igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity
igneum/docs/analysis/class-v6/coexist-rows.md

13 KiB
Raw Blame History

Class v6 coexistence rows: the 5.5 GiB miner beside the shard prover on an RTX 3060 12 GB and an RTX 4060 8 GB

Measured 8 October 2026, 16:10 to 16:38 BST, on two Vast.ai instances rented for the rows and destroyed after them (RTX 3060: instance 54866832, Norway, driver 610.57.04, power limit 170 W, 0.276 h, USD 0.02; RTX 4060: instance 54866835, Mexico, driver 570.211.01, power limit 115 W, 0.545 h, USD 0.07). Record for the Igneum 2.0 pin D1 ("5.5 GiB coexistence rows") and the proving statement's memory condition. Every row below is the RESULT line as the pod wrote it to its run log (scratchpad v6coexist/fb-v6c-3060/run.log, fb-v6c-4060/run.log and run2.log; the 1 Hz samples beside them). Times in the logs are UTC.

What ran

  • Miner: the ds55 kit's own worker (igneum-ca3-ds55-kit-20261008.zip, sha 2d7f55e8..., worker sha d43be4625b78baf7) on packs/ds55, the 5.5 GiB dataset (1,476,395,008 words), --bench --batch-log2 24 --block-warps 1 --device 0; 100 batches alone, 400 batches as the beside run.
  • Prover: igneum-prove-host-0317 (sha 71bc2438856bb141) on the fixed shard fees-v1-shards2.json shard 0 (4,717,439 cycles, fixture sha 20a108f159c61ff9), --mode compressed --shard 0, SP1_GPU_ELEMENT_THRESHOLD=67108864 (2^26), against the served floor tarballs: sm_86 (igneum-floor-sm86.tgz, patched server sha 224200d317cea579) on the 3060, sm_89 (igneum-floor-sm89.tgz, patched server sha 882bd34f7f69dc97) on the 4060.
  • Sampler: nvidia-smi --query-gpu=memory.used,power.draw,utilization.gpu once a second per phase. Watts are the mean of power.draw over the busy samples from 8 s in; the 4060 host reports power.draw as N/A, so the 4060 has no watts.
  • Register windows: the class v5 kit's worker (packs-ca3-v5-20261008T085619Z.zip, worker sha d84b1b6cb09cebdc) on hl-reg64c (the full chain) and hl-reg64 (arithmetic only), 100 batches each.

RTX 3060 12 GB: the rows verbatim

RESULT start 2026-10-08T15:10:03Z card=NVIDIAGeForceRTX3060 total_mib=12288 driver=610.57.04 power_limit_w=170.00
RESULT bins worker=d43be4625b78baf7 host=71bc2438856bb141 server=224200d317cea579 fixture=20a108f159c61ff9
RESULT idle mem_mib=1 power_w=21.32
RESULT cmd miner_alone: /root/coex/kit/bin/linux/igneum-worker-cuda --bench --pack /root/coex/kit/packs/ds55 --batches 100 --batch-log2 24 --block-warps 1 --device 0
RESULT miner_alone rc=0 wall_s=67 peak_mib=6129 check=PASS fingerprint=23ced07a4d28b465 mhs=26.824 self-test PASS
RESULT cmd proof_alone: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=67108864 timeout 900 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode compressed --shard 0 --out /root/coex/proof-alone.json
RESULT proof_alone rc=0 wall_s=31 peak_mib=7525 prove_s=13.2 verified=1 err=""
RESULT proof_alone_line RESULT compressed shard 0: prove 13.2 s, proof 1272897 bytes, verify 0.110 s, VERIFIED; statement 0x6adcc7fab21512b58cfa3778756718fd37aea6cccb311e12d2ca33b5f4bf10c0 proof sha256 0x57e133da05ea5f6ecbcbd731b1b26c9bed2f10c5a1aa1bd04a5a5e3dab1e
RESULT cmd miner_beside_comp26: /root/coex/kit/bin/linux/igneum-worker-cuda --bench --pack /root/coex/kit/packs/ds55 --batches 400 --batch-log2 24 --block-warps 1 --device 0
RESULT cmd proof_beside_comp26: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=67108864 timeout 600 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode compressed --shard 0 --out /root/coex/proof-beside-comp26.json (started 1s+10s after the miner, miner resident mem_mib=6129)
RESULT proof_beside_comp26 mode=compressed thr=67108864 rc=1 wall_s=34 proof_window=1791472334..1791472368 miner_start=1791472323 prove_s= verified=0 miner_alive_after=yes err="thread 'tokio-rt-worker' (1790) panicked at slop/crates/alloc/src/raw_buffer.rs:271:9:"
RESULT miner_beside_comp26 rc=0 wall_s=256 peak_mib=11893 check=PASS fingerprint=23ced07a4d28b465 mhs=26.512 self-test PASS
RESULT beside_fallback compressed 2^26 beside the miner did not verify; core 2^25 beside the miner next
RESULT cmd proof_beside_core25: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=33554432 timeout 600 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode core --shard 0 --out /root/coex/proof-beside-core25.json (started 3s+10s after the miner, miner resident mem_mib=6129)
RESULT proof_beside_core25 mode=core thr=33554432 rc=1 wall_s=18 proof_window=1791472598..1791472616 miner_start=1791472585 prove_s= verified=0 miner_alive_after=yes err="Error: unknown mode core"
RESULT miner_beside_core25 rc=0 wall_s=254 peak_mib=11013 check=PASS fingerprint=23ced07a4d28b465 mhs=26.775 self-test PASS
RESULT beside_failed core 2^25 beside the miner did not verify either
RESULT idle_after mem_mib=1
RESULT window_hl-reg64c rc=0 wall_s=130 peak_mib=1521 regs=87 blocks_per_sm=16 check=PASS fingerprint=4e7cc25967eba280 mhs=13.467 self-test PASS
RESULT window_hl-reg64 rc=0 wall_s=128 peak_mib=1521 regs=104 blocks_per_sm=16 check=PASS fingerprint=70e786af1a457653 mhs=13.476 self-test PASS

Watts from the samples (busy mean of power.draw from 8 s in): miner alone 117.4 W; proof alone 122.2 W over the 9 busy seconds; miner with the failed compressed attempt beside it 118.6 W; hl-reg64c 120.8 W; hl-reg64 119.3 W. The card touched its 170 W limit on none of them.

RTX 4060 8 GB: the rows verbatim

The first pass (run.log) ran the proof rows against the stock SDK server, because the served sm_89 tarball carries the stock sp1-gpu-server in home/.sp1/bin (sha c2642ad1c42e85d8, the 24 GB gate at builder.rs:38) and the patched one only in bin/ (882bd34f7f69dc97); those two rows are kept here as the record of the fault. The patched server was copied over home/.sp1/bin at 16:13 BST and the proof rows re-ran (run2.log).

RESULT start 2026-10-08T15:10:44Z card=NVIDIAGeForceRTX4060 total_mib=8188 driver=570.211.01 power_limit_w=115.00
RESULT bins worker=d43be4625b78baf7 host=71bc2438856bb141 server=882bd34f7f69dc97 fixture=20a108f159c61ff9
RESULT idle mem_mib=2 power_w=[N/A]
RESULT cmd miner_alone: /root/coex/kit/bin/linux/igneum-worker-cuda --bench --pack /root/coex/kit/packs/ds55 --batches 100 --batch-log2 24 --block-warps 1 --device 0
RESULT miner_alone rc=0 wall_s=91 peak_mib=6116 check=PASS fingerprint=23ced07a4d28b465 mhs=18.837 self-test PASS
RESULT proof_alone rc=1 wall_s=1 peak_mib=2 prove_s= verified=0 err="thread 'tokio-rt-worker' (1470) panicked at sp1-gpu/crates/prover_components/src/builder.rs:38:9:"   (stock server: "Unsupported GPU memory: 12, must be at least 24GB")
RESULT proof_beside_comp26 mode=compressed thr=67108864 rc=1 wall_s=1 proof_window=1791472371..1791472372 miner_start=1791472359 prove_s= verified=0 miner_alive_after=yes err="thread 'tokio-rt-worker' (1811) panicked at sp1-gpu/crates/prover_components/src/builder.rs:38:9:"   (stock server, as above)
RESULT miner_beside_comp26 rc=0 wall_s=359 peak_mib=6128 check=PASS fingerprint=23ced07a4d28b465 mhs=18.837 self-test PASS
RESULT cmd proof_beside_core25: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=33554432 timeout 600 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode core --shard 0 --out /root/coex/proof-beside-core25.json (started 1s+10s after the miner, miner resident mem_mib=6116)
RESULT proof_beside_core25 mode=core thr=33554432 rc=1 wall_s=5 proof_window=1791472735..1791472740 miner_start=1791472724 prove_s= verified=0 miner_alive_after=yes err="thread 'tokio-rt-worker' (4670) panicked at slop/crates/tensor/src/inner.rs:51:51:"
RESULT miner_beside_core25 rc=0 wall_s=358 peak_mib=7459 check=PASS fingerprint=23ced07a4d28b465 mhs=18.834 self-test PASS
RESULT idle_after mem_mib=2
RESULT window_hl-reg64c rc=0 wall_s=179 peak_mib=1508 regs=87 blocks_per_sm=20 check=PASS fingerprint=4e7cc25967eba280 mhs=9.508 self-test PASS
RESULT window_hl-reg64 rc=0 wall_s=178 peak_mib=1508 regs=104 blocks_per_sm=16 check=PASS fingerprint=70e786af1a457653 mhs=9.574 self-test PASS

The rerun on the patched server (run2.log):

RESULT start 2026-10-08T15:31:00Z card=NVIDIAGeForceRTX4060 total_mib=8188 driver=570.211.01 power_limit_w=115.00
RESULT bins worker=d43be4625b78baf7 host=71bc2438856bb141 server=882bd34f7f69dc97 fixture=20a108f159c61ff9
RESULT cmd proof_alone: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=67108864 timeout 900 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode compressed --shard 0 --out /root/coex/proof-alone-rerun.json
RESULT proof_alone_rerun rc=0 wall_s=15 peak_mib=7532 prove_s=8.2 verified=1 err=""
RESULT proof_alone_rerun_line RESULT compressed shard 0: prove 8.2 s, proof 1272897 bytes, verify 0.036 s, VERIFIED; statement 0x6adcc7fab21512b58cfa3778756718fd37aea6cccb311e12d2ca33b5f4bf10c0 proof sha256 0x7b887af9426039e61345f11d2c9faf10d4b4d0d16e4850d10ee08c5fa7002
RESULT cmd proof_beside_comp26: env HOME=/opt/igneum-floor/home SP1_PROVER=cuda RUST_LOG=off SP1_GPU_FLOOR_LOG=1 SP1_GPU_ELEMENT_THRESHOLD=67108864 timeout 600 /opt/igneum-floor/bin-0317/igneum-prove-host /root/coex/fees-v1-shards2.json --mode compressed --shard 0 --out /root/coex/proof-beside-comp26-rerun.json (started 3s+10s after the miner, miner resident mem_mib=6116)
RESULT proof_beside_comp26 mode=compressed thr=67108864 rc=1 wall_s=5 proof_window=1791473509..1791473514 miner_start=1791473495 prove_s= verified=0 miner_alive_after=yes err="thread 'tokio-rt-worker' (10627) panicked at slop/crates/tensor/src/inner.rs:51:51:"
RESULT miner_beside_comp26 rc=0 wall_s=359 peak_mib=7811 check=PASS fingerprint=23ced07a4d28b465 mhs=18.833 self-test PASS
RESULT beside_fallback compressed 2^26 beside the miner did not verify (the core 2^25 beside row is in run.log)
RESULT idle_after mem_mib=2

The table

Card Miner alone (ds55) Proof alone (compressed 2^26) Together Register windows (v5 kit worker)
RTX 3060 12 GB 26.82 MH/s at 117.4 W, 6,129 MiB resident, fingerprint 23ced07a4d28b465, PASS 13.2 s, VERIFIED, peak 7,525 MiB, 122 W 6,129 + 7,525 = 13,654 MiB against 12,288: TIME-SHARING NEEDED. Live attempt: the card filled to 11,893 MiB and the prover died in a device allocation (raw_buffer.rs:271) after 34 s; the miner held 26.51 MH/s through it (1.2 percent under alone). Proof with the miner paused = the proof-alone row hl-reg64c 13.47 MH/s, 87 regs, 16 of 24 blocks per SM, fingerprint 4e7cc25967eba280 MATCH, 120.8 W; hl-reg64 13.48 MH/s, 104 regs, 16 of 24, 70e786af1a457653 MATCH, 119.3 W
RTX 4060 8 GB 18.84 MH/s, 6,116 MiB resident, fingerprint 23ced07a4d28b465, PASS (no watts: host sensor N/A) 8.2 s, VERIFIED, peak 7,532 MiB 6,116 + 7,532 = 13,648 MiB against 8,188: TIME-SHARING NEEDED. Live attempt: the server died in a tensor allocation (inner.rs:51) at 7,811 MiB after 5 s; the miner held 18.83 MH/s hl-reg64c 9.51 MH/s, 87 regs, 20 of 24 blocks per SM, fingerprint MATCH; hl-reg64 9.57 MH/s, 104 regs, 16 of 24, MATCH

Not measured and why: a core-only proof beside the miner (igneum-prove-host-0317 has no --mode core: "Error: unknown mode core"; its modes are native, execute, shard, compressed, block, all); watts on the 4060 (the host's power sensor reads N/A).

Reconciliation with the served row

The served sentence (litepaper, "Proving", from docs/analysis/prover-tiers-real-cards.md, 6 October 2026) says an RTX 3060 (12 GB) "mines at 23.78 MH/s and proves the v1 shard beside its miner at an 8.9 GB peak in 37.5 s". Today's row on the same card tier reads a 7.5 GiB compressed peak that kills the prover beside a 6.1 GiB miner. The two are not in conflict; they measured different things. The 6 October row is the matrix's miner-comp-26-v1 point (tools/fleet/box-matrix.sh section 7): the patched server at threshold 2^26 in compressed mode, driven by the matrix host (the segment host build at /opt/igneum-segal/.../igneum-prove-host, which also carries --mode core; the matrix's core rows come from it), beside igneum-miner mine on the 1 GiB class v3 pack, whose resident set was 1.4 GB: 1.4 + 7.5 = 8.9 GB, inside 12 GB, proof in 37.5 s with the miner running. Today's row is igneum-prove-host-0317 in compressed mode at the same threshold 2^26 (the same 7.5 GiB own footprint, 7,525 to 7,532 MiB peak alone), beside the ds55 miner on the 5.5 GiB dataset of the genesis floor, whose resident set is 6.1 GB: 6.1 + 7.5 = 13.6 GB, outside 12 GB and 8 GB alike, so the prover's allocation fails while the miner keeps mining. What changed between the rows is the miner's dataset (1 GiB then, 5.5 GiB now), not the prover. The rule that follows: at the 5.5 GiB floor a compressed shard proof beside a running miner needs a 16 GB card (13.6 GB together; the 16 GB tier's own peak is 18 GB on the stock sizes and 7.8 GB patched, so 16 GB holds both with about 2 GB spare); 8 GB and 12 GB cards time-share, proving with the miner paused (13.2 s on the 3060, 8.2 s on the 4060) and mining otherwise. The 6 October beside rows stand only for the 1 GiB dataset; the 6 October core-only beside rows (5.6 GB own on the 3060, 27.2 s) stand only for core mode on the segment host, which the 0317 host does not expose, and are not a coexistence claim at the floor. The served sentence is read as a 1 GiB-dataset row until the site lane rewrites it against this record.