Tile primitives for speedy kernels
Cuda · active 2025-04-01 → 2026-08-07 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 494 days from 2025-04-01 to 2026-08-07. Pushes: 632 total, peak 21 in a day. Pull requests: 34 total, peak 4 in a day. Issues: 29 total, peak 2 in a day. Comments: 45 total, peak 4 in a day. Stars: 507 total, peak 17 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| StuartSul | 386 | 371 | 5 | 5 |
| jordan-benjamin | 121 | 121 | 0 | 0 |
| dylanllim | 78 | 76 | 2 | 0 |
| benjaminfspector | 37 | 36 | 1 | 0 |
| simran-arora | 36 | 18 | 5 | 6 |
| druidowm | 7 | 7 | 0 | 0 |
| Edenzzzz | 7 | 0 | 0 | 6 |
| DanFu09 | 6 | 2 | 1 | 3 |
| XiaobingSuper | 5 | 0 | 1 | 3 |
| Kushagra481 | 4 | 0 | 1 | 1 |
| irasin | 3 | 0 | 0 | 2 |
| tomflinda | 3 | 0 | 1 | 2 |
| Lazarus-931 | 3 | 0 | 1 | 2 |
| technillogue | 2 | 0 | 2 | 0 |
| goldhuang | 2 | 0 | 0 | 0 |
| Maharshi-Pandya | 2 | 0 | 0 | 1 |
| MaoZiming | 2 | 0 | 0 | 1 |
| zhiweij1 | 2 | 0 | 2 | 0 |
| tugrul512bit | 2 | 0 | 0 | 0 |
| Xiang-cd | 2 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#195Lazarus-9312026-04-20 04:21m3 + benchmark harness
- Pull request#195Lazarus-9312026-04-20 04:18
- Pull request#193Emre-Dinc2026-04-09 17:39
- Issue comment#192Lazarus-9312026-04-08 01:56[Question] Mamba 3
- Issue#187jeromeku2026-03-26 03:07TK vs FA4 Benchmark
- Pull request#186nashbrown2026-03-25 03:37
- Pull request#185nashbrown2026-03-24 19:20
- Issue comment#178StuartSul2026-03-16 04:29[Bug] `namespace "std" has no member "copy_n"` in include/common/util.cuh when building kernels/attention/mha_h100
- Pull request#172enbaocao2026-03-16 00:33
- Issue#173JINO-ROHIT2026-03-04 04:04can i run these kernels for adalovelace arch?
- Issue#175tugrul512bit2026-02-25 18:35Do you support batched gemm operations in a custom kernel with custom branching?
- Issue#175tugrul512bit2026-02-18 13:55Do you support batched gemm operations in a custom kernel with custom branching?
- Issue comment#174faiziktk2026-02-12 10:34Windows with Visual Studio and Cuda tool kit
- Pull request#172enbaocao2026-02-07 01:16
- Issue#168StuartSul2026-02-04 02:50Will EP be supported in the future?
- Issue comment#166StuartSul2026-02-04 02:48Will parallel kittens support decoding in the future?
- Issue#169FireKingY2026-01-29 03:23[Request] microbenchmark code for ParallelKittens: TMA vs ld/st SM utilization analysis
- Issue comment#167MaoZiming2026-01-25 19:28H100: ThunderKittens/include/types/system/vmm.cuh:152 'invalid argument'
- Issue#168ziyuhuang1232026-01-25 07:20Will EP be supported in the future?
- Issue#167MaoZiming2026-01-24 23:09H100: ThunderKittens/include/types/system/vmm.cuh:152 'invalid argument'
- Issue comment#165Eigensystem2025-12-10 08:40Fix build failure on Hopper caused by missing fp4 definition
- Issue comment#159KevinZeng082025-12-03 06:56Is there a runnable TK-based Flash Attention on B200?
- Issue#162siddagra2025-11-26 08:42Simple Conv2D implementation?
- Issue comment#160Maharshi-Pandya2025-11-21 13:19Compile error on B200 for parallel kernels: too few arguments in function call
- Issue#160Maharshi-Pandya2025-11-21 13:19Compile error on B200 for parallel kernels: too few arguments in function call
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 507 stars here means stars gained during the window, not the repo's star count.