grenade

grenade pushed to main at blackbeard/miner

2026-09-13 22:24:09 +00:00

8b01387e97 Merge pull request 'release: tag-triggered binaries and a README that states the measured rates' (#28) from release/binaries-and-readme into main

7831323362 release: tag-triggered binaries, README section with measured rates

c1cac91419 engine-cuda: LAIR_MINBLOCKS launch-bounds knob, with the occupancy results

Compare 3 commits »

grenade merged pull request blackbeard/miner#28

2026-09-13 22:24:08 +00:00

release: tag-triggered binaries and a README that states the measured rates

grenade created pull request blackbeard/miner#28

2026-09-13 22:14:06 +00:00

release: tag-triggered binaries and a README that states the measured rates

grenade created branch release/binaries-and-readme in blackbeard/miner

2026-09-13 22:13:54 +00:00

grenade pushed to release/binaries-and-readme at blackbeard/miner

2026-09-13 22:13:54 +00:00

7831323362 release: tag-triggered binaries, README section with measured rates

grenade commented on issue blackbeard/miner#27

2026-09-13 22:03:50 +00:00

quanpool-miner 6.2.4 measured on beast: 1234 MH/s per 5090, what it does differently, what was tried

Upstream's own engine measured for comparison. origin/main 0ba8ddf (v4.2.0, "CUDA: software-pipelined internal rounds, free-addend products, no early exit"), built on beast, `quantus-miner…

grenade commented on issue blackbeard/miner#27

2026-09-13 21:49:20 +00:00

quanpool-miner 6.2.4 measured on beast: 1234 MH/s per 5090, what it does differently, what was tried

The LAIR_MINBLOCKS launch-bounds knob used for the register experiments is on branch perf/minblocks-knob (c1cac91), pushed but not merged: it changes no default, and merging would only trigger…

grenade pushed to perf/minblocks-knob at blackbeard/miner

2026-09-13 21:49:17 +00:00

c1cac91419 engine-cuda: LAIR_MINBLOCKS launch-bounds knob, with the occupancy results

grenade created branch perf/minblocks-knob in blackbeard/miner

2026-09-13 21:49:16 +00:00

grenade opened issue blackbeard/miner#27

2026-09-13 21:48:55 +00:00

quanpool-miner 6.2.4 measured on beast: 1234 MH/s per 5090, what it does differently, what was tried

grenade pushed to main at blackbeard/miner

2026-09-13 21:06:35 +00:00

17caeabc10 Merge pull request 'perf: whole-grid batches, early reject on element 0, precomputed nonce direction (+2.9% on beast)' (#26) from perf/scheduling into main

2e6fce162e engine-cuda: batches of whole grids, early reject on element 0, nonce direction

Compare 2 commits »

grenade deleted branch perf/scheduling from blackbeard/miner

2026-09-13 21:06:35 +00:00

grenade merged pull request blackbeard/miner#26

2026-09-13 21:06:33 +00:00

perf: whole-grid batches, early reject on element 0, precomputed nonce direction (+2.9% on beast)

grenade created pull request blackbeard/miner#26

2026-09-13 20:58:42 +00:00

perf: whole-grid batches, early reject on element 0, precomputed nonce direction (+2.9% on beast)

grenade created branch perf/scheduling in blackbeard/miner

2026-09-13 20:58:19 +00:00

grenade pushed to perf/scheduling at blackbeard/miner

2026-09-13 20:58:19 +00:00

2e6fce162e engine-cuda: batches of whole grids, early reject on element 0, nonce direction

grenade commented on issue blackbeard/miner#25

2026-09-13 20:19:26 +00:00

ptxas 13.0 -O1+ miscompiles unrolled inline-PTX carry chains in a probe kernel; deployed kernel unaffected by parity, root cause open

Probe source for the record (also beast:/tmp/kregs/tcmul.cu). Build `nvcc -O3 -arch=sm_120 [-DPROD_TMP

grenade commented on issue blackbeard/miner#25

2026-09-13 20:19:11 +00:00

ptxas 13.0 -O1+ miscompiles unrolled inline-PTX carry chains in a probe kernel; deployed kernel unaffected by parity, root cause open

Post-deploy guard run on beast with the #23 kernel (bench-harness, miner paused): parity OK, 2000/2000 jobs found solutions, all verified against CPU. Each job is a random 32-byte header and…

grenade opened issue blackbeard/miner#25

2026-09-13 20:18:38 +00:00

ptxas 13.0 -O1+ miscompiles unrolled inline-PTX carry chains in a probe kernel; deployed kernel unaffected by parity, root cause open

grenade closed issue blackbeard/miner#24

2026-09-13 20:18:21 +00:00

research: Goldilocks multiplies on the INT8 tensor cores