Bench table: the class v4 watts rerun, first six rows (RTX 5070 176.1 W, 4070 Ti Super 137.8, 4070 Ti 155.3, 4070 Super 108.4 under a 110 W cap, 4060 Ti 16 GB 102.3, 4080 185.6; each with the class v3 watts paired on the same host as the shadow premium, +9 to +56 percent for no rate; the 5070 and 4070 Ti Super rates as ranges by board), MH per watt ranked again on those rows
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs
parent
5d1044ca25
commit
ba08d6b8a3
2 changed files with 34 additions and 38 deletions
|
|
@ -854,13 +854,13 @@
|
|||
"group": "buy",
|
||||
"card": "NVIDIA RTX 4070 Ti Super (16 GB)",
|
||||
"mh_s": 41.24,
|
||||
"watts": 142.6,
|
||||
"mh_per_w": 0.289,
|
||||
"watts": 137.8,
|
||||
"mh_per_w": 0.268,
|
||||
"driver_os": "NVIDIA driver 615.71.09, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 2 (a rented card, result cb2-rtx-4070-ti-super-7325-result.json; 09:26Z)",
|
||||
"note": "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. 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",
|
||||
"watts_class": "v3",
|
||||
"watts_display": "142.6 (class v3)"
|
||||
"source": "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)",
|
||||
"note": "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",
|
||||
"watts_display": "137.8 (class v4)",
|
||||
"mh_s_display": "37 to 41 by board (36.93 on this host)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
@ -902,13 +902,12 @@
|
|||
"group": "buy",
|
||||
"card": "NVIDIA RTX 4080 (16 GB)",
|
||||
"mh_s": 40.71,
|
||||
"watts": 128.6,
|
||||
"mh_per_w": 0.317,
|
||||
"watts": 185.6,
|
||||
"mh_per_w": 0.219,
|
||||
"driver_os": "NVIDIA driver 595.84, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 4 (a rented card, result cb2-rtx-4080-16-gb-7347-result.json; 09:28Z)",
|
||||
"note": "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. 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",
|
||||
"watts_class": "v3",
|
||||
"watts_display": "128.6 (class v3)"
|
||||
"source": "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)",
|
||||
"note": "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",
|
||||
"watts_display": "185.6 (class v4)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
@ -968,19 +967,18 @@
|
|||
"group": "buy",
|
||||
"card": "NVIDIA RTX 4060 Ti (16 GB)",
|
||||
"mh_s": 20.07,
|
||||
"watts": 79.2,
|
||||
"mh_per_w": 0.253,
|
||||
"watts": 102.3,
|
||||
"mh_per_w": 0.196,
|
||||
"driver_os": "NVIDIA driver 595.71.05, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 7 (a rented card, result cb2-rtx-4060-ti-16-gb-7364-result.json; 09:33Z)",
|
||||
"note": "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. 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",
|
||||
"source": "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)",
|
||||
"note": "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)",
|
||||
"hive": {
|
||||
"core_mhz": null,
|
||||
"mem_mhz": null,
|
||||
"pl_w": null,
|
||||
"label": "stock (rented card, no clock control on the host; no measured tune point)"
|
||||
},
|
||||
"watts_class": "v3",
|
||||
"watts_display": "79.2 (class v3)"
|
||||
"watts_display": "102.3 (class v4)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
@ -992,19 +990,19 @@
|
|||
"group": "buy",
|
||||
"card": "NVIDIA RTX 5070 (12 GB)",
|
||||
"mh_s": 51.81,
|
||||
"watts": 110.8,
|
||||
"mh_per_w": 0.468,
|
||||
"watts": 176.1,
|
||||
"mh_per_w": 0.334,
|
||||
"driver_os": "NVIDIA driver 580.173.02, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 8 (a rented card, result cb2-rtx-5070-12-gb-8750-result.json; 09:35Z)",
|
||||
"note": "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. 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",
|
||||
"source": "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)",
|
||||
"note": "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",
|
||||
"hive": {
|
||||
"core_mhz": null,
|
||||
"mem_mhz": null,
|
||||
"pl_w": null,
|
||||
"label": "stock (rented card, no clock control on the host; no measured tune point)"
|
||||
},
|
||||
"watts_class": "v3",
|
||||
"watts_display": "110.8 (class v3)"
|
||||
"watts_display": "176.1 (class v4)",
|
||||
"mh_s_display": "52 to 59 by board (58.89 on this host)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
@ -1016,19 +1014,18 @@
|
|||
"group": "buy",
|
||||
"card": "NVIDIA RTX 4070 Super (12 GB)",
|
||||
"mh_s": 31.24,
|
||||
"watts": 100.2,
|
||||
"mh_per_w": 0.312,
|
||||
"watts": 108.4,
|
||||
"mh_per_w": 0.288,
|
||||
"driver_os": "NVIDIA driver 580.178.04, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 9 (a rented card, result cb2-rtx-4070-super-12--8582-result.json; 09:36Z)",
|
||||
"note": "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. 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",
|
||||
"source": "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)",
|
||||
"note": "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",
|
||||
"hive": {
|
||||
"core_mhz": null,
|
||||
"mem_mhz": null,
|
||||
"pl_w": null,
|
||||
"label": "stock (rented card, no clock control on the host; no measured tune point)"
|
||||
},
|
||||
"watts_class": "v3",
|
||||
"watts_display": "100.2 (class v3)"
|
||||
"watts_display": "108.4 (class v4)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
@ -1279,19 +1276,18 @@
|
|||
"card": "NVIDIA RTX 4070 Ti (12 GB)",
|
||||
"group": "buy",
|
||||
"mh_s": 31.24,
|
||||
"watts": 95.4,
|
||||
"mh_per_w": 0.327,
|
||||
"watts": 155.3,
|
||||
"mh_per_w": 0.201,
|
||||
"driver_os": "NVIDIA driver 580.126.09, Ubuntu 24.04 (CUDA 12.8.1 image)",
|
||||
"source": "model sweep 2026-10-08, row 20 (a rented card, result cb2-rtx-4070-ti-12-gb-0695-result.json; 09:57Z)",
|
||||
"note": "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. 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",
|
||||
"source": "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)",
|
||||
"note": "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",
|
||||
"hive": {
|
||||
"core_mhz": null,
|
||||
"mem_mhz": null,
|
||||
"pl_w": null,
|
||||
"label": "stock (rented card, no clock control on the host; no measured tune point)"
|
||||
},
|
||||
"watts_class": "v3",
|
||||
"watts_display": "95.4 (class v3)"
|
||||
"watts_display": "155.3 (class v4)"
|
||||
},
|
||||
{
|
||||
"generator": "v2",
|
||||
|
|
|
|||
File diff suppressed because one or more lines are too long
Loading…
Reference in a new issue