issues
search
ROCm
/
triton
Development repository for the Triton language and compiler
MIT License
79
stars
22
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Add more unit tests to FA fwd kernels.
#609
xinyazhang
opened
4 days ago
2
Couple of FA optimizations
#608
vgokhale
opened
5 days ago
0
non atomic implementation of stream-k
#607
xiaohuguo2023
closed
4 days ago
0
[tuning] gemm tuning script v3.3
#606
zhanglx13
opened
6 days ago
0
Add a script for tuning flash attention kernels
#605
yiqian1
opened
1 week ago
0
[Issue]: FA Performance forward Kernel Segfault under certain cases
#604
xinyazhang
opened
1 week ago
1
basic nightly
#603
micmelesse
opened
1 week ago
0
Fixed streamk kernel bug
#602
ravil-mobile
opened
1 week ago
3
Fixed streamk kernel bug
#601
ravil-mobile
closed
1 week ago
0
prune LDS usage for the new pipeliner
#600
jtang10
closed
1 week ago
0
Update one_config.py to support new input parameters
#599
yiqian1
closed
2 weeks ago
0
fix correctness test issue of tune_streamk
#598
xiaohuguo2023
closed
3 weeks ago
0
experiment cpu timer for do_bench
#597
xiaohuguo2023
closed
2 weeks ago
5
[Issue]: Triton Compiler Takes Indefinite Time in ttgir -> llir Stage.
#596
xinyazhang
closed
1 week ago
20
[Issue]: `tl.exp`, `tl.sin`, etc. result in segmentation fault on Fedora 40
#595
kenneth-ge
opened
1 month ago
0
Update tune_gemm.py to save benchmarking results in a file
#594
yiqian1
closed
1 month ago
4
[Issue]: Triton not building/stuck
#593
radna0
closed
1 month ago
1
Fix post-processing to exclude local_load and local_alloc
#591
zhanglx13
closed
1 month ago
0
Add infrastructure to allow for kernel instrumentation passes to be inserted into the pass pipeline
#590
CRobeck
closed
1 month ago
1
[Feature]: Debugging Support
#589
Hprairie
opened
1 month ago
2
Add rotating tensor, icache flush, and bias to GEMM tuning script
#588
scxiao
closed
3 weeks ago
2
Add support for layouts
#587
vgokhale
closed
1 month ago
0
Skip test_op_bwd
#586
micmelesse
closed
1 month ago
0
Change all block pointers to tensor pointers
#585
vgokhale
closed
1 month ago
0
Change all block pointers to regular tensor pointers
#584
vgokhale
closed
1 month ago
0
test
#583
micmelesse
closed
1 month ago
0
change
#582
micmelesse
closed
1 month ago
0
seperate ci on triton-mlir and main
#581
micmelesse
closed
1 month ago
0
add perf-kernels
#580
micmelesse
closed
1 month ago
0
new stream-k kernel implementations
#579
xiaohuguo2023
closed
1 month ago
0
Use absolute paths in tune_gemm.py
#578
yiqian1
closed
1 month ago
0
[Feature]: Mark dev version as such
#577
fxmarty
closed
3 weeks ago
2
alibi backward
#576
micmelesse
closed
1 month ago
0
Groenenboomj/fixes causal
#575
groenenboomj
opened
1 month ago
0
Change all block pointers to tensor pointers
#574
vgokhale
closed
1 month ago
1
CI for Perf Kernels and test_core_amd.py
#573
micmelesse
closed
1 month ago
0
Update perf kernels readme
#571
vgokhale
closed
1 month ago
0
[Feature]: Do not recommend HIP_FORCE_DEV_KERNARG=1
#570
fxmarty
closed
1 month ago
2
[Triton][FA] Change block pointers to tensor pointers
#569
vgokhale
closed
6 days ago
0
Could amd triton load a fixed length list with `tl.load`?
#568
xinji1
closed
1 month ago
2
Mqa gqa bugfix
#567
vgokhale
closed
2 months ago
3
Fix varlen mqa test
#566
vgokhale
closed
2 months ago
0
Add read only benchmark and cmd line config capability
#565
vgokhale
closed
2 months ago
0
[Issue]: `error: operand #0 does not dominate this use`
#564
xinyazhang
opened
2 months ago
2
New base backwards kernel
#563
groenenboomj
closed
2 months ago
1
add kpack and matrix_instr_nonkdim for stream-k implementation
#562
xiaohuguo2023
closed
2 months ago
0
[release/internal/2.2.x] Only include HIP headers for triton
#561
jithunnair-amd
opened
2 months ago
0
Support head size <= 256
#560
vgokhale
closed
2 months ago
0
[Upstream backend] [PyTorch UT]: `Callback: Queue 0x7ef32c400000 aborting with error : HSA_STATUS_ERROR_EXCEPTION: An HSAIL operation resulted in a hardware exception. code: 0x1016`
#559
jataylo
closed
2 months ago
1
[Upstream backend] [Pytorch UT]: `TypeError: function takes exactly 6 arguments (34 given)`
#558
jataylo
closed
2 months ago
1
Next