class v5: the verifier measured on igneum-build-1 (tools/class-v5/verify-bench-remote.sh, the ladder's method, box at load 70): class v5 adds 0.28 ms cold and 0.20 ms on average over class v4 on the loaded reference core (8.74 to 9.03 ms cold with the sibling loaded), both under the 10 ms gate; the raw lines under docs/design/class-v5-bench; the page's time line with the fork suites and the packs test

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

docs/design/class-v5-bench/20261007T083716Z/alone-v4.txt

@@ -0,0 +1,8 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 540.7 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+program: class mx8+sh256x27, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 536.5 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 8.110 ms, 4096 items derived, lane0 2576769ee4a14c8d lane31 c58ddcb717dd3370
+warp base 4096: single cold run 7.636 ms, 4096 items derived, lane0 1ce77a600ec573b4 lane31 03600a05ffba0055
+warp base 1000000: single cold run 7.620 ms, 4096 items derived, lane0 6b390e64bbdd91ce lane31 91c944d603539c62
+CPU verify: 7.830 ms per 32-lane warp, avg of 50 (checksum 17e36e7905b81375)

docs/design/class-v5-bench/20261007T083716Z/alone-v5.txt

@@ -0,0 +1,9 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 551.4 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+state stream /srv/builds/_log/v5-class/node1-state.igsd1: chain block 159357 af89be5ddbadb6f6b4aee28ac8f249713be5d4c12621e3cea7f83ceada3c66b3, root 1c583d352bb9c75a06dadb8d82d42836ebe8afa82d0b413be87bf921741f1526, 93 records, 93 leaves
+program: class mx8+sh256x27+state, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 555.1 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 8.712 ms, 4096 items derived, lane0 18d47ad5e5f0db00 lane31 d1b36ed0ff68e575
+warp base 4096: single cold run 8.478 ms, 4095 items derived, lane0 a1f55f0b0cae0ed2 lane31 8c548231ef24eb73
+warp base 1000000: single cold run 8.549 ms, 4096 items derived, lane0 bd47078fc688b540 lane31 01f4d9a0bef81008
+CPU verify: 8.538 ms per 32-lane warp, avg of 50 (checksum 60bcc82b11d41f0f)

docs/design/class-v5-bench/20261007T083716Z/loaded-v4.txt

@@ -0,0 +1,8 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 589.9 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+program: class mx8+sh256x27, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 556.0 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 8.744 ms, 4096 items derived, lane0 2576769ee4a14c8d lane31 c58ddcb717dd3370
+warp base 4096: single cold run 8.694 ms, 4096 items derived, lane0 1ce77a600ec573b4 lane31 03600a05ffba0055
+warp base 1000000: single cold run 8.738 ms, 4096 items derived, lane0 6b390e64bbdd91ce lane31 91c944d603539c62
+CPU verify: 8.629 ms per 32-lane warp, avg of 50 (checksum 17e36e7905b81375)

docs/design/class-v5-bench/20261007T083716Z/loaded-v5.txt

@@ -0,0 +1,9 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 566.6 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+state stream /srv/builds/_log/v5-class/node1-state.igsd1: chain block 159357 af89be5ddbadb6f6b4aee28ac8f249713be5d4c12621e3cea7f83ceada3c66b3, root 1c583d352bb9c75a06dadb8d82d42836ebe8afa82d0b413be87bf921741f1526, 93 records, 93 leaves
+program: class mx8+sh256x27+state, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 588.0 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 9.025 ms, 4096 items derived, lane0 18d47ad5e5f0db00 lane31 d1b36ed0ff68e575
+warp base 4096: single cold run 8.877 ms, 4095 items derived, lane0 a1f55f0b0cae0ed2 lane31 8c548231ef24eb73
+warp base 1000000: single cold run 8.955 ms, 4096 items derived, lane0 bd47078fc688b540 lane31 01f4d9a0bef81008
+CPU verify: 8.833 ms per 32-lane warp, avg of 50 (checksum 60bcc82b11d41f0f)

docs/design/class-v5-bench/20261007T083716Z/meta.txt

@@ -0,0 +1,2 @@
+host igneum-build-1 load 70.78 56.14 34.92 freq40 3724858 kHz siblings 40,88 pow 978c9a64d265746f state abb5800350b02dd9
+load after 69.49 56.35 35.22

docs/design/class-v5-bench/20261007T083716Z/sibling-v4.txt

@@ -0,0 +1,8 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 546.6 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+program: class mx8+sh256x27, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 542.9 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 8.165 ms, 4096 items derived, lane0 2576769ee4a14c8d lane31 c58ddcb717dd3370
+warp base 4096: single cold run 8.058 ms, 4096 items derived, lane0 1ce77a600ec573b4 lane31 03600a05ffba0055
+warp base 1000000: single cold run 8.080 ms, 4096 items derived, lane0 6b390e64bbdd91ce lane31 91c944d603539c62
+CPU verify: 10.536 ms per 32-lane warp, avg of 4000 (checksum 5abf8f93f330d440)

docs/design/class-v5-bench/20261007T083716Z/sibling-v5.txt

@@ -0,0 +1,8 @@
+igneum-pow bench: seed "igneum-genesis", day "2026-10-03", dataset 2^28 words (memory-hard)
+cache: fill 757.6 ms on one core (2^26 words, 256 MiB, 65536 chains of 64 ChaCha12 blocks), FNV-1a 64 48c4f5bf24166b2e
+state stream /srv/builds/_log/v5-class/node1-state.igsd1: chain block 159357 af89be5ddbadb6f6b4aee28ac8f249713be5d4c12621e3cea7f83ceada3c66b3, root 1c583d352bb9c75a06dadb8d82d42836ebe8afa82d0b413be87bf921741f1526, 93 records, 93 leaves
+program: class mx8+sh256x27+state, 128 loads/hash, 512 bytes/hash, widths (1,4,16 words) [16, 0, 0], 4096 items/warp, mixer x8 (72 mixers/item), cache 2^26 words, op mix load=16 add=8 shfl=8 xor=6 mad=5 mul=5 mulhi=5 sub=4 rotl=3 rotr=3 or=1; epoch built in 1062.6 ms
+shadow: 256 instructions x 27 passes per iteration, 55296 shadow instructions per hash, op mix add=47 rotl=30 xor=30 shfl=29 mad=27 mul=22 sub=21 rotr=20 mulhi=18 or=12
+warp base 0: single cold run 17.767 ms, 4096 items derived, lane0 18d47ad5e5f0db00 lane31 d1b36ed0ff68e575
+warp base 4096: single cold run 17.585 ms, 4095 items derived, lane0 a1f55f0b0cae0ed2 lane31 8c548231ef24eb73
+warp base 1000000: single cold run 14.724 ms, 4096 items derived, lane0 bd47078fc688b540 lane31 01f4d9a0bef81008

docs/design/class-v5-stored-state.md

@@ -14,7 +14,10 @@
 | 08:42 | Section 2a committed (3cbe8c0f) |
 | 08:56 | FIRST GREEN: igneum-pow class v5 (c6bbde78): blake2b, state leaves, the leaf XOR behind `Shape.state`, `V5_CLASS`, generator 5, the emitters' leaf buffer, the CLI's `--state`; 66 unit and 39 integration tests on igneum-build-1, the known-failed case (a stateless hasher) first |
 | 09:10 | The fork builds (class-v5-node): `ProgramClass::V5`, `program_class_v5_activation_daa` (never, in the digest once set), the byte-5 tally and the one-step rule, the executor's per-epoch state captures and `igneum_getPowStateLeaves`, the engine's `DayStateProvider` and `DayStateUnavailable`, the miner's RPC provider with `--exec-rpc` and `--freeze-state`, the template's v5 fields (37 to 40); kaspa-pow 16 tests green; the harness `infra/fast-time/class-v5-signal.mjs` and the 4090 bench committed (e528cb05) |
-| 09:1x | The remaining fork suites wait behind another lane's measurement hold on igneum-build-1 |
+| 09:30 | Fork suites on igneum-build-1: kaspa-pow with the feature 16 passed (the engine's known-failed case `class_v5_refuses_without_state_and_refreshes_the_leaves_per_epoch` among them), igneum-exec 28 passed (`a_tampered_stream_is_refused`, `epoch_state_captures_follow_the_cut_and_unwind`), kaspa-consensus `class_signal` 1 passed, kaspa-consensus-core 122 of 123: the one failure is `fast_time_60x_file_is_the_devnet_at_60x` demanding the `emission` field of the 0.3.18 fork's `OverrideParams` in the repo's 60x file, which master's file lacks as well (the economy lane's field, not this lane's; the v5 field is in the file and read) |
+| 09:33 | The pinned packs exported on the box from the devnet's IGSD1 stream (`proto-cuda/packs-ca3-v5/`, 4d10ac56); the packs test 20 passed: the v4 control pack agrees with the v5 epoch on 0 of 32 lanes, the hash kernel text equal, the build kernel and leaves the difference |
+| 09:37 | The verifier measured (section 7): v5 adds 0.28 ms cold and 0.20 ms on average over v4 on the loaded reference core; both under 10 ms |
+| 09:37 | The harness's known-failed case started on the box (`tools/class-v5/harness-remote.sh failed-case -- --signal 5,5,4 --expect flip`) |
 
 ## 1. The claim, in one paragraph
 
@@ -174,7 +177,7 @@ The same machinery as class v4 and the ladder (counter-asic-3-node.md section 6;
 | Row | Value | Where |
 |---|---|---|
 | The devnet's state (node 1's exec snapshot, tip 159,357) | 93 records, 8,619 bytes; stream built in 0.1 ms; rebuilt and root-checked in 0.1 ms | `igneum-day-stream`, 08:33 UK, core 40 at nice 19 |
-| Verifier per unit, class v4 against class v5 (cold warp, core 40, sibling loaded; the ladder's method) | owed (section 0 carries the time) | `tools/ladder/verify-bench-remote.sh` shape with `--program-class v5 --state <stream>` |
+| Verifier per unit, class v4 against class v5 (one 32-lane warp, core 40 at 3.72 GHz, nice 19, the ladder's method; the box at load 70 from other lanes' builds, so every figure is a loaded-box figure: the ladder's rung-0 row read 5.14 alone and 8.77 loaded at load 25) | v4: cold 8.11 ms alone, 8.74 ms with the sibling loaded; average of 50: 7.83 and 8.63. v5 over the devnet's 93 leaves: cold 8.71 alone, 9.03 loaded; average of 50: 8.54 and 8.83. So v5 adds 0.28 ms cold and 0.20 ms on average with the sibling loaded (+3.2 and +2.3 percent), inside the prototype's +0.11 to 0.21 ms row, and both classes stay under the 10 ms gate on the loaded core at load 70 | `tools/class-v5/verify-bench-remote.sh`, 09:37 UK, raw lines under `docs/design/class-v5-bench/20261007T083716Z/` |
 | 4090: hash rate and watts, v4 against v5; dataset build with the devnet's leaves | owed | the fleet pod, `proto-newpow/state-dataset` harness on the class v4 pack text with the real leaf array |
 
 The prototype's rows stand meanwhile (new-pow 5.2): 63.083 against 63.088 MH/s, 205 to 208 W, build 30.55 against 31.98 ms resident, verifier +0.11 to 0.21 ms per unit.

tools/class-v5/verify-bench-remote.sh

@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+# The class v5 verifier cost against class v4 on igneum-build-1 (docs/design/class-v5-stored-state.md section 7), the
+# latency ladder's method (tools/ladder/verify-bench-remote.sh): the cold verify of one 32-lane warp on the reference
+# core (core 40, nice 19), alone and with its SMT sibling (core 88) running the same bench, plus the average of 50
+# warps; class v4 at rung 0 against class v5 over the devnet's state stream (node 1's exec snapshot, 93 records).
+#
+#   tools/class-v5/verify-bench-remote.sh [--pow <path on the box>] [--state <IGSD1 path on the box>] [--out <dir on the Mac>]
+#
+# Runs ON the box under its measure hold (infra/build-server/remote-run.sh BR_MEASURE=1), as the ladder's does.
+set -euo pipefail
+HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+BS_TOOL=class-v5-verify-bench
+# shellcheck source=../../infra/build-server/lib.sh
+. "$HERE/../../infra/build-server/lib.sh"
+POW=""; STATE="/srv/builds/_log/v5-class/node1-state.igsd1"; OUT=""
+while [ $# -gt 0 ]; do
+  case "$1" in
+    --pow) POW="$2"; shift 2 ;; --state) STATE="$2"; shift 2 ;; --out) OUT="$2"; shift 2 ;;
+    *) echo "unknown argument $1" >&2; exit 2 ;;
+  esac
+done
+bs_host
+pushd "$HERE/../../igneum-pow" > /dev/null; bs_context; popd > /dev/null
+WT_ROOT="$BS_WT_ROOT"
+WT_NAME="$BS_WT"
+POW="${POW:-/srv/builds/$WT_NAME/igneum-pow/target/release/igneum-pow}"
+OUT="${OUT:-$WT_ROOT/docs/design/class-v5-bench}"
+STAMP=$(date -u +%Y%m%dT%H%M%SZ)
+REMOTE_OUT="/srv/builds/_log/v5-class/bench/$STAMP"
+mkdir -p "$OUT"
+read -r -d '' CMD <<EOF || true
+set -u; mkdir -p '$REMOTE_OUT'; cd '$REMOTE_OUT'
+echo "host \$(hostname) load \$(cut -d' ' -f1-3 /proc/loadavg) freq40 \$(cat /sys/devices/system/cpu/cpu40/cpufreq/scaling_cur_freq 2>/dev/null || echo ?) kHz siblings \$(cat /sys/devices/system/cpu/cpu40/topology/thread_siblings_list) pow \$(sha256sum '$POW' | cut -c1-16) state \$(sha256sum '$STATE' | cut -c1-16)" > meta.txt
+for cls in v4 v5; do
+  extra=""; [ "\$cls" = v5 ] && extra="--state '$STATE'"
+  eval nice -n 19 taskset -c 40 '$POW' bench --seed igneum-genesis --day 2026-10-03 --program-class \$cls \$extra --warps 50 > alone-\$cls.txt 2>&1
+  eval nice -n 19 taskset -c 88 '$POW' bench --seed igneum-genesis --day 2026-10-03 --program-class \$cls \$extra --warps 4000 > sibling-\$cls.txt 2>&1 &
+  sib=\$!
+  sleep 1.5
+  eval nice -n 19 taskset -c 40 '$POW' bench --seed igneum-genesis --day 2026-10-03 --program-class \$cls \$extra --warps 50 > loaded-\$cls.txt 2>&1
+  kill \$sib 2>/dev/null; wait \$sib 2>/dev/null
+  echo "class \$cls done \$(date -u +%H:%M:%SZ)"
+done
+echo "load after \$(cut -d' ' -f1-3 /proc/loadavg)" >> meta.txt
+EOF
+bs_log "measuring class v4 and v5 verifier on $BS_HOST under the measure hold (pow $POW, state $STATE); results to $OUT"
+BR_MEASURE=1 BR_KIND=measure BR_COMMAND="igneum-pow bench class v4 and v5 (core 40 alone, then with core 88 loaded)" BR_TARGET=x86_64-unknown-linux-gnu \
+  bs_remote_run "$(dirname "$POW")" "class v5 verifier bench" "$CMD"
+bs_rsync -a "$BS_HOST:$REMOTE_OUT/" "$OUT/$STAMP/"
+for cls in v4 v5; do
+  for kind in alone loaded; do
+    f="$OUT/$STAMP/$kind-$cls.txt"
+    cold=$(grep -o "single cold run[^0-9]*[0-9.]* ms" "$f" | head -1 || true)
+    avg=$(grep -i "CPU verify" "$f" | head -1 || true)
+    echo "$cls $kind: $cold | $avg"
+  done
+done
+cat "$OUT/$STAMP/meta.txt"