Skip to content

Pull requests: ROCm/composable_kernel

New pull request New
51 Open 1,555 Closed
51 Open 1,555 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
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix pk_int4 cast and add pk_int4 dtype in ck tile
#1854 opened Jan 31, 2025 by bartekxk Loading…
7 tasks done
@bartekxk
CK Tile GEMM Compute V2 (2 LDS Ping Pong mechanism)
#1853 opened Jan 31, 2025 by ThomasNing Loading…
5 tasks
23
Add pre_softmax fnctor
#1852 opened Jan 31, 2025 by amd-hhashemi Loading…
7 tasks
1
add new pk_i4 cvt for fp16 and bf16
#1847 opened Jan 30, 2025 by zjing14 • Draft
7 tasks
Add flex attention example
#1846 opened Jan 29, 2025 by tenpercent • Draft
7 tasks
1
Support for dtypes (fp8, bf8, bf16 and fp16) for the ck_tile/03_gemm example.
#1845 opened Jan 28, 2025 by kylasa Loading…
7 tasks
13
[CK_TILE] moe sorting ex kernel to support expert > 128
#1840 opened Jan 26, 2025 by carlushuang • Draft
3 of 5 tasks
Added Int4 mixed batch gemm support
#1839 opened Jan 26, 2025 by mtgu0705 Loading…
7 tasks
6
Add make_kernel_pt for specific architecture compilation guards
#1824 opened Jan 17, 2025 by alugorey Loading…
7 tasks
4
Prec param new
#1813 opened Jan 14, 2025 by aledudek Loading…
6 tasks
6
[CK_TILE] Use the GEMM example prec argument
#1811 opened Jan 12, 2025 by aledudek • Draft
1 of 6 tasks
1
CK Tile Gemm API and heuristics changes
#1809 opened Jan 10, 2025 by jakpiase Loading…
6 tasks done
32
Update LICENSE to 2025 (#1797) ci:docs-only Skip most non-doc CI for this PR documentation Improvements or additions to documentation
#1801 opened Jan 7, 2025 by spolifroni-amd Loading…
6 tasks
[Draft] | GPUAI-3720 - Integrate Universal GEMM into Grouped GEMM - Pt 1
#1800 opened Jan 7, 2025 by rtmadduri • Draft
6 tasks
35
Cross GPU Reduce Operator Initial Development
#1795 opened Jan 6, 2025 by ThomasNing Loading…
4 of 6 tasks
1
[CK_TILE] Support moe with up gemm
#1793 opened Jan 4, 2025 by huaiguxu • Draft
5 of 6 tasks
4
[CK_TILE] Add GetName for GEMM kernels
#1791 opened Jan 3, 2025 by aledudek Loading…
6 tasks done
25
[CK_TILE] Sync fmha fwd splitkv minor optimizations
#1785 opened Jan 1, 2025 by poyenc Loading…
2 tasks done
@poyenc
device_prop.hpp: move static map to helper function and initialize there
#1763 opened Dec 18, 2024 by coconutruben Loading…
3 of 6 tasks
3
[Ck tile] Use raw store to improve layernorm performance
#1752 opened Dec 16, 2024 by rocking5566 Loading…
1
disable atomicAdd for C Output Vector Length = 1 with 16bit data type
#1737 opened Dec 10, 2024 by zjing14 Loading…
4
Add Grouped Convolution and GEMM documentation
#1719 opened Dec 5, 2024 by bartekxk Loading…
@bartekxk
11
[CK TILE] Support pure quantization
#1703 opened Nov 29, 2024 by rocking5566 Loading…
[BUGFix] Reorder GetSmemSize Related API
#1683 opened Nov 21, 2024 by LeiWang1999 Loading…
@LeiWang1999
1
Apply universal gemm to bwd_weight_cshuffle operator
#1658 opened Nov 12, 2024 by mozga-amd Loading…
23
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.