Owner

openxla

5 indexed repositories · View on GitHub

xla
A machine learning compiler for GPUs, CPUs, and ML accelerators

C++ · 4542 stars
stablehlo
Backward compatible ML compute opset inspired by HLO/MHLO

MLIR · 702 stars
xprof
A profiling and performance analysis tool for machine learning

C++ · 581 stars
tokamax
Tokamax: A GPU and TPU kernel library.

Python · 295 stars
shardy
MLIR-based partitioning system

MLIR · 210 stars

Beginner friendly Everything

Open All

[XLA:GPU] cuDNN int8 convolution with `feature_group_count=2` returns a wrong output channel Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#49123 · Opened Sep 19, 2026
[XLA] SPMD partitioner clamps a partitioned `dynamic-slice` per shard instead of globally Open

Difficulty 3/5 1-2 days Newbie friendliness 78/100

openxla/xla#49122 · Opened Sep 19, 2026
[XLA] AlgebraicSimplifier folds nested dynamic-slice / dynamic-update-slice without clamping the outer start Open

Difficulty 4/5 3-5 days Newbie friendliness 68/100

openxla/xla#49121 · Opened Sep 19, 2026
[XLA:GPU] dynamic-slice copy path writes a constant-offset `dynamic-update-slice` at the unclamped offse Open

Difficulty 4/5 3-5 days Newbie friendliness 68/100

openxla/xla#49120 · Opened Sep 19, 2026
[XLA:GPU] `IndexingMap` composition through `dynamic-update-slice` binds runtime variables to the wrong start operands Open

Difficulty 4/5 3-5 days Newbie friendliness 42/100

openxla/xla#49119 · Opened Sep 19, 2026
[XLA] SPMD partitioner drops a dynamic-update-slice whose start index needs clamping Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

openxla/xla#49118 · Opened Sep 19, 2026
GPU compilation is not reproducible: GpuHloCostAnalysis sums float utilizations in pointer-hash order, PriorityFusion near-ties flip between processes Open

Difficulty 3/5 1-2 days Newbie friendliness 72/100

openxla/xla#49047 · Opened Sep 18, 2026
[SPMD] TopK produces incorrect local index shapes when returned directly Open

bug Google-TPU sharding stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#49005 · 1 comment · Opened Sep 17, 2026
[GPU] WSL2 large-allocation regression after switching device memory to exportable VMM Open

bug GPU NVIDIA-GPU

Difficulty 5/5 Over a week Newbie friendliness 35/100

openxla/xla#48979 · Opened Sep 17, 2026
Interpreter crashes on convolution with non-canonical output dimension ordering Open

Difficulty 3/5 1-2 days Newbie friendliness 72/100

openxla/stablehlo#3011 · Opened Sep 15, 2026
PallasTritonGatedLinearUnit tile heuristic ignores per-block shared-memory capacity (RESOURCE_EXHAUSTED on Blackwell parts with <=96 SMs) Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

openxla/tokamax#1361 · Opened Sep 15, 2026
StableHLO interpreter errors on f32 complex types on fft op Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

openxla/stablehlo#3010 · Opened Sep 15, 2026
Specfication of what happens in the case of integer divison by zero Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

openxla/stablehlo#3009 · Opened Sep 13, 2026
[XLA] RET_CHECK abort compiling reduce_window with an all-1 window and only negative padding Open

bug CPU err: Runtime stat:awaiting response from contributor

Difficulty 3/5 1-2 days Newbie friendliness 68/100

openxla/xla#48719 · 1 comment · Opened Sep 11, 2026
[XLA] ScatterExpander builds the index-validity bound in the index type, dropping valid updates with narrow index types Open

bug CPU stat:awaiting openxla-eng

Difficulty 3/5 1-2 days Newbie friendliness 72/100

openxla/xla#48718 · 1 comment · Opened Sep 11, 2026
[XLA] Constant unsigned gather / dynamic_slice index with the MSB set folds to index 0 (getSExtValue) Open

bug CPU err: Runtime

Difficulty 3/5 1-2 days Newbie friendliness 78/100

openxla/xla#48717 · 1 comment · Opened Sep 11, 2026
[XLA] AlgebraicSimplifier folds a negative (cropping) pad into reduce_window / convolution window padding Open

bug CPU python

Difficulty 3/5 1-2 days Newbie friendliness 78/100

openxla/xla#48715 · Opened Sep 11, 2026
[XLA:CPU] reduce_window on the YNNPACK path ignores base_dilation when window size and stride are 1 Open

bug CPU err: Runtime python stat:awaiting openxla-eng

Difficulty 3/5 1-2 days Newbie friendliness 75/100

openxla/xla#48714 · 1 comment · Opened Sep 11, 2026
[XLA:CPU] Wrong results / out-of-bounds read in f32 convolution with negative low padding Open

bug CPU

Difficulty 3/5 1-2 days Newbie friendliness 72/100

openxla/xla#48713 · 1 comment · Opened Sep 11, 2026
Mosaic-GPU flash attention (sm90/sm100) fails to trace when `jax_enable_x64` is enabled Open

Difficulty 3/5 1-2 days Newbie friendliness 74/100

openxla/tokamax#1349 · Opened Sep 11, 2026
[GPU] Populate GB200 matmul perf table for sol cost model Open

enhancement GPU NVIDIA-GPU

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#48627 · Opened Sep 10, 2026
[Unit Test][Python] socket_bulk_transport_test is flaky Open

bug stat:awaiting openxla-eng

Difficulty 3/5 1-2 days Newbie friendliness 72/100

openxla/xla#48239 · 2 comments · Opened Sep 1, 2026
[XLA:GPU] TopK custom kernel (__gpu$TopK for k <= 16) violates IEEE 754 TotalOrder and corrupts output when -NaN is present Open

GPU

openxla/xla#48119 · 1 assignee · Opened Aug 31, 2026
[XLA:Benchmarks] Generate distinct inputs for each benchmark run Open

err:performance

openxla/xla#48047 · 1 assignee · Opened Aug 28, 2026
[GPU] SM107 compilation fails with CUDA 12.9 because sm_107a requires PTX 9.4 Open

err:Build GPU

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47982 · Opened Aug 27, 2026
ShardingGroupMap: un-densified sdy.sharding_group group_id crashes standalone propagation passes (OOB / OOM) Open

Difficulty 3/5 1-2 days Newbie friendliness 64/100

openxla/shardy#1555 · Opened Aug 27, 2026
[SPMD] Batch-sharded grouped convolution all-gathers the full batch, redundantly computes on every partition, then slices Open

err:performance sharding

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47945 · Opened Aug 26, 2026
Fixed CTA-per-update assignment limits ragged-all-to-all device kernel throughput Open

err:performance GPU

Difficulty 5/5 Over a week Newbie friendliness 35/100

openxla/xla#47868 · 1 comment · 2 reactions · Opened Aug 25, 2026
ROCm BLASLt algorithm ordinal can resolve to different solutions across processes Open

AMD-GPU question

openxla/xla#47659 · 3 comments · 1 assignee · Opened Aug 21, 2026
Dynamic-slice fusion hides Mosaic-GPU collective kernels from collective-memory coloring Open

err: Runtime NVIDIA-GPU

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47461 · 2 comments · Opened Aug 17, 2026
452 StableHLO modules crash without producing a meaningful stack trace Open

bug

Difficulty 5/5 Over a week Newbie friendliness 25/100

openxla/xla#47398 · Opened Aug 16, 2026
hlo-translate failure: xla_builder.h:199 builder_ != nullptr Open

bug stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 68/100

openxla/xla#47395 · 1 comment · Opened Aug 16, 2026
hlo-translate failure: segv:mlir::stablehlo::ReduceOp::build Open

bug stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47394 · 1 comment · Opened Aug 16, 2026
hlo-translate failure: segv:mlir::stablehlo::foldConvert Open

bug stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 62/100

openxla/xla#47393 · 1 comment · Opened Aug 16, 2026
Failure on CPU: copy_insertion.cc:427 param != root Open

bug CPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47392 · 1 comment · Opened Aug 16, 2026
Failure on CPU: common_pjrt_client.cc:1617] ExecutePrepareWithOomRetries failed: INVALID_ARGUMENT: Buffer… Open

bug CPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47391 · 1 comment · Opened Aug 16, 2026
Failure on Interpreter: dynamic_dimension_inference.cc:1962 operand_shape.IsTuple() Only tuple typed inputs can have dynamic dimension… Open

bug stat:awaiting response from contributor

Difficulty 4/5 3-5 days Newbie friendliness 50/100

openxla/xla#47390 · 1 comment · Opened Aug 16, 2026
2 StableHLO IRs abort at the check: `shape.h:803 state Expected a tuple shape. Got …` Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 35/100

openxla/xla#47389 · Opened Aug 16, 2026
2 StableHLO IRs abort at the check: `shape.cc:254 size >= N (…) the N-th dimension is static and must …` Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47388 · Opened Aug 16, 2026
9 StableHLO IRs abort at the check: `indexing_analysis.cc:1670 ComputeOutputToInputIndexing is not implemented for opcode …` Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 35/100

openxla/xla#47387 · Opened Aug 16, 2026
2 StableHLO IRs abort at the check: `hlo_instructions.cc:2415 fused_root->IsFusible() …` Open

bug CPU NVIDIA-GPU

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47386 · Opened Aug 16, 2026
Crash on GPU: all_reduce_thunk.cc:87 reduction_kind.has_value() Open

bug NVIDIA-GPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 58/100

openxla/xla#47383 · 1 comment · Opened Aug 16, 2026
Crash on GPU: triton_tiling_propagation.cc:627 subdim_size_accumulator == dim_size Open

bug NVIDIA-GPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 52/100

openxla/xla#47381 · 1 comment · Opened Aug 16, 2026
Crash on GPU: host_offload_utils.cc:98 !tmp_shape_index.empty() Expected shape index to be non-empty. Open

bug NVIDIA-GPU stat:awaiting openxla-eng

openxla/xla#47380 · 1 comment · 1 assignee · Opened Aug 16, 2026
Crash on CPU: abstract_cpu_buffer.cc:92 literal.size_bytes() == size Open

bug CPU stat:awaiting response from contributor

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47379 · 1 comment · Opened Aug 16, 2026
Crash on GPU: hoist_fused_bitcasts.cc:168 ShapeUtil::ArrayDataSize(…) == ShapeUtil::ArrayDataSize(instruction->shape())… Open

bug NVIDIA-GPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 40/100

openxla/xla#47378 · 1 comment · Opened Aug 16, 2026
Crash on GPU: hlo_instruction.cc:5794 parent_ != nullptr Instruction <I> must have a parent in order to have a unique… Open

bug NVIDIA-GPU stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47377 · 1 comment · Opened Aug 16, 2026
Crash on Interpreter: shape.h:852 buffer != nullptr Open

bug stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47376 · 1 comment · Opened Aug 16, 2026
Crash on Interpreter: while_util.cc:230 while_instr->shape().IsTuple() Open

bug stat:awaiting openxla-eng

Difficulty 4/5 3-5 days Newbie friendliness 45/100

openxla/xla#47375 · 1 comment · Opened Aug 16, 2026
Crash on CPU: ynn_emitter.cc:456 reduce_instr->to_apply()->instruction_count() == N Open

bug CPU stat:awaiting response from contributor

Difficulty 4/5 3-5 days Newbie friendliness 48/100

openxla/xla#47374 · 1 comment · Opened Aug 16, 2026

Showing the newest 100

This page lists what was indexed most recently. Advanced filter has the whole inventory, narrowed by language, difficulty and how long a task takes.

Open advanced filter

openxla

Showing the newest 100

Get new issues in your inbox