/></svg></button></div>
</div>
<div class="nav-controls">
<button class="btn icon-btn" type="button" data-theme-toggle aria-label="Switch to light theme"><svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><circle cx="12" cy="12" r="4"/><path d="M12 2v2m0 16v2M2 12h2m16 0h2M5 5l1 1m12 12 1 1M5 19l1-1M18 6l1-1"/></svg></button>
<a href="/download" class="btn primary small cta" data-nav="download">Download</a>
<button class="burger" id="nav-burger" type="button" aria-expanded="false" aria-controls="nav-sheet" aria-label="Open the menu" aria-haspopup="true"><svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path class="bars" d="M4 6h16M4 12h16M4 18h16"/><path class="cross" d="m6 6 12 12M6 18 18 6"/></svg></button>
</div>
</div>
<div class="panels" id="nav-panels">
<div class="panel-wrap container">
<div class="nav-panel" id="nav-panel-mine" data-group="mine" role="region" aria-labelledby="nav-btn-mine" hidden>
<div class="panel-list">
<a href="/miner" data-nav="miner"><b>Ember, the miner</b><span>One click installs the node, the miner and the prover.</span></a>
<a href="/download" data-nav="download"><b>Download</b><span>Windows, macOS, Linux and HiveOS.</span></a>
<a href="/app" data-nav="app"><b>The app</b><span>What you see while the card mines.</span></a>
<a href="/wallet" data-nav="wallet"><b>The wallet</b><span>Your coins. Final means final.</span></a>
<a href="/miners" data-nav="miners" aria-current="page"><b>GPU bench table</b><span>Measured rates, card by card.</span></a>
<a href="/dev-fee" data-nav="dev-fee"><b>The dev fee</b><span>The optional one, in full view, off with one flag.</span></a>
<a href="/metamask" data-nav="metamask"><b>Add to MetaMask</b><span>Igneum as a network in your wallet.</span></a>
</div>
</div>
<div class="nav-panel" id="nav-panel-network" data-group="network" role="region" aria-labelledby="nav-btn-network" hidden>
<div class="panel-list">
<a href="/live" data-nav="live"><b>Live devnet</b><span>The chain as it grows, one lane per miner.</span></a>
<a href="/explorer" data-nav="explorer"><b>Explorer</b><span>Look a block or an address up.</span></a>
<a href="/evidence" data-nav="evidence"><b>Evidence</b><span>Every claim and how far it is tested.</span></a>
<a href="/light" data-nav="light"><b>Light wallet</b><span>A balance your browser proves, trusting no node.</span></a>
<a href="/receipt" data-nav="receipt"><b>Payment receipt</b><span>A receipt any third party re-verifies offline.</span></a>
<a href="/oracle" data-nav="oracle"><b>State oracle on Sepolia</b><span>Balances on the devnet, read from another chain.</span></a>
</div>
</div>
<div class="nav-panel" id="nav-panel-learn" data-group="learn" role="region" aria-labelledby="nav-btn-learn" hidden>
<div class="panel-list">
<a href="/litepaper" data-nav="litepaper"><b>Litepaper</b><span>The design, as published.</span></a>
<a href="/income" data-nav="income"><b>Income</b><span>What your card would mine, from the live network.</span></a>
<a href="/economics" data-nav="economics"><b>Economics</b><span>The emission, the split and the fees, each with its line in the node.</span></a>
<a href="/ledger" data-nav="ledger"><b>Ledger</b><span>Every criticism, answered or conceded.</span></a>
<a href="/claims" data-nav="claims"><b>What Igneum does not claim</b><span>The limits, stated first.</span></a>
<a href="/randomx" data-nav="randomx"><b>Igneum vs RandomX</b><span>What was kept and what was rebuilt for GPUs.</span></a>
<a href="/provenance" data-nav="provenance"><b>Built on the shoulders</b><span>Every borrowed part, credited.</span></a>
</div>
</div>
<div class="nav-panel" id="nav-panel-build" data-group="build" role="region" aria-labelledby="nav-btn-build" hidden>
<div class="panel-list">
<a href="/build" data-nav="build"><b>Build on Igneum</b><span>Chain ids, RPC, a contract in five minutes.</span></a>
<a href="/faucet" data-nav="faucet"><b>Devnet faucet</b><span>10 IGN of devnet coin, once a day.</span></a>
<a href="/swap" data-nav="swap"><b>Swap</b><span>Test tokens on the devnet; every swap is a proven block.</span></a>
<a href="/grants" data-nav="grants"><b>Grants</b><span>Tooling, apps, infrastructure, research.</span></a>
</div>
</div>
</div>
</div>
<div class="nav-sheet" id="nav-sheet" role="dialog" aria-modal="true" aria-label="Menu" hidden>
<div class="sheet-scroll">
<div class="sheet-group"><div class="sheet-head">Mine</div>
<a href="/miner" data-nav="miner"><b>Ember, the miner</b><span>One click installs the node, the miner and the prover.</span></a>
<a href="/download" data-nav="download"><b>Download</b><span>Windows, macOS, Linux and HiveOS.</span></a>
<a href="/app" data-nav="app"><b>The app</b><span>What you see while the card mines.</span></a>
<a href="/wallet" data-nav="wallet"><b>The wallet</b><span>Your coins. Final means final.</span></a>
<a href="/miners" data-nav="miners" aria-current="page"><b>GPU bench table</b><span>Measured rates, card by card.</span></a>
<a href="/dev-fee" data-nav="dev-fee"><b>The dev fee</b><span>The optional one, in full view, off with one flag.</span></a>
<a href="/metamask" data-nav="metamask"><b>Add to MetaMask</b><span>Igneum as a network in your wallet.</span></a>
</div>
<div class="sheet-group"><div class="sheet-head">Network</div>
<a href="/live" data-nav="live"><b>Live devnet</b><span>The chain as it grows, one lane per miner.</span></a>
<a href="/explorer" data-nav="explorer"><b>Explorer</b><span>Look a block or an address up.</span></a>
<a href="/evidence" data-nav="evidence"><b>Evidence</b><span>Every claim and how far it is tested.</span></a>
<a href="/light" data-nav="light"><b>Light wallet</b><span>A balance your browser proves, trusting no node.</span></a>
<a href="/receipt" data-nav="receipt"><b>Payment receipt</b><span>A receipt any third party re-verifies offline.</span></a>
<a href="/oracle" data-nav="oracle"><b>State oracle on Sepolia</b><span>Balances on the devnet, read from another chain.</span></a>
</div>
<div class="sheet-group"><div class="sheet-head">Learn</div>
<a href="/litepaper" data-nav="litepaper"><b>Litepaper</b><span>The design, as published.</span></a>
<a href="/income" data-nav="income"><b>Income</b><span>What your card would mine, from the live network.</span></a>
<a href="/economics" data-nav="economics"><b>Economics</b><span>The emission, the split and the fees, each with its line in the node.</span></a>
<a href="/ledger" data-nav="ledger"><b>Ledger</b><span>Every criticism, answered or conceded.</span></a>
<a href="/claims" data-nav="claims"><b>What Igneum does not claim</b><span>The limits, stated first.</span></a>
<a href="/randomx" data-nav="randomx"><b>Igneum vs RandomX</b><span>What was kept and what was rebuilt for GPUs.</span></a>
<a href="/provenance" data-nav="provenance"><b>Built on the shoulders</b><span>Every borrowed part, credited.</span></a>
</div>
<div class="sheet-group"><div class="sheet-head">Build</div>
<a href="/build" data-nav="build"><b>Build on Igneum</b><span>Chain ids, RPC, a contract in five minutes.</span></a>
<a href="/faucet" data-nav="faucet"><b>Devnet faucet</b><span>10 IGN of devnet coin, once a day.</span></a>
<a href="/swap" data-nav="swap"><b>Swap</b><span>Test tokens on the devnet; every swap is a proven block.</span></a>
<a href="/grants" data-nav="grants"><b>Grants</b><span>Tooling, apps, infrastructure, research.</span></a>
</div>
<div class="sheet-foot">
<a href="/download" class="btn primary big" data-nav="download">Download</a>
<div class="sheet-social"><a href="https://discord.gg/igneum" target="_blank" rel="noopener">Discord</a><a href="https://git.igneum.network/igneum-network/spec" target="_blank" rel="noopener">Source</a><a href="https://www.reddit.com/user/Igneum_network/" target="_blank" rel="noopener">Reddit</a></div>
</div>
</div>
</div>
</nav>
<script>
(function(){
// the slim bar (site-ui-5, 7 Oct 2026): three group buttons open one panel each; the burger opens the sheet on phones.
// Keyboard: Enter or Space opens a group, Escape closes it and returns focus to its button, arrows move between the
// group buttons, Tab walks the open panel's links, a click outside closes. The theme toggle flips data-theme and stores it.
var n=document.querySelector('.nav');if(!n)return;
var btns=[].slice.call(n.querySelectorAll('.group-btn')),panels=[].slice.call(n.querySelectorAll('.nav-panel')),wrap=n.querySelector('.panels');
var burger=document.getElementById('nav-burger'),sheet=document.getElementById('nav-sheet');
function panelOf(g){return panels.filter(function(p){return p.getAttribute('data-group')===g;})[0];}
function closeAll(focusBack){var was=null;btns.forEach(function(b){if(b.getAttribute('aria-expanded')==='true')was=b;b.setAttribute('aria-expanded','false');});panels.forEach(function(p){p.hidden=true;});n.classList.remove('panel-open');if(focusBack&&was)was.focus();}
function open(g){closeAll(false);var b=btns.filter(function(x){return x.getAttribute('data-group')===g;})[0],p=panelOf(g);if(!b||!p)return;b.setAttribute('aria-expanded','true');p.hidden=false;n.classList.add('panel-open');anchor(b,p);}
// the panel sits under its button; a panel that would run past the container's right edge aligns its right edge to the button's right edge instead (never past the viewport)
function anchor(b,p){var pw=n.querySelector('.panel-wrap');if(!pw)return;var cr=pw.getBoundingClientRect(),br=b.getBoundingClientRect(),w=p.offsetWidth,left=br.left-cr.left;if(left+w>cr.width)left=(br.right-cr.left)-w;if(left<0)left=0;p.style.setProperty('--panel-left',Math.round(left)+'px');}
window.addEventListener('resize',function(){var o=btns.filter(function(x){return x.getAttribute('aria-expanded')==='true';})[0];if(o)anchor(o,panelOf(o.getAttribute('data-group')));}); btns.forEach(function(b,i){
b.addEventListener('click',function(){var on=b.getAttribute('aria-expanded')==='true';if(on)closeAll(false);else open(b.getAttribute('data-group'));});
b.addEventListener('keydown',function(e){
var k=e.key;
if(k==='ArrowRight'||k==='ArrowLeft'){e.preventDefault();var j=(i+(k==='ArrowRight'?1:btns.length-1))%btns.length;btns[j].focus();if(n.classList.contains('panel-open'))open(btns[j].getAttribute('data-group'));}
else if(k==='ArrowDown'){e.preventDefault();open(b.getAttribute('data-group'));var f=panelOf(b.getAttribute('data-group')).querySelector('a');if(f)f.focus();}
else if(k==='Escape'){closeAll(true);}
});
});
if(wrap)wrap.addEventListener('keydown',function(e){if(e.key==='Escape'){e.stopPropagation();closeAll(true);}});
document.addEventListener('click',function(e){if(!n.classList.contains('panel-open'))return;if(e.target.closest('.group-btn')||e.target.closest('.nav-panel'))return;closeAll(false);});
document.addEventListener('focusin',function(e){if(!n.classList.contains('panel-open'))return;if(n.contains(e.target))return;closeAll(false);});
document.addEventListener('keydown',function(e){if(e.key==='Escape'&&n.classList.contains('panel-open'))closeAll(true);});
// the sheet on phones
function setSheet(o){if(!burger||!sheet)return;n.classList.toggle('open',o);sheet.hidden=!o;burger.setAttribute('aria-expanded',o?'true':'false');burger.setAttribute('aria-label',o?'Close the menu':'Open the menu');document.documentElement.classList.toggle('nav-locked',o);if(o){var f=sheet.querySelector('a');if(f)f.focus();}}
if(burger&&sheet){
burger.addEventListener('click',function(){setSheet(!n.classList.contains('open'));});
sheet.addEventListener('click',function(e){if(e.target.closest('a'))setSheet(false);});
sheet.addEventListener('keydown',function(e){
if(e.key==='Escape'){setSheet(false);burger.focus();return;}
if(e.key!=='Tab')return;var f=sheet.querySelectorAll('a,button');if(!f.length)return;var first=f[0],last=f[f.length-1];
if(e.shiftKey&&document.activeElement===first){e.preventDefault();burger.focus();}else if(!e.shiftKey&&document.activeElement===last){e.preventDefault();burger.focus();}
});
burger.addEventListener('keydown',function(e){if(e.key==='Tab'&&!e.shiftKey&&n.classList.contains('open')){e.preventDefault();var f=sheet.querySelector('a');if(f)f.focus();}});
var mq=window.matchMedia('(min-width:861px)');if(mq.addEventListener)mq.addEventListener('change',function(){setSheet(false);});
}
// the theme toggle
var t=n.querySelector('[data-theme-toggle]');
function paintToggle(){var dark=document.documentElement.getAttribute('data-theme')!=='light';t.setAttribute('aria-label',dark?'Switch to light theme':'Switch to dark theme');t.innerHTML=dark?'<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><circle cx="12" cy="12" r="4"/><path d="M12 2v2m0 16v2M2 12h2m16 0h2M5 5l1 1m12 12 1 1M5 19l1-1M18 6l1-1"/></svg>':'<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="M20.7 13.3A8.5 8.5 0 0 1 10.7 3.3 8.5 8.5 0 1 0 20.7 13.3Z"/></svg>';}
if(t){t.addEventListener('click',function(){var v=document.documentElement.getAttribute('data-theme')==='light'?'dark':'light';document.documentElement.setAttribute('data-theme',v);try{localStorage.setItem('igneum-theme',v);}catch(e){}document.dispatchEvent(new CustomEvent('igneum-theme'));});paintToggle();document.addEventListener('igneum-theme',paintToggle);}
})();
</script>
<!-- nav:end -->
<main id="main">
<section class="page-hero">
<div class="container">
<div class="breadcrumb"><a href="/">Igneum</a><span>/</span><span>GPU bench table</span></div>
<div class="page-heading"><div><div class="eyebrow"><span class="line"></span>38 cards you can buy, 13 datacentre, 0 fleet tuning models</div>
<h1>GPU bench table</h1>
<p class="lead">Measured hash rates per card on the Igneum lottery hash, on the current program, with the date and the record behind each number.</p></div></div>
</div>
</section>
<section class="section compact"><div class="container">
<div class="doc-tools"><div class="searchbar"><svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><circle cx="10.5" cy="10.5" r="6.5"/><path d="m16 16 5 5"/></svg><label for="doc-q" class="visually-hidden">Filter the sections</label><input id="doc-q" data-filter-input="miners" placeholder="Filter the sections: a card, a date, a word" autocomplete="off" spellcheck="false"><span class="small" data-filter-count="miners"></span></div></div>
<div class="docs-layout">
<nav class="doc-nav" aria-label="Contents"><div class="eyebrow">Contents</div><a href="#table">The table</a><a href="#how">How a row gets here</a><a href="#priors">Fleet tuning priors</a></nav>
<article class="doc-body" data-filter-list="miners"><section class="doc-section"><style>
.lede{font-size:18px;line-height:1.5;color:var(--bone);margin:0 0 var(--s-4)}
.grp{font:500 10px/1.4 var(--mono);letter-spacing:.1em;text-transform:uppercase;color:var(--ash);margin:var(--s-5) 0 10px;padding-bottom:8px;border-bottom:1px solid var(--line)}
details.grp summary{cursor:pointer;list-style:none}details.grp summary::-webkit-details-marker{display:none}details.grp summary::after{content:" \u25BE";opacity:.6}details.grp[open] summary::after{content:" \u25B4"}
.tbl.bench2wrap{overflow-x:auto;border:0;background:none;border-radius:0;max-width:100%}
table.bench2{min-width:0;width:100%;border-collapse:collapse;font-size:14px;line-height:1.4}
table.bench2 th{font:500 10px/1.4 var(--mono);letter-spacing:.08em;text-transform:uppercase;color:var(--ash);text-align:left;padding:0 12px 10px;border-bottom:1px solid var(--line);white-space:nowrap}
table.bench2 th .sort{all:unset;cursor:pointer;font:inherit;color:inherit;letter-spacing:inherit;text-transform:inherit;padding-bottom:3px;border-bottom:2px solid transparent}
table.bench2 th .sort::after{content:" \u2195";opacity:.4}
table.bench2 th[aria-sort="descending"] .sort,table.bench2 th[aria-sort="ascending"] .sort{color:var(--bone);border-bottom-color:var(--ember)}
table.bench2 th[aria-sort="descending"] .sort::after{content:" \u2193";opacity:1;color:var(--ember)}table.bench2 th[aria-sort="ascending"] .sort::after{content:" \u2191";opacity:1;color:var(--ember)}
table.bench2 tr.row td{height:56px;padding:0 12px;border-bottom:1px solid var(--line);background:none;vertical-align:middle;white-space:nowrap;color:var(--ink-2)}
table.bench2 tr.row:hover td{background:var(--hover)}
table.bench2 td.lead{color:var(--bone)}@media(min-width:1101px){table.bench2 tr.row td.lead{white-space:normal;min-width:220px;max-width:360px;padding-top:8px;padding-bottom:8px}}
table.bench2 td.lead .cardname{display:inline-flex;align-items:center;gap:10px}
table.bench2 td.lead .gen{display:block;font:12px var(--mono);color:var(--ash);letter-spacing:.02em;margin-top:3px;white-space:normal;overflow-wrap:anywhere}table.bench2 td.lead .gen .dot{color:var(--line-2);margin:0 4px}
table.bench2 td.big{font:600 22px/1 var(--f-display,var(--f-sans));color:var(--bone);font-variant-numeric:tabular-nums}
table.bench2 td.tnum{font-variant-numeric:tabular-nums}
table.bench2 .eff{font:500 15px/1 var(--f-sans);color:var(--ink-2)}table.bench2 .eff.best{color:var(--ember-text,var(--ember));font-weight:600}
table.bench2 .hollow{color:var(--ash)}
table.bench2 td.quiet{color:var(--ash);font-size:13px}table.bench2 .basis{font:10px var(--mono);letter-spacing:.06em;text-transform:uppercase;color:var(--ash);margin-left:6px}
table.bench2 .pill.state{padding:5px 8px;font-size:9px}table.bench2 .pill.state.tuned{color:var(--ember-text,var(--ember));border-color:color-mix(in srgb,var(--ember) 35%,transparent);background:color-mix(in srgb,var(--ember) 8%,transparent)}table.bench2 .pill.state.stock{color:var(--ash)}table.bench2 .pill.state.nolever{color:var(--ash);border-style:dashed}
table.bench2 button.more{all:unset;cursor:pointer;font:500 10px/1.4 var(--mono);letter-spacing:.08em;text-transform:uppercase;color:var(--ash);padding:6px 0;border-bottom:1px solid var(--line-2)}table.bench2 button.more:hover{color:var(--bone)}
table.bench2 tr.detail td{padding:12px 12px 16px;border-bottom:1px solid var(--line);font-size:13px;line-height:1.55;color:var(--ash);white-space:normal;overflow-wrap:anywhere}
table.bench2 tr.detail .d{display:block;margin:0 0 4px}table.bench2 tr.detail b{color:var(--ink-2);font-weight:600;margin-right:6px}
@media(max-width:1100px){
table.bench2,table.bench2 tbody,table.bench2 tr.detail,table.bench2 tr.detail td{display:block}
table.bench2 thead{display:none}
table.bench2 tr.row{display:grid;grid-template-columns:minmax(0,1fr) auto;grid-template-areas:"lead big" "eff eff" "w pill" "more more";gap:4px 12px;padding:10px 0;min-height:44px;border-bottom:1px solid var(--line);align-items:center}
table.bench2 tr.row td{display:block;height:auto;padding:0;border:0;white-space:normal;min-width:0}
table.bench2 tr.row td.lead{grid-area:lead;max-width:none;min-width:0}table.bench2 td.lead .cardname{display:flex;align-items:center;gap:8px}table.bench2 td.lead .cardname .badge{flex:none}
table.bench2 td.big{grid-area:big;font-size:20px;text-align:right;white-space:nowrap}
table.bench2 td.effc{grid-area:eff;font-size:13px}table.bench2 td.effc .eff::before,table.bench2 td.effc .hollow::before{content:"MH per wall watt ";color:var(--ash);font-weight:400}
table.bench2 td.quiet.tnum{grid-area:w}table.bench2 td.pillc{grid-area:pill;text-align:right}
table.bench2 td.morec{grid-area:more;text-align:right}
table.bench2 tr.detail td{padding:8px 0 14px}
}
</style>
</section>
<section class="doc-section"><h2 id="table">The table</h2>
<p class="lede"><strong>Best desktop card:</strong> NVIDIA RTX 5090, 136.1 MH/s (measured, 2026-10-06). <strong>Best stock MH per wall watt:</strong> NVIDIA RTX 5070, 0.334 at 176.1 W, untuned (measured by the fleet, 2026-10-08). <strong>Best tuned:</strong> NVIDIA RTX 5090, 0.563 at 226.8 W (measured, 2026-10-06).</p>
<p class="lede sweep">The rented model sweep of 8 October 2026 (23 rows, 22 models, every row untuned) found the rate cap-proof: a card on a host power cap gives the same MH/s as an uncapped one. Class v5 held the rate on every uncapped card and read 17 to 18 percent under it on the two power-capped hosts (an A4000 at 114 W, a GTX 1080 Ti at 130 W), so a class v5 rate needs the card's power state beside it. The sweep's watts, and the 7 October fleet rows' watts, are class v3 figures: the fleet's bench pack ran the class v3 program, and class v4 draws more for the same rate (a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4, 8 October 2026). Those rows are ranked by rate alone; the per-watt claims above come from rows measured under class v4; the class v4 watts rerun lands on each row with its date, the v3 watts kept beside it as the shadow premium per card.</p>
<p>One row per card on the current class: the class v4 program (the latency-shadow block over the class v3 hash), or a class v3 row re-measured with its class v4 cost on 6 October 2026 or later. MH per wall watt uses board or wall power; a row whose watts are the chip's (Apple silicon: GPU plus DRAM from IOReport) says so and is not ranked on that column. Integrated GPUs are not listed.</p>
<h3 class="grp" id="buy">Cards you can buy</h3>
<div class="tbl bench2wrap"><table class="bench2" id="bench-current"><thead><tr><th data-key="card" data-type="text" aria-sort="none"><button type="button" class="sort">Card</button></th><th data-key="mh_s" data-type="num" aria-sort="descending"><button type="button" class="sort">MH/s</button></th><th data-key="mh_per_w" data-type="num" aria-sort="none"><button type="button" class="sort">MH per wall watt</button></th><th data-key="watts" data-type="num" aria-sort="none"><button type="button" class="sort">Watts</button></th><th data-key="tuned" data-type="text" aria-sort="none"><button type="button" class="sort">Tuned</button></th><th></th></tr></thead><tbody><tr class="row"><td class="lead" data-v="NVIDIA RTX 5090 (32 GB)">NVIDIA RTX 5090 (32 GB)<span class="gen">class v4 133.8 MH/s, 305 W, 0.439 <span class="dot">·</span> <span class="date">2026-10-06</span></span></td><td class="big tnum" data-v="136.1">136.1 stock (127.7 tuned)</td><td class="tnum effc" data-v="0.563"><span class="eff best">0.563</span></td><td class="quiet tnum" data-v="226.8">226.8 tuned (350 stock) W</td><td class="pillc" data-v="Tuned"><span class="pill state tuned">Tuned</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d0">Details</button></td></tr><tr class="detail" id="bench-current-d0" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> core lock 1,854 MHz, mem 13,801 MHz, PL 460 W (measured 6 October 2026 (Ember run 6: the clock lock 1,854 MHz, the memory clock as read, the limit 460 W of 575 as the cap did not bind))</span><span class="d"><b>Class v4 cost</b> class v4 at the 1,200 MHz knee: 133.80 MH/s at 305.1 W (0.439 MH/W); the premium over class v3 81.8 W at the best points and 145.3 W unlocked, measured 7 October 2026</span><span class="d"><b>Tuned</b> full Ember Tune: 1,854 MHz core lock at the 100 percent cap</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Windows 11</span><span class="d"><b>Source</b> bench log: 6 October 2026, Ember run 6; the chip-model measurement record: item 8's 5090 rows and the class v4 efficiency pass (7 October 2026) · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> one row for the card: best rate 136.1 MH/s at 350 W stock unlocked on the class v3 control (bench, 6 October 2026); best MH per wall watt 127.71 MH/s at 226.8 W at the full Ember Tune point, 1,854 MHz (Ember run 6, 6 October 2026); the class v4 efficiency pass (7 October 2026, 60 s steps, every fingerprint matched): unlocked 136.84 MH/s at 475.5 W (0.288), the 1,400 MHz lock 134.98 at 316.3 W (0.427), the knee 1,300 MHz, the best point 1,200 MHz 133.80 at 305.1 W (0.439); the fleet's standing 5090 reads 100.6 MH/s at 308 W beside its prover (reported by the fleet)</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5070 Ti (16 GB)">NVIDIA RTX 5070 Ti (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="78.45">69 to 78 by board (68.94 on this host)</td><td class="tnum effc" data-v="0.322"><span class="eff">0.322</span></td><td class="quiet tnum" data-v="214">214 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d1">Details</button></td></tr><tr class="detail" id="bench-current-d1" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.84, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 1 (a rented card, result cb2-rtx-5070-ti-16-gb-7324-result.json; 09:23Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-5070-ti-16-gb-1858-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.06 h; power.draw mean 135.9 W (instant 139.0, max 140.0, limit 300 W), SM 2,795 MHz mean, memory 13,801 MHz, 47 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the same rate on this card (78.72 MH/s, fingerprint matched); read ceiling 79.6 MH/s, so the hash runs at 0.99 of it. Class v4 watts (8 October 2026, the rerun, self-test PASS): 214 W at 68.94 MH/s, 0.322 MH per watt; the class v3 pack on the same host read 140.2 W at the same rate, so the shadow premium on this card is +52.6 percent of watts for no rate. two 5070 Ti memory populations exist: this board and one 7 October board read 69 MH/s, three others 78.4 to 78.8, all reporting the same 13,801 MHz memory clock, so the rate reads as a range with the host; the grid lane's own pair on a 78 MH/s board (141 W class v3, 224 W class v4, +59 percent) agrees on the watts to 10 W and gives 0.350 MH per watt there</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5070 Ti (16 GB)">NVIDIA RTX 5070 Ti (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="78.43">78.4</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.539</span></td><td class="quiet tnum" data-v="145.6">145.6 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d2">Details</button></td></tr><tr class="detail" id="bench-current-d2" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5080 (16 GB)">NVIDIA RTX 5080 (16 GB)<span class="gen">class v4 +83 W unlocked, +41 W at the best points <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="71.2">71.4 stock (71.2 tuned)</td><td class="tnum effc" data-v="0.486"><span class="eff">0.486</span></td><td class="quiet tnum" data-v="146.6">146.6 tuned (253 stock) W</td><td class="pillc" data-v="Tuned"><span class="pill state tuned">Tuned</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d3">Details</button></td></tr><tr class="detail" id="bench-current-d3" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> core lock 1,100 MHz, mem 0 MHz, PL 0 W (core lock 1,100 MHz, memory stock, power limit untouched (measured 8 October 2026, the efficiency pass))</span><span class="d"><b>Class v4 cost</b> 83.4 W unlocked and 41 W at the best points for 0.2 percent of rate (class v4 71.41 MH/s at 253.1 W against class v3 71.28 at 169.7 W unlocked; measured 8 October 2026)</span><span class="d"><b>Tuned</b> core lock 1,100 MHz (the efficiency pass's best point: 71.20 MH/s at 146.6 W, 0.486 MH/W, 106.5 W recovered for 0.29 percent of rate), memory 14,801 MHz, the driver's power limit untouched</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 617.14, Windows 11</span><span class="d"><b>Source</b> bench log: 7 to 8 October 2026, the class v4 efficiency passes: the core clock lock on the RTX 5090 and the RTX 5080 · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> the knee sits between 1,000 and 900 MHz (the rate within 0.3 percent of unlocked down to 1,000 MHz, 5.2 percent off at 900 on class v4); the draw floors from 1,500 MHz down. The rented fleet's 5080 read 71.16 MH/s at 143.4 W stock (the hive package, Ubuntu 24.04, 7 October 2026), 110 W under this card's stock draw: the method differs per row (nvidia-smi on Linux, the runner's card switch on Windows) and neither is taken as the card's figure until the fleet re-measures</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3090 Ti (24 GB)">NVIDIA RTX 3090 Ti (24 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="61.95">62</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.248</span></td><td class="quiet tnum" data-v="249.5">249.5 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d4">Details</button></td></tr><tr class="detail" id="bench-current-d4" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.65.06, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3080 Ti (12 GB)">NVIDIA RTX 3080 Ti (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="58.95">59</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.221</span></td><td class="quiet tnum" data-v="267.3">267.3 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d5">Details</button></td></tr><tr class="detail" id="bench-current-d5" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.65.06, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> 283.0 W maximum. Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3080 Ti (12 GB)">NVIDIA RTX 3080 Ti (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="58.86">58.9</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.201</span></td><td class="quiet tnum" data-v="293.5">293.5 (class v3, uncapped board) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d6">Details</button></td></tr><tr class="detail" id="bench-current-d6" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.65.06, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 3 (a rented card, result cb2-rtx-3080-ti-12-gb-efll-result.json; 09:28Z) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.04 h; power.draw mean 293.5 W (max 299.6, limit 320 W), SM 1,936 MHz mean, memory 9,251 MHz, 74 C, the SW power-cap bit set 26 W under the limit; self-test PASS (96 of 96 vector lanes); the class v5 kit runs 2.5 percent faster on this card (60.38 MH/s, fingerprint matched); read ceiling 58.9 MH/s, the hash at 1.00 of it. Matches the 7 October row (58.95 MH/s) across hosts; the worst MH per watt of the sweep and the cheapest MH per rented hour, so a buyer on a power bill prefers the 5070 Ti and a renter this card. Class v4 pair (8 October 2026, result cb4-rtx-3080-ti-12-gb-7647-result.json, a different and deep-capped board: 220 W host cap on a 350 W card, the SM clock collapsed to 749 MHz): 42.24 MH/s mean at 216.2 W under class v4 against 44.93 MH/s at 166.2 W under class v3 on that board, the shadow premium +30.1 percent of watts and, on this cap, 5 percent of rate: the first cap that costs rate (a cap near 63 percent of stock on a 350 W GDDR6X card starves the SM clock the memory loop needs, where the 4070 Super at half its 220 W and the 3070 at 88 percent held full rate). The uncapped morning rate (58.86 MH/s) stands as this card's rate; its class v4 watts on an uncapped board are not yet measured and would read about 380 W by this pair's ratio (an estimate). The rule from the pairs so far: a power cap holds the rate until the SM clock falls below roughly 1.8 GHz on Ampere and 2.4 GHz on Ada</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4090 (24 GB)">NVIDIA RTX 4090 (24 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="52.25">52.3</td><td class="tnum effc" data-v="0.285"><span class="eff">0.285</span></td><td class="quiet tnum" data-v="183.1">183.1 W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d7">Details</button></td></tr><tr class="detail" id="bench-current-d7" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, the RTX 4090 hands row · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> the hands row: the card mines and proves (17.4 GB peak while proving); the standing 4090 voters read 44.5 to 50.8 MH/s this hour beside their provers</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5070 (12 GB)">NVIDIA RTX 5070 (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="52.01">52</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.506</span></td><td class="quiet tnum" data-v="102.8">102.8 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d8">Details</button></td></tr><tr class="detail" id="bench-current-d8" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5070 (12 GB)">NVIDIA RTX 5070 (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="51.81">52 to 59 by board (58.89 on this host)</td><td class="tnum effc" data-v="0.334"><span class="eff">0.334</span></td><td class="quiet tnum" data-v="176.1">176.1 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d9">Details</button></td></tr><tr class="detail" id="bench-current-d9" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.173.02, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 8 (a rented card, result cb2-rtx-5070-12-gb-8750-result.json; 09:35Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-5070-12-gb-5732-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.09 h; power.draw mean 110.8 W (max 112.0, limit 250 W), SM 2,851 MHz mean, memory 13,801 MHz, 60 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (51.98 MH/s, fingerprint matched); read ceiling 52.3 MH/s, the hash at 0.99 of it. Repeats the 7 October row (52.01 then, 51.82 now) at 8 W more on this host; second to the 5070 Ti in MH per watt among consumer cards (GDDR7 lifts Blackwell here, not SM count); 12 GB holds the class v4 set with 10 GB spare. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 176.1 W at 58.89 MH/s, 0.334 MH per watt; the class v3 pack on the same host read 113.1 W at the same rate, so the shadow premium on this card is +55.7 percent of watts for no rate (the class is memory-bound). board-to-board spread: this host 58.9 MH/s on both classes, the morning host 51.8 (GDDR7 memory clocks differ by board), so the rate reads as a range with the host noted</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3090 (24 GB)">NVIDIA RTX 3090 (24 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="50.04">50</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d10">Details</button></td></tr><tr class="detail" id="bench-current-d10" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, mining</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Ubuntu 24.04</span><span class="d"><b>Source</b> the fleet's standing voters (status lines, 7 October 2026) · <b>Who</b> reported by the fleet</span><span class="d"><b>Note</b> the best of four standing voters this hour (42.2 to 50.0 MH/s) beside their provers; watts not read</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3080 (10 GB)">NVIDIA RTX 3080 (10 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="43.72">43.7</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d11">Details</button></td></tr><tr class="detail" id="bench-current-d11" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, mining</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Ubuntu 24.04</span><span class="d"><b>Source</b> the fleet's standing voter (status line, 7 October 2026) and the sweep's hands row · <b>Who</b> reported by the fleet</span><span class="d"><b>Note</b> a standing voter's status line, 18:00Z; the hands row of the sweep reads 40.82 MH/s at 204.9 W; 10 GB, no prover</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4080 Super (16 GB)">NVIDIA RTX 4080 Super (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="42.6">42.6</td><td class="tnum effc" data-v="0.218"><span class="eff">0.218</span></td><td class="quiet tnum" data-v="195.7">195.7 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d12">Details</button></td></tr><tr class="detail" id="bench-current-d12" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.126.09, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 23 (a rented card, result cb2-rtx-4080-super-16-g-2453-result.json; 10:03Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4080-super-16-g-8479-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.12 h; power.draw mean 134.9 W (max 138.4, limit 320 W), SM 2,740 MHz mean, memory 11,254 MHz, 51 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (42.68 MH/s, fingerprint matched); read ceiling 43.1 MH/s, the hash at 0.99 of it. The 23 Gbps GDDR6X gives it 4.6 percent over the 4080 for 6 W more at the same MH per watt: the Ada 256-bit tier is 41 to 43 MH/s whichever card. Class v4 watts (8 October 2026, the rerun, self-test PASS): 195.7 W at 42.63 MH/s, 0.218 MH per watt; the class v3 pack on the same host read 139.8 W at the same rate, so the shadow premium on this card is +40.0 percent of watts for no rate. a sibling board of the morning's host; the rate and the v3 draw repeat the morning; the Ada 256-bit tier reads 0.22 MH per watt under class v4 whichever card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4070 Ti Super (16 GB)">NVIDIA RTX 4070 Ti Super (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="41.24">37 to 41 by board (36.93 on this host)</td><td class="tnum effc" data-v="0.268"><span class="eff">0.268</span></td><td class="quiet tnum" data-v="137.8">137.8 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d13">Details</button></td></tr><tr class="detail" id="bench-current-d13" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 615.71.09, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 2 (a rented card, result cb2-rtx-4070-ti-super-7325-result.json; 09:26Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4070-ti-super-5733-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.10 h; power.draw mean 142.6 W (max 145.2, limit 285 W), SM 2,885 MHz mean, memory 10,251 MHz, 47 C, the SW thermal bit set with no rate effect (three runs within 0.02 percent); self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (41.37 MH/s, fingerprint matched); read ceiling 41.6 MH/s, the hash at 0.99 of it. The Ada 192-bit bus class: the same MH per watt as the 4070 Ti; the extra 4 GB buys headroom, not rate. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 137.8 W at 36.93 MH/s, 0.268 MH per watt; the class v3 pack on the same host read 102.3 W at the same rate, so the shadow premium on this card is +34.7 percent of watts for no rate (the class is memory-bound). this host is a 210 W, 2,400 MHz board at 36.9 MH/s on both classes against the morning board's 41.2, so the rate reads as a range with the host noted; the premium is smaller than the 5070's because this board sits nearer its cap</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4080 (16 GB)">NVIDIA RTX 4080 (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="40.71">40.7</td><td class="tnum effc" data-v="0.219"><span class="eff">0.219</span></td><td class="quiet tnum" data-v="185.6">185.6 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d14">Details</button></td></tr><tr class="detail" id="bench-current-d14" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.84, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 4 (a rented card, result cb2-rtx-4080-16-gb-7347-result.json; 09:28Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4080-16-gb-7107-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.15 h; power.draw mean 128.6 W (max 130.5, limit 320 W), SM 2,776 MHz mean, memory 10,801 MHz, 38 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (40.80 MH/s, fingerprint matched); read ceiling 41.0 MH/s, the hash at 0.99 of it. One percent under the 4070 Ti Super at 14 W less: the Ada 16 GB tier is a 41 MH/s tier whichever card. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 185.6 W at 40.73 MH/s, 0.219 MH per watt; the class v3 pack on the same host read 128.1 W at the same rate, so the shadow premium on this card is +44.9 percent of watts for no rate (the class is memory-bound). the same board as the morning row (offer 54551253; the v3 draw repeats to half a watt); the 256-bit board's premium sits above the 192-bit cards' third</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3070 Ti (8 GB)">NVIDIA RTX 3070 Ti (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="38.98">39</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.219</span></td><td class="quiet tnum" data-v="178.3">178.3 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d15">Details</button></td></tr><tr class="detail" id="bench-current-d15" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.71.05, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3070 Ti (8 GB)">NVIDIA RTX 3070 Ti (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="38.98">39</td><td class="tnum effc" data-v="0.146"><span class="eff">0.146</span></td><td class="quiet tnum" data-v="267.6">267.6 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d16">Details</button></td></tr><tr class="detail" id="bench-current-d16" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.71.05, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 6 (a rented card, result cb2-rtx-3070-ti-8-gb-8528-result.json; 09:33Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-3070-ti-8-gb-7705-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.07 h; power.draw mean 177.1 W (max 183.6, limit 290 W), SM 1,955 MHz mean, memory 9,251 MHz, 67 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (39.12 MH/s, fingerprint matched); read ceiling 39.1 MH/s, the hash at 1.00 of it. Repeats the 7 October row on another host (38.98 then, 38.98 now; 178 then 177 W), so the class v4 number is host-independent to 0.1 percent. Class v4 watts (8 October 2026, the rerun, self-test PASS): 267.6 W at 39.1 MH/s, 0.146 MH per watt; the class v3 pack on the same host read 180.4 W at the same rate, so the shadow premium on this card is +48.3 percent of watts for no rate. the same board as the 7 October row (offer 36538839), uncapped (a 310 W AIB); the uncapped GDDR6X Ampere premium sits with the 4080's and the 5070's rather than the 192-bit Ada third, so the premium scales with the SM the board had idle under class v3</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3070 (8 GB)">NVIDIA RTX 3070 (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="37.09">37.1</td><td class="tnum effc" data-v="0.189"><span class="eff">0.189</span></td><td class="quiet tnum" data-v="196.4">196.4 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d17">Details</button></td></tr><tr class="detail" id="bench-current-d17" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.65.06, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 5 (a rented card, result cb2-rtx-3070-8-gb-qh9j-result.json; 09:32Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-3070-8-gb-l6mu-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.06 h; power.draw mean 142.3 W (max 146.7, limit 225 W), SM 2,002 MHz mean, memory 6,801 MHz, 57 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (37.14 MH/s, fingerprint matched); read ceiling 37.2 MH/s, the hash at 1.00 of it. Against the 3070 Ti: 5 percent less rate for 25 percent less power, the better owner card of the two; 8 GB holds the class v4 set with room. Class v4 watts (8 October 2026, the rerun, self-test PASS): 196.4 W at 37.11 MH/s, 0.189 MH per watt; the class v3 pack on the same host read 153.4 W at the same rate, so the shadow premium on this card is +28 percent of watts for no rate. this board is capped at 193 W by the host (stock 220) and the class v4 draw sits on the cap with the power-cap bit set while the rate holds, the same shape as the capped 4070 Super; the first Ampere pair</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3070 (8 GB)">NVIDIA RTX 3070 (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="33.66">33.7</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d18">Details</button></td></tr><tr class="detail" id="bench-current-d18" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, mining</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Ubuntu 24.04</span><span class="d"><b>Source</b> a fleet canary (status line, 7 October 2026) · <b>Who</b> reported by the fleet</span><span class="d"><b>Note</b> a fleet canary’s status line, 17:07Z, beside its node; watts not read</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3060 Ti (8 GB)">NVIDIA RTX 3060 Ti (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="33.1">33.1</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.256</span></td><td class="quiet tnum" data-v="129.5">129.5 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d19">Details</button></td></tr><tr class="detail" id="bench-current-d19" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3060 Ti (8 GB)">NVIDIA RTX 3060 Ti (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="31.7">32 to 33 by board (33.06 on this host)</td><td class="tnum effc" data-v="0.256"><span class="eff">0.256</span></td><td class="quiet tnum" data-v="128.7">128.7 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d20">Details</button></td></tr><tr class="detail" id="bench-current-d20" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.91.07, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 10 (a rented card, result cb2-rtx-3060-ti-8-gb-8771-result.json; 09:37Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-3060-ti-8-gb-8620-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.12 h; power.draw mean 100.4 W (max 104.7, limit 180 W), SM 1,560 MHz mean, memory 5,001 MHz on this host (below the 7,000 MHz GDDR6 stock), 61 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (31.75 MH/s, fingerprint matched); read ceiling 31.8 MH/s, the hash at 1.00 of it. Four percent under the 7 October 3060 Ti row (33.1 on another host), consistent with this host's lower memory clock; the best MH per watt of the Ampere cards. Class v4 watts (8 October 2026, the rerun, self-test PASS): 128.7 W at 32.92 MH/s, 0.256 MH per watt; the class v3 pack on the same host read 129.9 W at the same rate, so the shadow premium on this card is -0.9 percent of watts for no rate. the cleanest cap reading of the set: this board is capped at 130 W by the host (stock 200) and both classes sit on the cap, the SM giving a quarter of its clock (1,722 to 1,297 MHz) while the rate holds, so a 3060 Ti at 65 percent of stock runs class v4 at full rate with no premium; the rate on this board (33.1) matches the 7 October host and sits 4 percent over the morning's slow-memory board, so the rate reads as a range</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5060 (8 GB)">NVIDIA RTX 5060 (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="31.27">31.3</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.415</span></td><td class="quiet tnum" data-v="75.4">75.4 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d21">Details</button></td></tr><tr class="detail" id="bench-current-d21" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4070 Ti (12 GB)">NVIDIA RTX 4070 Ti (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="31.26">31.3</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.291</span></td><td class="quiet tnum" data-v="107.3">107.3 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d22">Details</button></td></tr><tr class="detail" id="bench-current-d22" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4070 Super (12 GB)">NVIDIA RTX 4070 Super (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="31.24">31.2</td><td class="tnum effc" data-v="0.288"><span class="eff">0.288</span></td><td class="quiet tnum" data-v="108.4">108.4 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d23">Details</button></td></tr><tr class="detail" id="bench-current-d23" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.178.04, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 9 (a rented card, result cb2-rtx-4070-super-12--8582-result.json; 09:36Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4070-super-12--6238-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.13 h; power.draw mean 100.2 W (max 103.9) under a 110 W host cap (stock 220 W), SM 2,764 MHz mean, memory 10,251 MHz, 53 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (31.29 MH/s, fingerprint matched); read ceiling 34.4 MH/s, the hash at 0.91 of it. Lands on the 4070 Ti's number (31.26 on 7 October) at 7 W less (the same 192-bit 21 Gbps bus); the host's 110 W cap cost no rate, a tune-down reading in itself: an owner can cap a 4070 Super at half its stock power and lose nothing on class v4. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 108.4 W at 31.25 MH/s, 0.288 MH per watt; the class v3 pack on the same host read 99.2 W at the same rate, so the shadow premium on this card is +9.3 percent of watts for no rate (the class is memory-bound). the same board as the morning row (host offer 51485188), capped at 110 W by the host (stock 220) with the SW power-cap bit set the whole run: the cap absorbs the class v4 stage's extra draw (the SM clock drops 10 percent, the rate does not move), so the premium is 9 percent here against a third on uncapped Ada boards; a power cap near half stock costs a 192-bit Ada card nothing on class v4</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4070 Ti (12 GB)">NVIDIA RTX 4070 Ti (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="31.24">31.2</td><td class="tnum effc" data-v="0.201"><span class="eff">0.201</span></td><td class="quiet tnum" data-v="155.3">155.3 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d24">Details</button></td></tr><tr class="detail" id="bench-current-d24" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.126.09, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 20 (a rented card, result cb2-rtx-4070-ti-12-gb-0695-result.json; 09:57Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4070-ti-12-gb-5744-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.18 h; power.draw mean 95.4 W (max 96.7, limit 285 W), SM 2,825 MHz mean, memory 10,251 MHz, 49 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (31.28 MH/s, fingerprint matched); read ceiling 31.7 MH/s, the hash at 0.99 of it. Repeats the 7 October rate to 0.07 percent on another provider at 12 W less; the 192-bit Ada pair (4070 Ti, 4070 Super) land on one number, 31.2 MH/s under 100 W. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 155.3 W at 31.26 MH/s, 0.201 MH per watt; the class v3 pack on the same host read 113.8 W at the same rate, so the shadow premium on this card is +36.5 percent of watts for no rate (the class is memory-bound). the rate matches the morning and 7 October rows on a third host (no host spread on this card); driver 570.169 ran the class v4 pack clean</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4070 (12 GB)">NVIDIA RTX 4070 (12 GB)<span class="gen">class v4 +30 W <span class="dot">·</span> <span class="date">2026-10-06</span></span></td><td class="big tnum" data-v="30.95">31</td><td class="tnum effc" data-v="0.389"><span class="eff">0.389</span></td><td class="quiet tnum" data-v="79.5">79.5 W</td><td class="pillc" data-v="Tuned"><span class="pill state tuned">Tuned</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d25">Details</button></td></tr><tr class="detail" id="bench-current-d25" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> core lock 1,863 MHz, mem 10,251 MHz, PL 100 W (measured 6 October 2026 (Ember run 6: 1,863 MHz at the 50 percent cap of 200 W, 75.6 W drawn; the item 8 rows at 1,860 MHz and 160 W))</span><span class="d"><b>Class v4 cost</b> +30 W (79 to 109 W) for +0.4 percent of rate at 102,100 ops per hash, measured 6 October 2026</span><span class="d"><b>Tuned</b> full Ember Tune: 1,860 MHz core lock, 160 W cap (Ember run 6: 1,863 MHz at the 50 percent cap, 75.6 W)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Windows 11</span><span class="d"><b>Source</b> the chip-model measurement record: item 8, the RTX 4070 rows (job run-ca3-pc1-4070-shadow-20261006) · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> 79.3 to 79.8 W at the tune point; 28.78 MH/s at 75.6 W (0.381) in Ember run 6 mining</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 5060 Ti (16 GB)">NVIDIA RTX 5060 Ti (16 GB)<span class="gen">class v4 0.1 percent rate <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="30.9">30.9</td><td class="tnum effc" data-v="0.269"><span class="eff">0.269</span></td><td class="quiet tnum" data-v="114.8">114.8 W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d26">Details</button></td></tr><tr class="detail" id="bench-current-d26" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> 0.1 percent of rate, measured 7 October 2026</span><span class="d"><b>Tuned</b> stock, bench only (180 W default cap, never tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Windows 11</span><span class="d"><b>Source</b> bench log: 7 October 2026, the first 16 GB card: an RTX 5060 Ti in a Thunderbolt enclosure (run b) · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> class v4 program, 128 loads per hash, 1 GiB dataset, 10-minute window on the card alone at the stock 180 W limit: 114.8 W mean; PCIe 4.0 x4 through the enclosure</span></td></tr><tr class="row"><td class="lead" data-v="Apple M5 Max (40 GPU cores, Metal)">Apple M5 Max (40 GPU cores, Metal)<span class="gen">class v4 +16 W <span class="dot">·</span> <span class="date">2026-10-06</span></span></td><td class="big tnum" data-v="27">27</td><td class="tnum effc" data-v="-1"><span class="hollow" title="chip watts (GPU plus DRAM), not wall; not ranked">○ 1.29</span></td><td class="quiet tnum" data-v="21">21 W <span class="basis">chip, not wall</span></td><td class="pillc" data-v="No lever"><span class="pill state nolever">No lever</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d27">Details</button></td></tr><tr class="detail" id="bench-current-d27" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> +16 W for 1.5 percent of rate at 102,100 ops per hash, measured 6 October 2026</span><span class="d"><b>Tuned</b> no lever on Apple silicon (no clock or power control exposed); stock</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> macOS, Metal</span><span class="d"><b>Source</b> the chip-model measurement record: item 8, the Apple M5 Max rows (IOReport GPU and DRAM watts) · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> GPU plus DRAM watts from IOReport, not wall power, so its MH per watt is not ranked against the cards' wall figures</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 3060 (12 GB)">NVIDIA RTX 3060 (12 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="26.89">26.9</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.241</span></td><td class="quiet tnum" data-v="111.6">111.6 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d28">Details</button></td></tr><tr class="detail" id="bench-current-d28" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.126.09, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> 114.5 W maximum; the devnet boxes read 23.7 to 25.7 MH/s beside their nodes. Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 2080 Ti (11 GB)">NVIDIA RTX 2080 Ti (11 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="26.01">26 to 28 by board (28.05 tuned on this host)</td><td class="tnum effc" data-v="0.174"><span class="eff">0.174</span></td><td class="quiet tnum" data-v="161.1">161.1 (class v4, a tuned board) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d29">Details</button></td></tr><tr class="detail" id="bench-current-d29" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.159.03, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 11 (a rented card, result cb2-rtx-2080-ti-9077-result.json; 09:40Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-2080-ti-0387-result.json, a different board: 22 GB modded, the core held at 1,095 MHz, a 170 W cap; the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.13 h; power.draw mean 224.6 W (max 234.7, limit 300 W; the driver reports no power average on Turing), SM 1,914 MHz mean, memory 6,802 MHz, 72 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (26.04 MH/s, fingerprint matched); read ceiling 29.9 MH/s, the hash at 0.87 of it. The first Turing row: the 7 October 2080 Ti pod that left no result was a host fault, not the card. The worst MH per watt measured on any card (the Titan RTX, the same silicon, read 28.3 MH/s at 226 W on 7 October): an owner mines at a loss on any real power price; 11 GB holds the set. Class v4 watts (8 October 2026, the rerun, self-test PASS) on a tuned board (22 GB memory mod, the core held at 1,095 MHz by the host, a 170 W cap): 161.1 W at 28.05 MH/s, 0.174 MH per watt, the class v3 pack on the same host 139.2 W at 28.04, a premium of +15.7 percent for no rate; this tuned board runs 8 percent over the morning's stock 11 GB board (26.0 MH/s at 1,914 MHz and 224.6 W class v3), so the 2080 Ti's rate is set by its memory and a core lock near 1.1 GHz with a 170 W cap is the way to run one; the morning's stock board under class v4 would read about 0.09 MH per watt by the 2070 Super's premium (an inference)</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4060 Ti (8 GB)">NVIDIA RTX 4060 Ti (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="20.1">20.1</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.259</span></td><td class="quiet tnum" data-v="77.5">77.5 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d30">Details</button></td></tr><tr class="detail" id="bench-current-d30" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4060 Ti (16 GB)">NVIDIA RTX 4060 Ti (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="20.07">20.1</td><td class="tnum effc" data-v="0.196"><span class="eff">0.196</span></td><td class="quiet tnum" data-v="102.3">102.3 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d31">Details</button></td></tr><tr class="detail" id="bench-current-d31" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.71.05, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 7 (a rented card, result cb2-rtx-4060-ti-16-gb-7364-result.json; 09:33Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-4060-ti-16-gb-6679-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.22 h; power.draw mean 79.2 W (max 86.2, limit 165 W), SM 2,836 MHz mean, memory 8,751 MHz, 63 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (20.10 MH/s, fingerprint matched). Repeats the 7 October number (20.10 then, 20.07 now); the one card of the sweep well under its probed read ceiling (0.65 of 30.9 MH/s: the 128-bit bus with the large L2 makes the chase probe optimistic), so 20 MH/s at 79 W is the card's rate; 16 GB holds any class with room. Class v4 watts (8 October 2026, the rerun on the frozen v4-devnet-epoch0 pack, self-test PASS): 102.3 W at 20.1 MH/s, 0.196 MH per watt; the class v3 pack on the same host read 76.6 W at the same rate, so the shadow premium on this card is +33.6 percent of watts for no rate (the class is memory-bound). the same board as the 7 October row (offer 47543124); the rate is the same on every 4060 Ti host and both classes (20.07 to 20.10)</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 2070 Super (8 GB)">NVIDIA RTX 2070 Super (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="19.09">19.1</td><td class="tnum effc" data-v="0.094"><span class="eff">0.094</span></td><td class="quiet tnum" data-v="204.2">204.2 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d32">Details</button></td></tr><tr class="detail" id="bench-current-d32" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 575.57.08, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 12 (a rented card, result cb2-rtx-2070-super-8-g-9129-result.json; 09:41Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-2070-super-8-g-9125-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.15 h; power.draw mean 166.0 W (max 167.8, limit 215 W; no power average on Turing), SM 1,955 MHz mean, memory 6,802 MHz, 64 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (19.12 MH/s, fingerprint matched); read ceiling 30.1 MH/s, the hash at 0.63 of it: Turing does not overlap the independent reads the hash needs (its indep probe 2.45 G loads/s binds). The Turing per-watt floor: a 4060's rate at twice the power. Class v4 watts (8 October 2026, the rerun, self-test PASS): 204.2 W at 19.1 MH/s, 0.094 MH per watt; the class v3 pack on the same host read 165.9 W at the same rate, so the shadow premium on this card is +23.1 percent of watts for no rate. the same board as the morning row; the first Turing pair, the smallest premium of the uncapped boards because Turing's SMs were busy under class v3 already (0.63 of its read ceiling); a 2060-class rate at a 3080-class draw: an owner mines at a loss on any real power price</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 4060 (8 GB)">NVIDIA RTX 4060 (8 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="19.09">19.1</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d33">Details</button></td></tr><tr class="detail" id="bench-current-d33" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.91.07, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 18 (a rented card, result cb2-rtx-4060-8-gb-0672-result.json; 09:53Z); the class v4 rerun, 8 October 2026 (result cb4-rtx-4060-8-gb-7165-result.json: the rate on both classes, no watts) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> watts not reported by any rented host (four 4060 boards across 7 and 8 October, one machine family, every power field absent; two other hosts never answered ssh); the class v4 rerun on 8 October read the same rate on both classes (19.09 MH/s class v4, 19.094 class v3) and still no watts; by the 4060 Ti pair on the same silicon (+33.6 percent) the class v4 draw is about 80 to 90 W from a class v3 draw of about 60 to 70 W under the 115 W limit, an estimate until an owner measures at the wall.12 h; SM 2,732 MHz mean, memory 8,251 MHz, 57 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (19.11 MH/s, fingerprint matched); read ceiling 30.2 MH/s, the hash at 0.63 of it (the 128-bit bus: the independent-read line binds, as on the 2070 Super). Repeats the 7 October row (19.09 then, 19.10 now)</span></td></tr><tr class="row"><td class="lead" data-v="AMD Radeon RX 9070 XT (16 GB)">AMD Radeon RX 9070 XT (16 GB)<span class="gen">class v4 +2 percent rate <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="18.96">19.0 stock (19.0 with the AMD knob)</td><td class="tnum effc" data-v="0.127"><span class="eff">0.127</span></td><td class="quiet tnum" data-v="149.3">149.3 with the AMD knob (195.8 stock) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d34">Details</button></td></tr><tr class="detail" id="bench-current-d34" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (the AMD knob is the app’s (core -500 MHz, power -30 percent); no Hive flight-sheet values yet)</span><span class="d"><b>Class v4 cost</b> +2 percent of rate (19.29 against 18.92 MH/s) at 102,100 ops per hash, watts owed, measured 6 October 2026</span><span class="d"><b>Tuned</b> core -500 MHz, power -30 percent, with the AMD knob: 2,921 MHz, 149.3 W at 18.96 MH/s, a 24 percent saving at the same rate (the rate flat at 18.93 to 18.98 MH/s on all 24 points, the knob moves watts only); measured 8 October 2026</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> Adrenalin 26.9.2, Windows 11</span><span class="d"><b>Source</b> the chip-model measurement record: the RX 9070 XT's first measured grid on the AMD knob (job run-ca3-pc1-amd-grid-9070-20261008-b, 24 of 24 points, 8 October 2026) for the rate and the watts; the 9070 XT Ember tune row (run-ca3-pc1-ember-9070-20261007) for the app's own stock reading · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> the grid's stock point reads 3,292 MHz, 195.8 W, 0.097 MH per watt (the app's own stock reading of 7 October was 202 W); the clock offset alone reaches 2,924 MHz at 159.2 W and the power limit alone 184.4 W; the best point is both together. Against the 5090 at its class v4 1,200 MHz lock (133.8 MH/s at 305 W, 0.43 MH per watt) the 9070 XT at its best point sits 3.5x behind per watt; against the 5090 at its 1,300 MHz knee (134.6 at 223 W, 0.60) about a fifth</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA GTX 1080 Ti (11 GB)">NVIDIA GTX 1080 Ti (11 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="16.26">16.3</td><td class="tnum effc" data-v="0.08"><span class="eff">0.08</span></td><td class="quiet tnum" data-v="202.7">202.7 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d35">Details</button></td></tr><tr class="detail" id="bench-current-d35" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card; Pascal has no clock lock; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented; Pascal has no clock lock on any host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.173.02, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, rows 17 and 22 (two rented hosts, results cb2-gtx-1080-ti-11-gb-0377-result.json and cb2-gtx-1080-ti-11-gb-1452-result.json; 09:49Z and 10:02Z); the class v4 watts rerun, 8 October 2026 (result cb4-gtx-1080-ti-11-gb-9935-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> the mean of the first host's two clean runs (16.26 MH/s at 124.3 W, the card uncapped at a 280 W limit); a second host read 16.29 MH/s at 129.4 W under a 130 W host cap with five of five self-tests passing, so the first host's one-in-three cache self-test failure was that host's VRAM and not Pascal, and the flag is cleared. Power.draw on the first host max 125.5 W (no power average on Pascal), SM 1,936 MHz mean, memory 5,005 MHz, 59 C; the class v5 kit held the rate on the uncapped host (16.27 MH/s) and read 13.27 MH/s on the 130 W-capped one (fingerprint matched on both); read ceiling 16.3 MH/s, the hash at 1.00 of it. The first Pascal row: a 2017 card that runs the class at its own ceiling. Class v4 watts (8 October 2026, the rerun, self-test PASS): 202.7 W at 16.29 MH/s, 0.08 MH per watt; the class v3 pack on the same host read 124.5 W at the same rate, so the shadow premium on this card is +62.8 percent of watts for no rate. an uncapped 280 W board (five of five self-tests PASS, so the first host's cache fault stays that host's); the largest premium of the pairs, pure added work (the SM clock did not move between classes): Pascal has no fast integer path for the shadow stage and ran class v3 at half its power limit; the worst MH per watt of every card measured, a 2017 card that cannot pay its own power on any tariff</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX 2060 (6 GB)">NVIDIA RTX 2060 (6 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="14.13">14 to 15 by board (14.90 on this 12 GB board)</td><td class="tnum effc" data-v="0.117"><span class="eff">0.117</span></td><td class="quiet tnum" data-v="127">127 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d36">Details</button></td></tr><tr class="detail" id="bench-current-d36" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.159.03, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 19 (a rented card, result cb2-rtx-2060-0673-result.json; 09:55Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-2060-0819-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.15 h; power.draw mean 120.6 W (max 124.6, limit 160 W; no power average on Turing), SM 1,948 MHz mean, memory 6,801 MHz, 69 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (14.14 MH/s, fingerprint matched); read ceiling 16.1 MH/s, the hash at 0.88 of it. The floor of the list and the smallest card on it: the class v4 set fits with 4.6 GB spare; the same Turing per-watt floor as the 2070 Super and the 2080 Ti, so every Turing card earns the same per watt and only the absolute rate differs. Class v4 watts (8 October 2026, the rerun, self-test PASS): 127 W at 14.9 MH/s, 0.117 MH per watt; the class v3 pack on the same host read 124.9 W at the same rate, so the shadow premium on this card is +1.7 percent of watts for no rate. the 12 GB variant on a 130 W host cap (stock 185), 5 percent over the morning's 6 GB board (its memory runs faster), so the rate reads as a range; the cap absorbs the premium with the SM giving 7 percent of its clock; the Turing floor under class v4 is 0.09 to 0.12 MH per watt: every Turing card mines at a loss on a real tariff</span></td></tr><tr class="row"><td class="lead" data-v="Intel Arc B580 (12 GB)">Intel Arc B580 (12 GB)<span class="gen">class v4 0.1 percent rate <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="11">11</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-current-d37">Details</button></td></tr><tr class="detail" id="bench-current-d37" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> 0.1 percent of rate, measured 7 October 2026</span><span class="d"><b>Tuned</b> stock, bench only</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> Intel driver, Windows 11</span><span class="d"><b>Source</b> bench log: 7 October 2026, the Intel Arc B580 (eGPU equals slot) · <b>Who</b> measured by the team</span><span class="d"><b>Note</b> the same rate in the Thunderbolt enclosure and the slot; watts not read on this run. Re-read 8 October 2026: the installed OpenCL worker on this card passes its own self-test with the devnet pack (96 of 96 vector lanes) and mined 54 CPU-re-checked blocks at 10.58 MH/s, so the row stands; a kit build from a branch without the Intel rotate-fold rewrite failed its self-test on this card the same morning, a build fault and not an Arc result</span></td></tr></tbody></table></div>
<p class="gen">38 cards. Details opens a card's Hive flight-sheet values, its full class v4 cost, source and note.</p>
<details class="grp"><summary>Datacentre (13 cards, rented for the measurement; about three times the rented dollars per hash of a desktop card)</summary>
<div class="tbl bench2wrap"><table class="bench2" id="bench-datacentre"><thead><tr><th data-key="card" data-type="text" aria-sort="none"><button type="button" class="sort">Card</button></th><th data-key="mh_s" data-type="num" aria-sort="descending"><button type="button" class="sort">MH/s</button></th><th data-key="mh_per_w" data-type="num" aria-sort="none"><button type="button" class="sort">MH per wall watt</button></th><th data-key="watts" data-type="num" aria-sort="none"><button type="button" class="sort">Watts</button></th><th data-key="tuned" data-type="text" aria-sort="none"><button type="button" class="sort">Tuned</button></th><th></th></tr></thead><tbody><tr class="row"><td class="lead" data-v="NVIDIA B200 (180 GB)">NVIDIA B200 (180 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="416.35">416.4</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.487</span></td><td class="quiet tnum" data-v="855.6">855.6 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d0">Details</button></td></tr><tr class="detail" id="bench-datacentre-d0" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA H200 SXM (141 GB)">NVIDIA H200 SXM (141 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="313.04">313</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.723</span></td><td class="quiet tnum" data-v="432.9">432.9 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d1">Details</button></td></tr><tr class="detail" id="bench-datacentre-d1" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA H100 SXM (80 GB)">NVIDIA H100 SXM (80 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="248.7">248.7</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.645</span></td><td class="quiet tnum" data-v="385.6">385.6 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d2">Details</button></td></tr><tr class="detail" id="bench-datacentre-d2" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.126.09, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep, a rented card) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> 424.0 W maximum; 98 percent of its random-read ceiling like the 5090; 1.78x the 5090's hash at 1.15x the tuned 5090's hash per watt and a third of the hash per rented dollar. Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA H100 SXM (80 GB HBM3)">NVIDIA H100 SXM (80 GB HBM3)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="248.3">248.3</td><td class="tnum effc" data-v="0.36"><span class="eff">0.36</span></td><td class="quiet tnum" data-v="695.1">695.1 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d3">Details</button></td></tr><tr class="detail" id="bench-datacentre-d3" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580.126.09, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 14 (a rented card, result cb2-h100-80-gb-rs3d-result.json; 09:43Z); the class v4 watts rerun, 8 October 2026 (result cb4-h100-80-gb-sbwo-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.03 h; power.draw mean 411.7 W (instant 434.1, the largest gap between the two fields in the sweep, 22 W; max 437.1, limit 700 W), SM 1,980 MHz, HBM3 2,619 MHz, 36 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit reads 252.64 MH/s, 1.5 percent over class v4 (fingerprint matched); read ceiling 252.4 MH/s, the hash at 0.99 of it. Repeats the 7 October rows (248.70, 248.21) on a third host; the best per-watt figure of the sweep on power.draw (0.573 on the instant field) and the worst MH per rented dollar: an owner's card on cheap power, never a renter's. Class v4 watts (8 October 2026, the rerun, self-test PASS): 695.1 W at 250.02 MH/s, 0.36 MH per watt; the class v3 pack on the same host read 456.9 W at the same rate, so the shadow premium on this card is +52.1 percent of watts for no rate. this board draws 45 W more than the morning's under class v3 at 1 percent more rate; under class v4 the H100 sits on its 700 W cap (SM 1,774 MHz, down from 1,980) and holds the rate; still the worst MH per rented dollar</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA A100 PCIe (80 GB)">NVIDIA A100 PCIe (80 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="154.97">155</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.517</span></td><td class="quiet tnum" data-v="299.6">299.6 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d4">Details</button></td></tr><tr class="detail" id="bench-datacentre-d4" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA A100 SXM (80 GB)">NVIDIA A100 SXM (80 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="138.39">138.4</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.52</span></td><td class="quiet tnum" data-v="266.3">266.3 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d5">Details</button></td></tr><tr class="detail" id="bench-datacentre-d5" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA A100 SXM4 (80 GB)">NVIDIA A100 SXM4 (80 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="138.1">138.1</td><td class="tnum effc" data-v="0.282"><span class="eff">0.282</span></td><td class="quiet tnum" data-v="489.3">489.3 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d6">Details</button></td></tr><tr class="detail" id="bench-datacentre-d6" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.71.05, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 13 (a rented card, result cb2-a100-80-gb-j79g-result.json; 09:43Z); the class v4 watts rerun, 8 October 2026 (result cb4-a100-80-gb-mdfz-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.04 h; power.draw mean 268.6 W (max 274.1, limit 400 W; no power average on this host), SM 1,410 MHz, HBM2e 1,593 MHz, 44 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit reads 136.48 MH/s, 1.3 percent under class v4 (fingerprint matched), the first card where v5 is not level with v4; read ceiling 138.6 MH/s, the hash at 1.00 of it. Repeats the three 7 October readings (138.39, 138.74, 138.76) to 0.3 percent on a fourth host; datacentre-class per watt, a third of a 3080 Ti's MH per rented dollar. Class v4 watts (8 October 2026, the rerun, self-test PASS): 489.3 W at 138 MH/s, 0.282 MH per watt; the class v3 pack on the same host read 274.3 W at the same rate, so the shadow premium on this card is +78.4 percent of watts for no rate. the 500 W SKU (the morning's board was the 400 W SKU; the class v3 draw holds across both); the largest premium of the pairs so far: the 108 SMs sit nearly idle under class v3 and the shadow stage puts them all to work, the card touching its 500 W cap without losing rate; under the live class the A100 falls from datacentre-class per watt to the capped 4070 Super's level; on the 400 W SKU the cap would bind and the rate should hold by the Ampere rule (an inference until a 400 W board is paired)</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX PRO 6000 Blackwell (96 GB)">NVIDIA RTX PRO 6000 Blackwell (96 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="130.49">130.5</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.452</span></td><td class="quiet tnum" data-v="288.7">288.7 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d7">Details</button></td></tr><tr class="detail" id="bench-datacentre-d7" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA L40S (48 GB)">NVIDIA L40S (48 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="56.41">56.4</td><td class="tnum effc" data-v="0.203"><span class="eff">0.203</span></td><td class="quiet tnum" data-v="277.6">277.6 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d8">Details</button></td></tr><tr class="detail" id="bench-datacentre-d8" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 570.133.20, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 21 (a rented card, result cb2-l40s-48-gb-0890-result.json; 09:57Z); the class v4 watts rerun, 8 October 2026 (result cb4-l40s-48-gb-1574-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.17 h; power.draw mean 220.7 W (max 227.2, limit 350 W), SM 2,520 MHz, memory 9,001 MHz, 45 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (56.51 MH/s, fingerprint matched); read ceiling 57.8 MH/s, the hash at 0.98 of it. Repeats the 7 October row (56.36 then, 56.41 now) at 20 W less on this host; a 384-bit GDDR6 card that lands where the 3090 Ti and the A6000 do at a datacentre price. Driver 570.133, the oldest of the sweep, ran the CUDA 12.8 worker clean: 570 is a safe floor. Class v4 watts (8 October 2026, the rerun, self-test PASS): 277.6 W at 56.4 MH/s, 0.203 MH per watt; the class v3 pack on the same host read 219.3 W at the same rate, so the shadow premium on this card is +26.6 percent of watts for no rate. a sibling board of the morning's host, uncapped; the smallest premium of the uncapped Ada boards, the 142 SMs running the shadow stage inside their idle margin without moving the clock; level with the 4070 Ti and the capped 3080 Ti per watt, at a datacentre price</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA L40S (48 GB)">NVIDIA L40S (48 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="56.36">56.4</td><td class="tnum effc" data-v="-1"><span class="hollow" title="class v3 watts: the fleet's bench pack ran the class v3 program; the class v4 watts land with the rerun; not ranked">○ 0.234</span></td><td class="quiet tnum" data-v="240.7">240.7 (class v3) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d9">Details</button></td></tr><tr class="detail" id="bench-datacentre-d9" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, not tuned)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 580, Ubuntu 24.04</span><span class="d"><b>Source</b> bench log: 7 October 2026, every rentable card on the hash (the fleet's sweep on rented cards; the hive package, Ubuntu 24.04) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> . Watts are class v3 (read 8 October 2026, 11:4x UK): the fleet's bench pack was the class v3 program (program_class v3, generator 3, load_class mx8-erad810f22d), so this row's watts and MH per watt are class v3 figures; the rate holds under class v4 (memory-bound: a 5070 Ti measured 78.7 MH/s at 141 W on class v3 and 78.8 at 224 W on class v4). The class v4 watts rerun on the same models lands with its date, the v3 watts kept beside as the shadow premium per card</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX A6000 (48 GB)">NVIDIA RTX A6000 (48 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="53.64">53.6</td><td class="tnum effc" data-v="0.182"><span class="eff">0.182</span></td><td class="quiet tnum" data-v="295">295 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d10">Details</button></td></tr><tr class="detail" id="bench-datacentre-d10" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.84, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 16 (a rented card, result cb2-rtx-a6000-48-gb-9600-result.json; 09:46Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-a6000-48-gb-1814-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.15 h; power.draw mean 239.1 W (max 251.0, limit 300 W), SM 1,913 MHz mean, memory 7,601 MHz, 65 C, no throttle reason; self-test PASS (96 of 96 vector lanes); the class v5 kit holds the rate (53.82 MH/s, fingerprint matched), so on an uncapped sm_86 card v5 is level with v4 and the A4000's v5 reading narrows to its power cap; read ceiling 53.8 MH/s, the hash at 1.00 of it. Repeats the 7 October row (53.66 then, 53.64 now) at 43 W less on this host; a 3090-class rate at a 3090-class draw. Class v4 watts (8 October 2026, the rerun, self-test PASS): 295.0 W at 53.67 MH/s, 0.182 MH per watt; the class v3 pack on the same host read 253.9 W at the same rate (the board 9 C warmer than the morning, 15 W over its morning class v3 reading), so the shadow premium on this card is +16.2 percent of watts for no rate: the A6000 lands on its stock 300 W cap under class v4 and holds the rate, the fourth card to do so with the H100, the A100 and the A4000, so a stock cap reached under class v4 is a note, not a penalty. The rerun is complete: 22 class v4 watts rows on 22 models, every pod destroyed</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX A5000 (24 GB)">NVIDIA RTX A5000 (24 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-07</span></span></td><td class="big tnum" data-v="47.64">47.6</td><td class="tnum effc" data-v="-1">not measured</td><td class="quiet tnum" data-v="-1">not read</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d11">Details</button></td></tr><tr class="detail" id="bench-datacentre-d11" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (no measured tune point; the 5080 and 9070 XT passes and the class v4 efficiency pass land theirs when read)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, mining</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver, Ubuntu 24.04</span><span class="d"><b>Source</b> the fleet's standing voter (status line, 7 October 2026) · <b>Who</b> reported by the fleet</span><span class="d"><b>Note</b> a standing voter's status line, 18:00Z; watts not read</span></td></tr><tr class="row"><td class="lead" data-v="NVIDIA RTX A4000 (16 GB)">NVIDIA RTX A4000 (16 GB)<span class="gen">class v4 not measured <span class="dot">·</span> <span class="date">2026-10-08</span></span></td><td class="big tnum" data-v="31.41">31.4</td><td class="tnum effc" data-v="0.232"><span class="eff">0.232</span></td><td class="quiet tnum" data-v="135.2">135.2 (class v4) W</td><td class="pillc" data-v="Stock"><span class="pill state stock">Stock</span></td><td class="morec" data-v=""><button type="button" class="more" aria-expanded="false" aria-controls="bench-datacentre-d12">Details</button></td></tr><tr class="detail" id="bench-datacentre-d12" hidden><td colspan="6"><span class="d"><b>Hive flight sheet</b> stock (rented card, no clock control on the host; no measured tune point)</span><span class="d"><b>Class v4 cost</b> not measured (class v4 program only)</span><span class="d"><b>Tuned</b> stock, bench only (rented, no clock control on the host)</span><span class="d"><b>Generator</b> v2 · <b>Driver</b> NVIDIA driver 595.91.07, Ubuntu 24.04 (CUDA 12.8.1 image)</span><span class="d"><b>Source</b> model sweep 2026-10-08, row 15 (a rented card, result cb2-rtx-a4000-16-gb-9473-result.json; 09:45Z); the class v4 watts rerun, 8 October 2026 (result cb4-rtx-a4000-16-gb-0549-result.json, the v4-devnet-epoch0 pack, sub-version 3, three runs of about 30 s, the class v3 pack paired on the same host) · <b>Who</b> measured by the fleet</span><span class="d"><b>Note</b> rented single-card host, 0.16 h; power.draw mean 112.8 W on a host that caps the card at 114 W (stock 140 W; the SW power-cap bit set the whole run), SM 1,562 MHz mean, memory 6,501 MHz, 81 C; self-test PASS (96 of 96 vector lanes); read ceiling 31.6 MH/s, the hash at 0.99 of it. Class v4 repeats the 7 October row (31.45 then, 31.41 now) under the cap. The class v5 kit's reading on this capped host is held off the page until an uncapped A4000 confirms it. Class v4 watts (8 October 2026, the rerun, self-test PASS): 135.2 W at 31.36 MH/s, 0.232 MH per watt; the class v3 pack on the same host read 136.1 W at the same rate, so the shadow premium on this card is -0.7 percent of watts for no rate. a different board from the morning's 114 W-capped one, at its 140 W stock limit and hot (96 C, the thermal bit set under both classes): the A4000 is a capped card already under class v3 on every host, so under class v4 it draws the same 135 to 140 W and holds the full rate with the SM giving a quarter of its clock; the morning's class v5 reading on the 114 W board follows from the same cap. The best renter's card under the live class after the capped 3060 Ti</span></td></tr></tbody></table></div>
</details>
</section>
<section class="doc-section"><h2 id="how">How a row gets here</h2>
<p>Every row names the engineering log entry or the job it came from. "Measured by the team" means our own hardware and our own log. "Reported by the fleet" or "measured by the fleet" means a machine we rent or do not own, read from the status lines its miner uploads or from a bench run on it.</p>
<p>MH per watt needs the card's power draw during the run. The app reads it on NVIDIA cards through the driver. Rows get the figure when a run records it. The class v4 cost column is what the latency-shadow work costs that card against the class v3 control, in watts and rate, where it was measured. The tuned column is the card's state at the row: a full Ember Tune names its point; stock means the card as it came, bench only or mining.</p>
<p>There is no other Igneum miner to compare with yet, so this table compares cards, not miners. The app that produces these rows: <a href="/miner">the miner page</a>.</p>
<script>
(function(){
document.querySelectorAll('table.bench2').forEach(function(t){
var ths=t.querySelectorAll('th[data-key]'),tb=t.querySelector('tbody');
ths.forEach(function(th){var i=Array.prototype.indexOf.call(th.parentNode.children,th);th.querySelector('button').addEventListener('click',function(){
var cur=th.getAttribute('aria-sort'),dir=cur==='descending'?'ascending':'descending',num=th.getAttribute('data-type')==='num';
ths.forEach(function(o){o.setAttribute('aria-sort','none');});th.setAttribute('aria-sort',dir);
var rows=Array.prototype.slice.call(tb.querySelectorAll('tr.row'));
var pairs=rows.map(function(r){return [r, r.nextElementSibling && r.nextElementSibling.classList.contains('detail') ? r.nextElementSibling : null];});
pairs.sort(function(a,b){var x=a[0].children[i].getAttribute('data-v'),y=b[0].children[i].getAttribute('data-v');var c=num?(parseFloat(x)-parseFloat(y)):x.localeCompare(y,'en');return dir==='descending'?-c:c;});
pairs.forEach(function(p){tb.appendChild(p[0]); if(p[1]) tb.appendChild(p[1]);});
});});
t.querySelectorAll('button.more').forEach(function(b){b.addEventListener('click',function(){var d=document.getElementById(b.getAttribute('aria-controls'));if(!d)return;var open=d.hasAttribute('hidden');if(open)d.removeAttribute('hidden');else d.setAttribute('hidden','');b.setAttribute('aria-expanded',open?'true':'false');b.textContent=open?'Close':'Details';});});
});
})();
</script>
</section>
<section class="doc-section"><h2 id="priors">Fleet tuning priors</h2>
<p>Ember Tune runs on every card the app mines with: the power limit and the core clock are stepped on the live program and the card keeps the point with the best MH per watt within 1% of its top rate. Every finished tune is reported back without anything that identifies the owner, and the fleet's median point per card model, driver major and program class comes back down inside the signed update manifest as the starting point for the next card of that model. A model needs 5 reports before its prior is used.</p>
<p>No tune reports yet. The first rows appear once five machines with the same card model have reported.</p>
<h3 id="team">Measured by the team</h3>
<ul><li><b>NVIDIA GeForce RTX 5090</b> (driver 617, class l128w0, 2026-10-06, job <code>ember-tune-pc1-6</code>): 1854 MHz core cap, power limit 100% (575 W). 84 W saved for 0.15% of rate: 127.71 MH/s at 226.8 W (0.563 MH/W) against 127.9 MH/s at 311 W untuned (0.411 MH/W), 37% more hashes per watt. The chosen clock is the ladder's floor (60% of the maximum), not the optimum; the next cut's ladder goes to 45% and the climb walks on from here.</li><li><b>NVIDIA GeForce RTX 4070</b> (driver 617, class l128w0, 2026-10-06, job <code>ember-tune-pc1-6</code>): 1863 MHz core cap, power limit 50% (100 W). 30 W saved for no rate lost: 28.78 MH/s at 75.6 W (0.381 MH/W) against 28.72 MH/s at 106 W untuned (0.271 MH/W), 41% more hashes per watt. The chosen clock is the ladder's floor (60% of the maximum), not the optimum; the next cut's ladder goes to 45%.</li></ul>
<p>Rows: 0. Written from the fleet's tune records at build time.</p></section></article>
</div>
<p class="gen">Generated from the repository at build time. Times are UTC. Machine names are model names.</p>
</div></section>
</main>
<!-- footer:start -->
<footer class="site-footer">
<div class="container">
<p class="positioning">A GPU-secured network for Ethereum-compatible applications and verifiable computation.</p>
<div class="footer-grid">
<div class="footer-brand">
<a href="/" class="logo" aria-label="Igneum home"><svg class="brand-mark" viewBox="0 0 1024 1024" aria-hidden="true"><rect width="1024" height="1024" rx="160" fill="#0C0C0E"></rect><g transform="translate(166.95 166.95) scale(6.901)"><polygon points="50,4 74,34 67,58 80,54 61,96 39,96 20,54 33,58 26,34" fill="#F2541B"></polygon><polygon points="50,42 59,58 50,82 41,58" fill="#0C0C0E"></polygon></g></svg><span class="word">IGNEUM</span><span class="stop">.</span></a>
<p>Mined by GPUs. Proven by fire.</p>
<div class="social" aria-label="Igneum elsewhere">
<a href="https://git.igneum.network/igneum-network/spec" target="_blank" rel="noopener" aria-label="Igneum source: the spec, the vectors and the issues"><svg viewBox="0 0 24 24" width="22" height="22" fill="currentColor" aria-hidden="true"><path d="M8.6 17.6 3 12l5.6-5.6 1.4 1.4L5.8 12l4.2 4.2zm6.8 0-1.4-1.4L18.2 12 14 7.8l1.4-1.4L21 12z"></path></svg></a>
<a href="https://discord.gg/igneum" target="_blank" rel="noopener" aria-label="The Igneum Discord"><svg viewBox="0 0 24 24" width="22" height="22" fill="currentColor" aria-hidden="true"><path d="M19.6 5.3A17 17 0 0 0 15.4 4l-.2.4a15.6 15.6 0 0 1 3.9 1.9 13.6 13.6 0 0 0-14.2 0A15.6 15.6 0 0 1 8.8 4.4L8.6 4a17 17 0 0 0-4.2 1.3C1.8 9.2 1.1 13 1.4 16.7a17.1 17.1 0 0 0 5.2 2.6l1.1-1.8a10.8 10.8 0 0 1-1.7-.8l.4-.3a12.2 12.2 0 0 0 11.2 0l.4.3-1.7.8 1.1 1.8a17 17 0 0 0 5.2-2.6c.4-4.3-.7-8-3-11.4zM8.7 14.4c-1 0-1.8-.9-1.8-2.1s.8-2.1 1.8-2.1 1.9 1 1.8 2.1c0 1.2-.8 2.1-1.8 2.1zm6.6 0c-1 0-1.8-.9-1.8-2.1s.8-2.1 1.8-2.1 1.9 1 1.8 2.1c0 1.2-.8 2.1-1.8 2.1z"></path></svg></a>
<a data-social="reddit" href="https://www.reddit.com/user/Igneum_network/" target="_blank" rel="noopener" aria-label="Igneum on Reddit"><svg viewBox="0 0 24 24" width="22" height="22" fill="currentColor" aria-hidden="true"><path d="M22 12.1a2.2 2.2 0 0 0-3.7-1.6 10.8 10.8 0 0 0-5.8-1.8l1-4.6 3.2.7a1.5 1.5 0 1 0 .2-.9l-3.6-.8a.5.5 0 0 0-.5.4l-1.1 5.2a10.8 10.8 0 0 0-5.9 1.8A2.2 2.2 0 1 0 3.4 14a4.3 4.3 0 0 0 0 .7c0 3.4 3.9 6.1 8.6 6.1s8.6-2.7 8.6-6.1a4.3 4.3 0 0 0 0-.7 2.2 2.2 0 0 0 1.4-1.9zM7 13.6a1.5 1.5 0 1 1 1.5 1.5A1.5 1.5 0 0 1 7 13.6zm8.6 4.1a5.7 5.7 0 0 1-3.6 1.1 5.7 5.7 0 0 1-3.6-1.1.4.4 0 0 1 .6-.6 4.9 4.9 0 0 0 3 .9 4.9 4.9 0 0 0 3-.9.4.4 0 0 1 .6.6zm-.3-2.6a1.5 1.5 0 1 1 1.5-1.5 1.5 1.5 0 0 1-1.5 1.5z"></path></svg></a>
</div>
<div class="footer-dl" aria-label="Download Ember"><a href="/download#windows"><span class="osmark mini" data-os="windows" title="Windows"><svg viewBox="0 0 24 24" width="22" height="22" aria-hidden="true" focusable="false" fill="currentColor"><path d="M3 5.6l7.3-1v7.1H3zM11.4 4.4L21 3v8.7h-9.6zM3 12.3h7.3v7.1L3 18.4zM11.4 12.3H21V21l-9.6-1.4z"/></svg></span>Windows</a><a href="/download#mac"><span class="osmark mini" data-os="mac" title="macOS"><svg viewBox="0 0 24 24" width="22" height="22" aria-hidden="true" focusable="false" fill="currentColor"><path d="M16.4 12.6c0-2.5 2-3.6 2.1-3.7-1.2-1.7-3-1.9-3.6-2-1.5-.2-3 .9-3.8.9-.8 0-2-.9-3.3-.8-1.7 0-3.2 1-4.1 2.5-1.8 3-.5 7.6 1.3 10.1.9 1.2 1.9 2.6 3.2 2.5 1.3 0 1.8-.8 3.3-.8 1.6 0 2 .8 3.3.8 1.4 0 2.3-1.2 3.1-2.5 1-1.4 1.4-2.8 1.4-2.9 0 0-2.7-1-2.9-4.1zM13.9 5.3c.7-.8 1.2-2 1-3.2-1 0-2.2.7-2.9 1.5-.6.7-1.2 1.9-1 3 1.1.1 2.2-.5 2.9-1.3z"/></svg></span>macOS</a><a href="/download#linux"><span class="osmark mini" data-os="linux" title="Linux"><svg viewBox="0 0 24 24" width="22" height="22" aria-hidden="true" focusable="false" fill="currentColor"><path d="M12 2c-2.4 0-4 1.9-4 4.6 0 1.2.2 2 0 2.8-.6 1.4-2.1 2.9-2.6 4.9-.3 1.1-.1 2 .3 2.6-.6.4-1.3 1-1.1 1.7.3 1 2.1 1.2 3.2 1.8.7.4 1.5.6 2.1.1.6.2 1.3.3 2.1.3s1.5-.1 2.1-.3c.6.5 1.4.3 2.1-.1 1.1-.6 2.9-.8 3.2-1.8.2-.7-.5-1.3-1.1-1.7.4-.6.6-1.5.3-2.6-.5-2-2-3.5-2.6-4.9-.2-.8 0-1.6 0-2.8C16 3.9 14.4 2 12 2zm-1.4 4.2c.5 0 .8.5.8 1.2s-.3 1.2-.8 1.2-.8-.5-.8-1.2.3-1.2.8-1.2zm2.8 0c.5 0 .8.5.8 1.2s-.3 1.2-.8 1.2-.8-.5-.8-1.2.3-1.2.8-1.2zM12 9.3c.9 0 1.9.5 1.9 1s-1 1.2-1.9 1.2-1.9-.7-1.9-1.2 1-1 1.9-1zm0 3.4c2.2 0 3.6 2.6 3.6 4.4 0 1.5-1.6 2.3-3.6 2.3s-3.6-.8-3.6-2.3c0-1.8 1.4-4.4 3.6-4.4z"/></svg></span>Linux</a><a href="/download#hive"><span class="osmark mini" data-os="hive" title="HiveOS"><svg viewBox="0 0 24 24" width="22" height="22" aria-hidden="true" focusable="false" fill="none" stroke="currentColor" stroke-width="1.7" stroke-linejoin="round" stroke-linecap="round"><path d="M12 2.6 20.2 7.3v9.4L12 21.4 3.8 16.7V7.3z"/><path d="M12 7.4 16 9.7v4.6L12 16.6 8 14.3V9.7z"/><path d="M12 7.4V2.6M16 9.7l4.2-2.4M16 14.3l4.2 2.4M12 16.6v4.8M8 14.3l-4.2 2.4M8 9.7 3.8 7.3"/></svg></span>HiveOS</a></div>
<a href="https://git.igneum.network/igneum-network/spec" target="_blank" rel="noopener" class="text-link">Specification and test vectors<svg class="icon" viewBox="0 0 24 24" aria-hidden="true"><path d="M6 18 18 6M6 6h12v12"/></svg></a>
</div>
<div class="footer-column"><h4>Run</h4><div>
<a href="/download">Download Ember</a>
<a href="/miner">The miner</a>
<a href="/app">The app</a>
<a href="/wallet">The wallet</a>
<a href="/miners">GPU bench table</a>
<a href="/dev-fee">The dev fee</a>
<a href="/faucet">Devnet faucet</a>
<a href="/metamask">Add Igneum to MetaMask</a>
</div></div>
<div class="footer-column"><h4>Read</h4><div>
<a href="/litepaper">Litepaper</a>
<a href="/evidence">Evidence</a>
<a href="/ledger">Ledger: every criticism</a>
<a href="/claims">What Igneum does not claim</a>
<a href="/randomx">Igneum vs RandomX</a>
<a href="/provenance">Built on the shoulders</a>
</div></div>
<div class="footer-column"><h4>Explore</h4><div>
<a href="/live">Live devnet</a>
<a href="/explorer">Explorer</a>
<a href="/build">Build on Igneum</a>
<a href="/swap">Swap</a>
<a href="/grants">Grants</a>
<a href="/download">Downloads: devnet build</a>
<a href="https://discord.gg/igneum" target="_blank" rel="noopener">Community Discord</a>
<a href="mailto:hello@igneum.network">hello@igneum.network</a>
<a href="https://git.igneum.network/igneum-network/spec/issues" rel="noopener">An issue on the spec</a>
</div></div>
</div>
<div class="footer-bottom">
<p>Igneum Labs LTD · Unit IH-00-01-01-OF-01, Level 01, Innovation One, Dubai International Financial Centre. Experimental software. Nothing on this site is an offer to sell anything.<br>Report a flaw by email or by an issue on the spec. Nobody from Igneum will ask for your seed. © 2026 Igneum · not an offer to sell anything</p>
<div class="theme-group" role="group" aria-label="Theme"><button type="button" data-theme-set="system" aria-pressed="true">System</button><button type="button" data-theme-set="light" aria-pressed="false">Light</button><button type="button" data-theme-set="dark" aria-pressed="false">Dark</button></div>
</div>
</div>
</footer>
<script>
(function(){
// theme control: stores light or dark under igneum-theme; system clears it and follows the OS. The head applied the value before paint.
var bs=document.querySelectorAll('[data-theme-set]');if(!bs.length)return;
var media=window.matchMedia('(prefers-color-scheme: light)');
function stored(){try{var t=localStorage.getItem('igneum-theme');return t==='light'||t==='dark'?t:'system';}catch(e){return 'system';}}
function apply(){var c=stored();document.documentElement.setAttribute('data-theme',c==='system'?(media.matches?'light':'dark'):c);bs.forEach(function(b){b.setAttribute('aria-pressed',b.getAttribute('data-theme-set')===c?'true':'false');});document.dispatchEvent(new CustomEvent('igneum-theme'));}
bs.forEach(function(b){b.addEventListener('click',function(){var v=b.getAttribute('data-theme-set');try{if(v==='system')localStorage.removeItem('igneum-theme');else localStorage.setItem('igneum-theme',v);}catch(e){}apply();});});
if(media.addEventListener)media.addEventListener('change',function(){if(stored()==='system')apply();});
document.addEventListener('igneum-theme',function(){var c=stored();bs.forEach(function(b){b.setAttribute('aria-pressed',b.getAttribute('data-theme-set')===c?'true':'false');});});
bs.forEach(function(b){b.setAttribute('aria-pressed',b.getAttribute('data-theme-set')===stored()?'true':'false');});
// the network state word in the status bar: one read of the observer per page view
var fs=document.getElementById('foot-state'),fd=document.getElementById('foot-dot');if(fs&&fd&&fs.getAttribute('data-devnet-state')==='live'){fetch('/api/live?window=30',{cache:'no-store'}).then(function(r){return r.json();}).then(function(d){var on=d&&d.ok&&d.state&&!d.state.stale;fs.textContent=on?'devnet live':'devnet, observer stale';fd.className='dot '+(on?'live':'off');}).catch(function(){fs.textContent='devnet';});}
})();
</script>
<!-- footer:end -->
<script src="/ui.js" defer></script>
</body>
</html>