Skip to content

Pull requests: tracel-ai/cubecl

New pull request New
23 Open 1,327 Closed
23 Open 1,327 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fix(topology): a cube's units are contiguous in ABSOLUTE_POS
#1595 opened Sep 2, 2026 by ThierryCantin-Demers Member • Draft
fix(throughput): key the cache on the device part, and version the probes
#1594 opened Sep 2, 2026 by ThierryCantin-Demers Member • Draft
8 tasks
fix(wgpu): close a device profile on its last compute pass
#1592 opened Sep 2, 2026 by ThierryCantin-Demers Member Loading…
8 tasks
1
fix(cuda): stop advertising MMA and tensor cores on dies that have none
#1591 opened Sep 2, 2026 by ThierryCantin-Demers Member Loading…
8 tasks
Refactor/runtime erasure
#1590 opened Sep 1, 2026 by nathanielsimard Member Loading…
fix(cpp): lower CUDA complex casts with cuComplex helpers
#1581 opened Aug 31, 2026 by shinaoka Contributor Loading…
8 tasks done
test(runtime): coordinate the profile tests with the refusal toggle
#1579 opened Aug 31, 2026 by ThierryCantin-Demers Member Loading…
perf(cpu): stop the pool starving the client, and run ahead unconditionally
#1566 opened Aug 27, 2026 by ThierryCantin-Demers Member • Draft
fix(spirv, wgsl): emit real fmas and fuse a product into the subtraction consuming it
#1562 opened Aug 27, 2026 by ThierryCantin-Demers Member Loading…
feat(wgpu/msl): register bf16 when the device probe-compiles bfloat
#1509 opened Aug 12, 2026 by scott-wyatt Loading…
1
feat(cuda): add resilient native and grouped GEMM dispatch
#1440 opened Jul 18, 2026 by ppodolsky Loading…
4 tasks done
1
refactor(benchmark): make Benchmark::sync fallible
#1414 opened Jul 9, 2026 by dcvz Contributor Loading…
3
fix(runtime): saturate FlushingPolicyState counters to prevent add-overflow panic (#1359)
#1413 opened Jul 8, 2026 by teddytennant Loading…
1
1
fix(cuda): reach pinned-DMA bandwidth for large host-to-device uploads
#1404 opened Jul 4, 2026 by matteospanio Loading…
1
fix(wgpu): clamp compute dispatch cube count to u16::MAX (NVIDIA Vulkan)
#1388 opened Jun 19, 2026 by thewtex Loading…
fix: Remove unused CUtensorMap import
#1377 opened Jun 8, 2026 by Lang-Qiu • Draft
1
Remove md5 deps
#1371 opened Jun 6, 2026 by xnorpx • Draft
4 tasks
Fuzz the #[cube] macro
#1364 opened May 30, 2026 by LucaCappelletti94 Contributor Loading…
Feat/distributed comm
#1353 opened May 26, 2026 by nathanielsimard Member • Draft
feat(runtime): pinned-host-buffer fast path for create_from_slice uploads
#1334 opened May 17, 2026 by lilith Contributor • Draft
3 of 5 tasks
2
fix(opt): visit input operand of AtomicOp::CompareAndSwap
#1331 opened May 14, 2026 by kimjune01 • Draft
1
1
fix(spirv): reject f64 transcendentals at compile time
#1328 opened May 10, 2026 by kimjune01 • Draft
1
1
feat: add exp2 and log2 operations
#1197 opened Feb 25, 2026 by jeff-hiner Loading…
9
ProTip! Adding no:label will show everything without a label.