Issues
- 1
Panic when do custom_call for gpu cuda
#15945 opened by knightXun - 2
StableHLO spec has signed integers instead of signless, but XLA appears to use signless integers
#15400 opened by efferifick - 3
- 8
[platform set error][GPU-cuda]NOT_FOUND: could not find registered compiler for platform CUDA -- was support for that platform linked in?
#15007 opened by FatJhon - 0
Error in compilation on GPU
#15744 opened by jakevdp - 5
[XLA:GPU] Failed to launch CUDA kernel: cutlass_gemm_with_dynamic_update_slice with Dynamic-Update Slice Fusion flags
#15265 opened by qGentry - 1
[XLA:CLIENT] Why force using U32 indices type when converting torch gather op
#15654 opened by Nullkooland - 1
- 1
Issue running AOT-compiled code: `proto.has_host_program_shape() No program shape found in the proto`
#15022 opened by joaospinto - 1
Autosharding fails with "absl::btree_map::at"
#15094 opened by rahul003 - 1
Cannot remove instruction: %all-reduce = f32[32]{0} all-reduce(f32[32]{0} %reduce.3), channel_id=1, replica_groups={{0}}, to_apply=%region_0.44, metadata={op_name="parallelize(train_step_shard_parallel)/jit(main)/reduce_sum[axes=(0, 1, 2)];"}
#15114 opened by huhuiqi7 - 13
Channel ids for collectives are not unique when custom sharding is inlined in XLA
#14600 opened by Tixxx - 5
[XLA:GPU] Check failed in collective_pipeliner when using gradient accumulation with non-unrolled loop
#14332 opened by qGentry - 2
Involuntary Full Rematerialization
#15207 opened by chaserileyroberts - 3
What is the difference between PJRT, IFRT, and TFRT?
#15168 opened by eaplatanios - 2
- 1
【test:se_gpu_pjrt_client_test】ERROR: No test targets were found, yet testing was requested
#15054 opened by FatJhon - 3
SPMD - Handle Gather/Scatter is Harcoded to IndexParallel
#13304 opened by ptoulme-aws - 3
ROCm CI build failed in all recent PRs
#14084 opened by apivovarov - 43
Build fails
#14352 opened by AleksKnezevic - 0
Strange Python dependencies required in `WORKSPACE` file for running C++ XLA code
#15024 opened by joaospinto - 1
Error occurred while building TensorFlow
#14129 opened by Prashant-Jagtap - 3
All Gather Combiner - Coalesces different dtypes
#13305 opened by ptoulme-aws - 5
[CPU] dot(f16, f32) -> f16 - crashes on cpu.
#14607 opened by apivovarov - 0
Add ppc64le support
#14826 opened by bhuntsman - 6
Error reported when building a project using an image
#14781 opened by yanminghui123 - 4
ReduceScatterCreator - Misses One Pattern
#13307 opened by ptoulme-aws - 12
[XLA:GPU] Fine-grained remat policy makes async/pipelined collectives execute in the main stream
#14397 opened by qGentry - 1
xla_extension.so: undefined symbol: _ZN3tsl15ProfilerSession6CreateERKN10tensorflow14ProfileOptionsE
#14462 opened by huhuiqi7 - 0
Error due to empty field
#14520 opened by LeoneChen - 0
JAX PJRT Client, GPU Build Fails. Linking error
#14475 opened by kranipa - 1
xla_extension.so: undefined symbol: descriptor_table_xla_2fxla_5fdata_2eproto
#14433 opened by huhuiqi7 - 9
How to use PJRT C API v.2
#13733 opened by janpfeifer - 3
Cannot use `XlaOp` from outer scope
#14299 opened by joelberkeley - 5
- 1
clang: error: linker command failed with exit code 1
#14321 opened by huhuiqi7 - 4
Compiling xla/stream_executor/gpu/gpu_timer_kernel_cuda.cu.cc [for tool] failed
#13889 opened by huhuiqi7 - 7
- 1
[auto sharding build issue] xla/service/gpu/runtime/gpublas_lt_matmul_thunk.h: No such file or directory
#13644 opened by dulingkang - 5
Build of target //xla/tools:run_hlo_module failed - no such target '@com_google_protobuf//:any_cc_proto'
#14172 opened by apivovarov - 11
Regression in using NVCC as a CUDA compiler: redzone_allocator_kernel_cuda compilation fails
#13460 opened by pearu - 1
Bug: Segfault when CUPTI is not correctly initialized.
#13853 opened by yliu120 - 0
- 0
- 4
- 2
HandleReshape in SPMD partitioner splits sharding in some cases where it is unoptimal
#13306 opened by ptoulme-aws - 6
[XLA:GPU] cuDNN norm rewriter fails with OOB shape access
#13361 opened by bchetioui - 2
Question : When do we not honor "preferred_element_type" ?
#13388 opened by HahTK - 3
How are random seeds set for RNG operations?
#13300 opened by Ebanflo42 - 1
Interleaved/Looping SPMD Pipelines
#13242 opened by ptoulme-aws