issues
search
ROCm
/
rocMLIR
123
stars
39
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Use half-precision math library calls
#1650
dhernandez0
opened
10 hours ago
1
Fix aliasing for LDS
#1649
dhernandez0
opened
15 hours ago
1
MLIR#1470: fix crash when blockPerCU == 0
#1648
krzysz00
closed
9 hours ago
2
Setup CMAKE_MODULE_PATH when LLVM_LIBDIR_SUFFIX is used
#1647
krzysz00
opened
1 day ago
0
[DO NOT SQUASH] Stop treating bf16 as i16
#1646
krzysz00
closed
9 hours ago
1
Add `.amdhsa_code_object_version` metadata serializing rocDL modules
#1645
umangyadav
opened
1 day ago
1
Use cmake -E touch for cross-platform compatibility
#1644
stefankoncarevic
closed
1 day ago
2
Move enableApplicability to allow ReuseLDSPass to fail if there is not enough LDS memory
#1643
dhernandez0
closed
2 days ago
3
[Issue]: Exhaustive tunning for attention fails
#1642
dhernandez0
closed
2 days ago
0
Upstream use of target attributes for compilation
#1641
umangyadav
opened
6 days ago
4
Fix issue 1620
#1640
icobg
closed
1 week ago
1
Add support for `pad` + removeSubDims in `removeUpperDims`
#1639
manupak
opened
1 week ago
2
Fix: Correctly lower FP8 instructions for specific architectures.
#1638
stefankoncarevic
closed
1 week ago
3
Fix overly-strict guards in LLVM conversions for fp8 intrinsic
#1637
krzysz00
closed
1 week ago
1
[Issue]: Installation Failes
#1636
Dustin-rpg
opened
1 week ago
3
disable occupancy warnings
#1635
umangyadav
closed
1 week ago
3
Handle the new F8 types in RockTuningImpl.cpp too. Oops.
#1634
pcf000
closed
6 days ago
4
Fix verifyGemmTypes for fp8
#1633
dhernandez0
closed
1 week ago
1
Report error when a library fails to load
#1632
dhernandez0
closed
1 week ago
3
Fix removeUpperDims
#1631
manupak
closed
2 weeks ago
1
Fix test for input vectorization traversal, use types correctly, add …
#1630
krzysz00
closed
2 weeks ago
1
[DO NOT SQUASH] Handle non-zero-preserving input fusions, make read_into track validity
#1629
krzysz00
opened
2 weeks ago
1
Handle process exception from calling rocminfo, to see why it sometimes fails.
#1628
pcf000
closed
2 weeks ago
1
Fix hard-coded '5' that needs to be inputDimension.size() to handle 3-D convolutions.
#1627
pcf000
closed
2 weeks ago
1
Fix too-strict test in fp8 emulation chenks
#1626
krzysz00
closed
2 weeks ago
1
Use separate call to check for `gfx11`
#1625
umangyadav
closed
2 weeks ago
3
Add HIP API logs for the Navi3x nightly builds
#1624
umangyadav
opened
2 weeks ago
0
Add 3-D layouts to conv regression tests and fix the problems exposed
#1623
pcf000
closed
2 weeks ago
1
Collected Jenkinsfile tweaks for reliability.
#1622
pcf000
closed
3 weeks ago
2
Fix crash arising from insufficient guards in WMMA instruction selector
#1621
krzysz00
closed
3 weeks ago
1
[Issue]: fatal error: 'mlir/Conversion/RocMLIRPasses.h.inc' file not found + patch
#1620
icobg
closed
2 days ago
3
[DO NOT SQUASH][EXTERNAL] Add a scheduling barrier guard around inlineAsm lds.barrier
#1619
manupak
closed
3 weeks ago
3
Error: 'tensor.expand_shape' op expected dimension 1 of collapsed type to be static value of 320
#1618
pfultz2
opened
3 weeks ago
1
Reduced split-k range
#1616
djramic
closed
1 week ago
20
[Attention] Fixed preSoftmaxElementwiseRegion input ordering
#1615
manupak
closed
3 weeks ago
2
[DO NOT SQUASH] Upstream merge August'24
#1614
manupak
closed
3 weeks ago
5
Add support for tuning split-k with convolutions
#1613
djramic
closed
1 week ago
3
[DO NOT SQUASH] Fp8 support for gfx12
#1612
giuseros
closed
3 weeks ago
3
Make LDS one big pool so we can allocate/deallocate/reuse it
#1611
dhernandez0
closed
2 weeks ago
7
Fix tests where no GPU is present
#1610
krzysz00
closed
4 weeks ago
0
[DO NOT SQUASH][EXTERNAL] fix plumbing of rocdl attrs: waves_per_eu & unsafe_atomics
#1609
manupak
closed
1 month ago
3
Buildbot improvements and fixes.
#1608
pcf000
closed
4 weeks ago
1
use removeUpperDims if possible
#1607
dhernandez0
closed
4 weeks ago
2
Fix code-coverage version mismatch -- PATH in Jenkinsfile doesn't apply to sh
#1606
pcf000
closed
1 month ago
1
[BACKPORT] Fix dequantizelinear definition
#1605
CharlieL7
closed
4 weeks ago
1
[CI] Created new CI job pipeline for Navi4x architecture
#1604
stefankoncarevic
closed
2 days ago
4
Move threadwise_copy ops in gridwise_gemm_accel, pipeline non-accel
#1603
krzysz00
closed
4 weeks ago
1
Update CAPI tests to include C++ tests
#1602
fabianmcg
closed
1 month ago
1
Do not do TransposeRewritePattern if the operation has more than one use
#1601
dhernandez0
closed
1 month ago
3
Move ops inside k-Loop into pipeline stages
#1600
manupak
closed
4 weeks ago
0
Next