2026-06-20 - 2026-09-20
Overview
7 Releases published by 1 user
Published
v4.0.2-lair.2
quantus-miner v4.0.2-lair.2
Published
v4.0.2-lair.1
quantus-miner v4.0.2-lair.1
Published
v4.0.2
Published
v4.0.1
Published
v4.0.0
Published
v3.3.1
Published
v3.3.0
20 Pull requests merged by 1 user
Merged
#32 deploy: require nvcc at build time
Merged
#31 engine-cuda: loop kernel diagonal from launch parameters (+1.3% loop kernel), multiply-add screening variants
Merged
#30 readme: 4090 at 727 MH/s on the loop kernel, how to select it
Merged
#29 engine-cuda: loop-structured mining_loop kernel, opt-in via MINER_CUDA_KERNEL=loop
Merged
#28 release: tag-triggered binaries and a README that states the measured rates
Merged
#26 perf: whole-grid batches, early reject on element 0, precomputed nonce direction (+2.9% on beast)
Merged
#23 perf: fused-PTX field arithmetic, +54..66% on every card
Merged
#22 perf: parameter-bank uniforms and per-arch carry path (+20.5% sm_120)
Merged
#21 deploy: miner credentials from the mainnet chain
Merged
#20 deploy: beast joins the miner fleet for mainnet
Merged
#19 deploy: chain id as a dispatch input for the miner credential path
Merged
#18 bench: record the memory clock alongside the SM clock (schema 2)
Merged
#17 engine-cuda: deferred carries, carry-flag arithmetic, one nonce per thread
Merged
#16 engine-cuda: native CUDA mining engine behind MinerEngine
Merged
#15 metrics: stale work is the batch in flight at cancellation, not the whole search
Merged
#14 metrics: attribute hashrate to build, device, kernel and job outcomes
Merged
#13 deploy: wait for the miner to connect before judging its metrics
Merged
#12 deploy: build on push to main and land the binary on the mining hosts
Merged
#11 bench: quantus-bench harness and a workflow that measures on a mining host
Merged
#10 ci: gitea workflow, drop inherited github workflows, embed commit in --version
10 Issues closed from 1 user
Closed
#27 quanpool-miner 6.2.4 measured on beast: 1234 MH/s per 5090, what it does differently, what was tried
Closed
#5 engine-gpu: double-buffered submission so the GPU never drains between batches
Closed
#7 NVIDIA WGSL kernel: port the Apple arithmetic wins without the code-size changes
Closed
#8 CI: replace inherited workflows with a Gitea fat-binary build and deploy-on-main
Closed
#24 research: Goldilocks multiplies on the INT8 tensor cores
Closed
#9 Continuous performance telemetry: attribute production hashrate to builds
Closed
#6 Batch size: raise the default and derive it from measured hashrate and a stale-work budget
Closed
#4 Allow more than one worker thread per GPU device
Closed
#3 engine-cuda: native CUDA kernel behind MinerEngine
Closed
#2 Benchmark and parity harness: the gate every perf change and every origin merge passes
12 Issues created by 1 user
Opened
#1 Architecture: loose coupling to origin for a performance-tuned miner
Opened
#2 Benchmark and parity harness: the gate every perf change and every origin merge passes
Opened
#3 engine-cuda: native CUDA kernel behind MinerEngine
Opened
#4 Allow more than one worker thread per GPU device
Opened
#5 engine-gpu: double-buffered submission so the GPU never drains between batches
Opened
#6 Batch size: raise the default and derive it from measured hashrate and a stale-work budget
Opened
#7 NVIDIA WGSL kernel: port the Apple arithmetic wins without the code-size changes
Opened
#8 CI: replace inherited workflows with a Gitea fat-binary build and deploy-on-main
Opened
#9 Continuous performance telemetry: attribute production hashrate to builds
Opened
#24 research: Goldilocks multiplies on the INT8 tensor cores
Opened
#25 ptxas 13.0 -O1+ miscompiles unrolled inline-PTX carry chains in a probe kernel; deployed kernel unaffected by parity, root cause open
Opened
#27 quanpool-miner 6.2.4 measured on beast: 1234 MH/s per 5090, what it does differently, what was tried