COMMITS
January 14, 2026
M
[CK][Examples] Fixing stride issues in ck examples 14/65/68/69 by workaround - Bypassing hostTensor validation
Michal Kulikowski committed
J
Update README.md files to match recent code changes
John Shumway committed
I
add aiter test_batch_prefill and simplify jenkins file a bit (#3570)
Illia Silin committed
E
Disable CK Tile Stream-K reduction tests (#3559)
Emily Martins committed
J
[CK_BUILDER] Update owners file for more reviews for CK Builder (#3572)
John Shumway committed
B
Disable ActiveWorkgroupsPerCU for different arch in wmma kernels (#3566)
Bartłomiej Kocot committed
B
Fix grouped conv bwd data wmma check (#3562)
Bartłomiej Kocot committed
K
[CK_Tile] Support for group size 128 for Preshuffle quant for 2d block scale gemm (#3462)
Khushbu Agarwal committed
V
Build CK on Windows (#3458)
Ville Pietilä committed
J
[CK] Refactor GPU verification kernel to gather error stats on GPU (#3551)
Johannes Graner committed
J
[CK Profiler] Initialize tensors on GPU in CK profiler (#3550)
Johannes Graner committed
L
[CK_TILE][FMHA] Enable gpt-oss sink (#3490)
Linjun-AMD committed
E
Add support for direct store in epilogue and padding support for wave transfer without transpose (#3465)
Enrico Degregori committed
T
[CK TILE ENGINE] CI fix for Basic Tile Engine (#3554)
Thrupti Raj Lakshmana Gowda committed
January 15, 2026
Y
Implement device_gemm_universal_preshuffle_instance for RDNA4 (#3429)
Yung-sheng Tu committed
E
Fix error when building with -DCMAKE_BUILD_TYPE=Debug (#3541)
Estevan Vedovelli committed
J
[FMHA] Enable page size 16 for batch prefill kernel (#3568)
Jeff Huang committed
J
[CK_BUILDER] Convert convolution traits to a struct with factory functions (#3547)
John Shumway committed
January 13, 2026
T
Shuffle fix for gfx950 (#3491)
Thomas Ning committed
V
[CK_BUILDER] Add bwd weight factories (#3509)
Ville Pietilä committed
P
fix incorrect List import in reduce_parameter.py (#3555)
Po Yen Chen committed
E
Implement grouped gemm tile loop for RDNA4 (#3304)
Erwin Terpstra committed
J
[CK Tile] Fix FMHA LSE calculation and potential division by zero (#3326)
Jeff Huang committed
J
[FMHA] Support page_size=1 (linear layout) in batch prefill pipeline (#3545)
Jeff Huang committed
Z
fix mxfp8-gemm example failure (#3531)
ZheWang committed
January 12, 2026
A
A
refactor: remove Default scheduler implementation as it not used anymore (#3542)
Aviral Goel committed
J
[CK profiler] Perform verification on GPU when using GPU reference (#3482)
Johannes Graner committed
K
adressed review comments from PR3459 (#3526)
kabrahamAMD committed
R
ck-builder: tensor input/output reflection (#3536)
Robin Voetter committed