Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 30, 2024

• [xla:gpu] Unify static and dynamic slice cases for AddressComputationFusionRewriter

  Show description for fd091d4
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  fd091d4

  Copy full SHA for fd091d4

• compat: Update forward compatibility horizon to 2024-03-30

  Show description for c65e205
  tensorflower-gardener

  committed
  c65e205

  Copy full SHA for c65e205

• Update GraphDef version to 1817.

  Show description for c96e6c5
  tensorflower-gardener

  committed
  c96e6c5

  Copy full SHA for c96e6c5

• [xla:gpu][NFC] Remove unused constexprs

  Show description for 7e9d002
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  7e9d002

  Copy full SHA for 7e9d002

• [xla:gpu] Unify GEMM emission for (Dynamic)AddressComputationFusion emitters

  Show description for ce7a132
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  ce7a132

  Copy full SHA for ce7a132

• [xla:gpu] AddressComputationFusionRewriter should run before other fusions

  Show description for 27bd888
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  27bd888

  Copy full SHA for 27bd888

Commits on Mar 29, 2024

• Add a contextmanager to temporarily disable XLA sharding support for ResourceVariables.

  Show description for 2c90827
  

  swachhandl

  authored and

  tensorflower-gardener

  committed
  2c90827

  Copy full SHA for 2c90827

• In general, avoid the suffix on StatusOr.

  Show description for 30464e6
  

  ghpvnist

  authored and

  tensorflower-gardener

  committed
  30464e6

  Copy full SHA for 30464e6

• [PJRT C API] Add a PJRT extension to register custom partitioner.

  Show description for d6b38af
  

  Jieying Luo

  authored and

  tensorflower-gardener

  committed
  d6b38af

  Copy full SHA for d6b38af

• [xla:gpu] Add a version of `HloPredicateIsOp` for `HloInstructionAdaptor`

  Show description for 7810e73
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  7810e73

  Copy full SHA for 7810e73

• Implementation of get / retrieve buffer attributes methods in GPU async kernel.

  Show description for b25ab71
  tensorflower-gardener

  committed
  b25ab71

  Copy full SHA for b25ab71

• Reverts 3a2cd8887ed96de6abbd26e46844b959aa42e481

  Show description for 5bbc002
  

  Marcello Maggioni

  authored and

  tensorflower-gardener

  committed
  5bbc002

  Copy full SHA for 5bbc002

• Move `tsl/python` to `xla/tsl/python`

  Show description for ea867d6
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  ea867d6

  Copy full SHA for ea867d6

• Use `bytes` proto field type for string values of `xla::PjRtValueType` to `bytes`

  Show description for f606d83
  

  hyeontaek

  authored and

  tensorflower-gardener

  committed
  f606d83

  Copy full SHA for f606d83

• PR #7849: [XLA:CPU] Add support for cross-process collectives using mpi.

  Show description for 8958c65
  

  inailuig

  authored and

  tensorflower-gardener

  committed
  8958c65

  Copy full SHA for 8958c65

• Add new methods in backend async kernel interface. Reflect to the design of go/async-io-coherency

  Show description for 66ee739
  tensorflower-gardener

  committed
  66ee739

  Copy full SHA for 66ee739

• Add support for LocalDeviceManager in tfrt_session.

  Show description for 8dc94a2
  

  CMinge

  authored and

  tensorflower-gardener

  committed
  8dc94a2

  Copy full SHA for 8dc94a2

• Integrate LLVM at llvm/llvm-project@80aa52d8c5a8

  Show description for 1faaace
  

  gribozavr

  authored and

  tensorflower-gardener

  committed
  1faaace

  Copy full SHA for 1faaace

• Change include order in `ml_dtypes.cc` to prevent errors.

  Show description for 8012c68
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  8012c68

  Copy full SHA for 8012c68

• [xla:gpu][NFC] Simplify `collect_slice_info`

  Show description for c9cffcd
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  c9cffcd

  Copy full SHA for c9cffcd

• Modify the matrix class to keep track of both memory and communication resharding costs for a given edge as part of one matrix object.

  Show description for 1a9dbe8
  tensorflower-gardener

  committed
  1a9dbe8

  Copy full SHA for 1a9dbe8

• Rollback https://github.com/openxla/xla/commit/0ab2be0b5a575da3206d2c2f92b85e6346708405.

  Show description for 5c66d02
  

  reedwm

  authored and

  tensorflower-gardener

  committed
  5c66d02

  Copy full SHA for 5c66d02

• [tflite] Add the composite lowering logic for hardswish

  Show description for 19efc8f
  

  majiddadashi

  authored and

  tensorflower-gardener

  committed
  19efc8f

  Copy full SHA for 19efc8f

• Enhanced zeta readability based on the article

  Show description for bc4b9d8
  

  Paweł Paruzel

  authored and

  tensorflower-gardener

  committed
  bc4b9d8

  Copy full SHA for bc4b9d8

• [odml] Remove MHLO from CHLO->StableHLO lowering. Migrate JAX random lowering to StableHLO

  Show description for b36854a
  

  GleasonK

  authored and

  tensorflower-gardener

  committed
  b36854a

  Copy full SHA for b36854a

• Restore GOOGLE_CUDA guard in scoped_annotation.h

  Show description for 210d156
  tensorflower-gardener

  committed
  210d156

  Copy full SHA for 210d156

• Delete the redundant compilation_cache_test

  Show description for 370c98d
  tensorflower-gardener

  committed
  370c98d

  Copy full SHA for 370c98d

• Deduplicate inferred mesh shapes when try_multiple_mesh_shapes=true.

  Show description for bcf3638
  tensorflower-gardener

  committed
  bcf3638

  Copy full SHA for bcf3638

• Split hybrid quantized dot-like StableHLO ops into TFLite dequantize and float op

  Show description for 147638c
  

  doyeonkim0

  authored and

  tensorflower-gardener

  committed
  147638c

  Copy full SHA for 147638c

• [xla:gpu][NFC] Use meaningful constexpr

  Show description for 06f2799
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  06f2799

  Copy full SHA for 06f2799

• Update GraphDef version to 1816.

  Show description for 3016662
  tensorflower-gardener

  committed
  3016662

  Copy full SHA for 3016662

• compat: Update forward compatibility horizon to 2024-03-29

  Show description for 17aaaa5
  tensorflower-gardener

  committed
  17aaaa5

  Copy full SHA for 17aaaa5

• [xla:gpu][NFC] Add AddressComputationThunk test with GEMM operands sharing the same buffer

  Show description for f2a78c0
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  f2a78c0

  Copy full SHA for f2a78c0

• Convert quantized stablehlo.constant to tfl.pseudo_qconst

  Show description for 06ef8a6
  

  doyeonkim0

  authored and

  tensorflower-gardener

  committed
  06ef8a6

  Copy full SHA for 06ef8a6

• [xla:gpu] Create fake buffer allocations for embedded thunk

  Show description for a829ac0
  

  tyb0807

  authored and

  tensorflower-gardener

  committed
  a829ac0

  Copy full SHA for a829ac0

Pagination

Previous

Next