-
Notifications
You must be signed in to change notification settings - Fork 96
Insights: ROCm/hipBLASLt
Overview
Could not load contribution data
Please try again later
16 Pull requests merged by 11 people
-
hipblaslt-bench: throw error if c_type is not equal to d_type
#1570 merged
Jan 20, 2025 -
[TensileLite] Support arbitrary M & K for swizzle-A kernels
#1558 merged
Jan 20, 2025 -
Optimize preloop by v_lshl_add
#1564 merged
Jan 20, 2025 -
Factor out argument parsing in TensileCreateLibrary
#1514 merged
Jan 17, 2025 -
code-gen: Allowed WaveGroups be distributed along n-dim for DTVA/SwizzledA
#1493 merged
Jan 17, 2025 -
Update BBS NN/NT/TN equality tuning for gfx942_80cu
#1557 merged
Jan 17, 2025 -
Added equality tuning for F8HS TN for gfx942
#1554 merged
Jan 16, 2025 -
Set default cmake code object version
#1553 merged
Jan 16, 2025 -
gfx942 BBS F8B8BS F8BS equality tuning
#1551 merged
Jan 16, 2025 -
fix typo in install.sh
#1560 merged
Jan 16, 2025 -
Install msgpack dependency for CentOS8
#1559 merged
Jan 16, 2025 -
Add emulation smoke/regression/extended tests.
#1556 merged
Jan 16, 2025 -
Check destination folder with yaml attribute while merging
#1555 merged
Jan 16, 2025 -
Use B64 instead of B32
#1548 merged
Jan 16, 2025 -
Update BBS NN/NT/TN Equality yamls.
#1549 merged
Jan 15, 2025 -
hipblaslt-bench: only print device caps of target device
#1539 merged
Jan 15, 2025
11 Pull requests opened by 10 people
-
Modify trig initialization on device to remove dependency on lda.
#1543 opened
Jan 14, 2025 -
Add tensilelite client performance args to hipblaslt-bench
#1544 opened
Jan 14, 2025 -
Remove global working path
#1546 opened
Jan 14, 2025 -
Fix device initialization 2^32 element limitation
#1552 opened
Jan 15, 2025 -
feature: DTVB with Swizzling (tensorB)
#1562 opened
Jan 16, 2025 -
Deprecate AMDGPU_TARGETS variable
#1563 opened
Jan 16, 2025 -
[Don't Merge] BF16 Swizzle Experiment
#1566 opened
Jan 17, 2025 -
[OPT] Tail Loop Optimization
#1567 opened
Jan 17, 2025 -
Stream-k libs for CPX mode
#1568 opened
Jan 17, 2025 -
Remove redundant code for gwvw > 1 route
#1573 opened
Jan 21, 2025 -
Support other types for Swizzling
#1574 opened
Jan 21, 2025
1 Issue closed by 1 person
-
[Issue]: Tensile should not be vendored
#1396 closed
Jan 14, 2025
4 Issues opened by 4 people
-
[Issue]: Inconsistent call counts between hipblaslt log and RPD tracing
#1572 opened
Jan 20, 2025 -
[Issue]: Build failure in TensileCreateExtOpLibraries
#1571 opened
Jan 20, 2025 -
[Issue]: fail to build hipBlasLt client
#1561 opened
Jan 16, 2025 -
Use of `clang-offload-bundler` needs to be updated
#1547 opened
Jan 15, 2025
4 Unresolved conversations
Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.
-
Change CMake to respect GPU_TARGETS variable
#1291 commented on
Jan 16, 2025 • 0 new comments -
Move frequency retrieval to the beginning and manual input when error
#1483 commented on
Jan 20, 2025 • 0 new comments -
Windows gfx1201
#1500 commented on
Jan 20, 2025 • 0 new comments -
[Experimental] Support on arbitrary M & K for siwzzle-A
#1521 commented on
Jan 20, 2025 • 0 new comments