Skip to content

Pull requests: ROCm/composable_kernel

New pull request New
87 Open 2,030 Closed
87 Open 2,030 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

[CK_TILE] Fix unmatched K size of WarpGemmMfmaBf16Bf16F32M16N16K32TransposedCDistribution on gfx950
#2393 opened Jun 24, 2025 by DDEle • Draft
7 tasks done
Enabling diff datatypes for tile_engine and build with more granularity
#2392 opened Jun 24, 2025 by amd-khushbu Loading…
1 of 7 tasks
Huaiguxu/moe fp8 pertoken scale fix
#2391 opened Jun 24, 2025 by huaiguxu Loading…
7 tasks
[CK_TILE] Eliminate fmha batch_prefill warning caused by failed to meet occupancy requirement
#2389 opened Jun 24, 2025 by poyenc Loading…
4 of 7 tasks
@poyenc
Grouped conv bwd wei optimize instances 16x16 on hold
#2387 opened Jun 21, 2025 by bartekxk Loading…
7 tasks done
@bartekxk
add a mx_fp8 client example
#2380 opened Jun 20, 2025 by JiaLuo-CAN Loading…
1 of 7 tasks
3
[CK][CONV] Support NCHW in class DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
#2375 opened Jun 20, 2025 by linqun Loading…
7 tasks
12
A workaround for the case that get_slice_tile() doesn't work. bug Something isn't working help wanted Extra attention is needed question Further information is requested
#2371 opened Jun 19, 2025 by ruanjm • Draft
1
[CK TILE] Fix FA build filter
#2369 opened Jun 19, 2025 by aska-0096 Loading…
5 of 7 tasks
2
ck_tile kernel for gemm with groupwise quantized A or B tensor.
#2362 opened Jun 18, 2025 by vj-krish Loading…
6 tasks
1
[CK_TILE] Blockwise GEMM Pipeline V5
#2360 opened Jun 17, 2025 by aledudek Loading…
7 tasks
7
[CK_TILE] Grouped Convolution Backward Weight Kernel
#2357 opened Jun 16, 2025 by jakpiase Loading…
6 tasks done
8
WMMA GEMM b_scale
#2350 opened Jun 16, 2025 by EnricoDeg Loading…
6 of 7 tasks
1
Fix recent changes of universal_gemm in tile_engine
#2344 opened Jun 13, 2025 by amd-khushbu Loading…
1 of 7 tasks
1
added cktile doc; edited doxygen to pull in as much as possible ci:docs-only Skip most non-doc CI for this PR documentation Improvements or additions to documentation
#2343 opened Jun 13, 2025 by spolifroni-amd • Draft
7 tasks
Report HIP occupancy-driven grid sizes in Stream-K CkProfiler output
#2340 opened Jun 13, 2025 by ozturkosu Loading…
2 tasks done
@ozturkosu
2
support y-direction step length greater than 1 for SimplifiedGenericAttentionMask
#2338 opened Jun 13, 2025 by shay-li77 Loading…
7 tasks done
updated mxfp4 moe gemm2 config
#2330 opened Jun 12, 2025 by mtgu0705 Loading…
7 tasks
Enable FP4 native conversion and tests
#2329 opened Jun 11, 2025 by geyyer • Draft
3 of 7 tasks
Fix amd_ck_fp8.hpp macro definitions
#2325 opened Jun 10, 2025 by xli Loading…
1 of 7 tasks
3
Add function to print slices of tensors to help debug documentation Improvements or additions to documentation
#2323 opened Jun 10, 2025 by AviralGoelAMD Loading…
2 of 7 tasks
[EARLY draft] Unify codegen
#2313 opened Jun 9, 2025 by tenpercent • Draft
7 tasks
[CK_TILE] Enable persistent kernel and tail handler in tile_engine
#2300 opened Jun 6, 2025 by samremes Loading…
7 tasks
2
2217 add io module for profiler enhancement New feature or request
#2287 opened Jun 4, 2025 by smedegaard Loading…
1
@smedegaard
[CK_TILE] Add fMHA forward batch_decode kernel/pipeline
#2286 opened Jun 4, 2025 by poyenc • Draft
7 tasks
@poyenc
ProTip! no:milestone will show everything without a milestone.