Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Sep 5, 2024

• Add folding for basic sum situations.

  Show description for 2eb9e9c
  

  LukeBoyer

  authored and

  tensorflower-gardener

  committed
  2eb9e9c

  Copy full SHA for 2eb9e9c

• Enable Cuda Graph in Xprof Analysis Server

  Show description for 67e441c
  tensorflower-gardener

  committed
  67e441c

  Copy full SHA for 67e441c

• Merge pull request #75100 from tensorflow:fixtypos03

  Show description for 8f45072
  tensorflower-gardener

  committed
  8f45072

  Copy full SHA for 8f45072

• Rolling back due to breakage

  Show description for 6b5ab9f
  

  SandSnip3r

  authored and

  tensorflower-gardener

  committed
  6b5ab9f

  Copy full SHA for 6b5ab9f

• Create a copy for the while instruction if the following two conditions hold.

  Show description for b2fead9
  

  ZixuanJiang

  authored and

  tensorflower-gardener

  committed
  b2fead9

  Copy full SHA for b2fead9

• Fix tanh for large inputs

  Show description for ff48a2f
  

  impjdi

  authored and

  tensorflower-gardener

  committed
  ff48a2f

  Copy full SHA for ff48a2f

• Ensure BFloat16Propagation respects if an instruction does not support mixed precision.

  Show description for 5b1c4dd
  

  SandSnip3r

  authored and

  tensorflower-gardener

  committed
  5b1c4dd

  Copy full SHA for 5b1c4dd

Commits on Sep 4, 2024

• Disable cholesky_op_test on gpu_h100

  Show description for 4aa9630
  tensorflower-gardener

  committed
  4aa9630

  Copy full SHA for 4aa9630

• Make GpuDriver::CreateStream return an absl::StatusOr<GpuStreamHandle>.

  Show description for 7643bed
  

  klucke

  authored and

  tensorflower-gardener

  committed
  7643bed

  Copy full SHA for 7643bed

• Don't require a docker container for a `Build` in build.py

  Show description for cc699ed
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  cc699ed

  Copy full SHA for cc699ed

• Move Exhaustive16BitBinaryTest range logging to inside the VLOG level check

  Show description for c99a295
  

  Gregory Pataky

  authored and

  tensorflower-gardener

  committed
  c99a295

  Copy full SHA for c99a295

• [IFRT] Change `CustomCallProgram::serialized_program_text` to use `absl::Cord`

  Show description for 427d57a
  

  hyeontaek

  authored and

  tensorflower-gardener

  committed
  427d57a

  Copy full SHA for 427d57a

• Rename generated_build_commands.txt to avoid confusion

  Show description for 842afe7
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  842afe7

  Copy full SHA for 842afe7

• Switch all GpuDriver APIs to work in terms of virtual Context objects rather than GpuContext concrete objects.

  Show description for 6e8a5fa
  

  klucke

  authored and

  tensorflower-gardener

  committed
  6e8a5fa

  Copy full SHA for 6e8a5fa

• Remove CUDA and NCCL repository rules calls from RBE configs.

  Show description for 2f7f2f9
  tensorflower-gardener

  committed
  2f7f2f9

  Copy full SHA for 2f7f2f9

• some args order are misplaced, swap them will not fix the msan issue, but at least it don't failed the old places.

  Show description for 5cd07a2
  

  trisolaran

  authored and

  tensorflower-gardener

  committed
  5cd07a2

  Copy full SHA for 5cd07a2

• Fix in-memory file feature detection test.

  Show description for 14cd8c7
  

  qukhan

  authored and

  tensorflower-gardener

  committed
  14cd8c7

  Copy full SHA for 14cd8c7

• Merge `exhaustive_binary_f32_f64_test` into `exhaustive_binary_16_bit_test` and rename to `exhaustive_binary_test`

  Show description for 90d1c70
  

  Gregory Pataky

  authored and

  tensorflower-gardener

  committed
  90d1c70

  Copy full SHA for 90d1c70

• Rollback: remove experimental GCC support in hermetic CUDA rules.

  Show description for 39d4959
  tensorflower-gardener

  committed
  39d4959

  Copy full SHA for 39d4959

• Add documentation for `config-cuda-only`

  Show description for bd89c04
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  bd89c04

  Copy full SHA for bd89c04

• Add folding for basic slice situations.

  Show description for 095ee95
  

  LukeBoyer

  authored and

  tensorflower-gardener

  committed
  095ee95

  Copy full SHA for 095ee95

• Merge pull request #75070 from tensorflow:lionelfeng-patch-1

  Show description for 9e99703
  tensorflower-gardener

  committed
  9e99703

  Copy full SHA for 9e99703

• Fix ClangTidy warning for missing header.

  Show description for 42192e2
  

  arfaian

  authored and

  tensorflower-gardener

  committed
  42192e2

  Copy full SHA for 42192e2

• Add direct legalizations for constant ops.

  Show description for 4b294ec
  

  LukeBoyer

  authored and

  tensorflower-gardener

  committed
  4b294ec

  Copy full SHA for 4b294ec

• Automated Code Change

  Show description for ba6b057
  tensorflower-gardener

  committed
  ba6b057

  Copy full SHA for ba6b057

• Reverts b0b76816f1315dfa73f05ab02a2b12b655e06ac6

  Show description for 6c39142
  tensorflower-gardener

  committed
  6c39142

  Copy full SHA for 6c39142

• Create goldens and `diff_test` for commands for all builds in `build.py`

  Show description for e4d2c23
  

  ddunl

  authored and

  tensorflower-gardener

  committed
  e4d2c23

  Copy full SHA for e4d2c23

• Fix typo in `weight_cache_test.cc`.

  Show description for 7b27eca
  

  qukhan

  authored and

  tensorflower-gardener

  committed
  7b27eca

  Copy full SHA for 7b27eca

• [XLA:GPU] Collect statistics for autotuning cache hits and misses.

  Show description for 90e094b
  

  dimitar-asenov

  authored and

  tensorflower-gardener

  committed
  90e094b

  Copy full SHA for 90e094b

• Default enable custom kernel fusions for V100 GPUs and test that we run triton gemms for using A100 GPUS and custom kernel fusions using V100.

  Show description for 85a79f8
  

  derdrdirk

  authored and

  tensorflower-gardener

  committed
  85a79f8

  Copy full SHA for 85a79f8

• Added check for using fregister-allocation compiler option.

  Show description for 2826678
  

  roserg

  authored and

  tensorflower-gardener

  committed
  2826678

  Copy full SHA for 2826678

• Introduce SetOptions interface and a visitor concept to `visit` passes and pipelines to `SetOptions`

  Show description for a902aaa
  

  vamsimanchala

  authored and

  tensorflower-gardener

  committed
  a902aaa

  Copy full SHA for a902aaa

• Make DeviceDescription mutable

  Show description for 43fbcb2
  

  beckerhe

  authored and

  tensorflower-gardener

  committed
  43fbcb2

  Copy full SHA for 43fbcb2

• Add experimental GCC support in hermetic CUDA rules.

  Show description for 12bfb5b
  tensorflower-gardener

  committed
  12bfb5b

  Copy full SHA for 12bfb5b

• [XLA:GPU] Fix a CHECK-fail by tightening the fusibility check in the SoftMax triton rewriter.

  Show description for 6a42067
  

  dimitar-asenov

  authored and

  tensorflower-gardener

  committed
  6a42067

  Copy full SHA for 6a42067