aboutsummaryrefslogtreecommitdiff
AgeCommit message (Expand)Author
16 hoursInternal code changeupstream-masterA. Unique TensorFlower
16 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
16 hoursAdd an optional MemoryKind parameter to IFRT IR shardings.Ionel Gog
16 hoursAlways propagate shardings on kAllReduce.Parker Schuh
18 hoursSupport delayed CPU memory allocation in PjRt CPU.Yue Sheng
18 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
19 hoursRollback of PR #12076Penporn Koanantakool
19 hoursFix unreachable code in TFLiteA. Unique TensorFlower
19 hours#tensorflow Skip failing test.Yang Chen
19 hoursImplement Freeze() in IfrtServingExecutable to release mlir moduleDeqiang Chen
19 hoursAdd `bzl_library` targets to XLA and TSLDavid Dunleavy
20 hoursUnified the GPU executable proto used in AOT and JIT caching path.Jieying Luo
20 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
20 hours[XLA] Introduce pass which ensures memory is in the right memory space before...Victor Stone
21 hoursMake CommandBuffer use StreamExecutorInterface rather than StreamExecutor.Kyle Lucke
21 hoursReport a failure when detecting top-level compilation marker in TF2XLA Non-ep...Jian Cai
22 hoursUse `ShapeUtil::HumanString` instead of calling `Shape::ToString` directly in...Gunhyun Park
22 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
22 hoursDelete chlo.dynamic_reshape and downstream opsMichael Levesque-Dion
22 hoursAdd `MhloDynamicReshape` to XLA client API.Gunhyun Park
22 hoursMake Kernel classes use StreamExecutorInterface rather than StreamExecutor.Kyle Lucke
23 hoursTurns off `.weighted_flat_map` random access compatibility because it does no...Jim Lin
23 hoursUse StreamExecutorInterface in Event rather than StreamExecutor.Kyle Lucke
23 hours[Triton] Fixing issues with pipelining that cause crashes for some use-cases ...Mohammed Anany
23 hoursXProf GPU: Cache CUPTI activity buffer during tracing, process them after tra...A. Unique TensorFlower
24 hoursAdd integration options in TFRT/IFRT with tensorflow serving.Deqiang Chen
24 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
24 hoursPR #12050: [XLA:GPU] Force nccl command buffer to update even memory pointers...Shawn Wang
25 hoursPR #12171: [XLA: GPU] add more checks for barrierCmd unittestShawn Wang
25 hoursEnhance error message when creating RepeatBufferKernel fails.Ruoxin Sang
26 hoursPR #12012: Update cuDNN frontend version.Elfie Guo
26 hoursUpdate MHLO README to indicate current statusMichael Levesque-Dion
26 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
27 hoursIntegrate LLVM at llvm/llvm-project@6217abce86b5Benjamin Kramer
27 hours[XLA:GPU] Added more logs in the autotuner to allow for better benchmark debu...Mohammed Anany
28 hoursPR #12031: add gemm autotune log dumpAmir Samani
28 hoursFix reduction group computation for MLIR emitter.Johannes Reifferscheid
28 hours[XLA:GPU] Cast bf16 matmuls to fp32 on pre-Ampere GPUsGeorge Karpenkov
28 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
29 hours[XLA:GPU] Make symbolic tiles use a single indexing map under the hoodTamás Danyluk
30 hoursPR #11967: [XLA:GPU] Add cuDNN flash attention dbias computation and broadcas...Shanbin Ke
30 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower
31 hoursIntegrate LLVM at llvm/llvm-project@18e7dcb7c576Benjamin Kramer
31 hoursDon't MOF computations that are already fused.Johannes Reifferscheid
31 hoursRewrite all reductions to race-free in MLIR mode.Johannes Reifferscheid
31 hoursUpdate GraphDef version to 1854.A. Unique TensorFlower
32 hourscompat: Update forward compatibility horizon to 2024-05-06A. Unique TensorFlower
32 hoursUse the correct dependency for the included gpu_init.h header.Adrian Kuegel
32 hoursCompute non-fusion computations just once.Johannes Reifferscheid
32 hoursUpdate ops-related pbtxt files.A. Unique TensorFlower