Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 4, 2025

• Added num_nonbatch_dims to the operator object.

  Show description for f38c111
  

  Aelphy

  authored and

  xnnpack-bot

  committed
  f38c111

  Copy full SHA for f38c111

• update googlebenchmark to v1.9.1-22-g4a805f9

  Show description for 411f9de
  xnnpack-bot

  committed
  411f9de

  Copy full SHA for 411f9de

Commits on Feb 3, 2025

• scripts/build-android-*.sh: bump ANDROID_PLATFORM to 24

  Show description for 7702e72
  xnnpack-bot

  committed
  7702e72

  Copy full SHA for 7702e72

• Fix warning "implicit conversion from 'int' to 'const float' changes value..."

  Show description for 5e223a0
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  5e223a0

  Copy full SHA for 5e223a0

• Update to latest Android NDK

  Show description for f43f507
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  f43f507

  Copy full SHA for f43f507

• Add LayerNorm benchmark

  Show description for a73fa69
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  a73fa69

  Copy full SHA for a73fa69

• Align stack pointer in x64 kernels

  Show description for c467d1a
  

  alankelly

  authored and

  xnnpack-bot

  committed
  c467d1a

  Copy full SHA for c467d1a

• Update KleidiAI version to v1.3.0.

  Show description for 6a834a0
  

  gonnet

  authored and

  xnnpack-bot

  committed
  6a834a0

  Copy full SHA for 6a834a0

• Increase the number of cores used the cmake-linux-riscv64 workflow to make builds a bit faster.

  Show description for 1077850
  

  gonnet

  authored and

  xnnpack-bot

  committed
  1077850

  Copy full SHA for 1077850

• Use fpritnf(stderr, ...) instead of xnn_log_info for the KleidiAI version information.

  Show description for 7fdfa1c
  

  gonnet

  authored and

  xnnpack-bot

  committed
  7fdfa1c

  Copy full SHA for 7fdfa1c

• Added transient buffers tests to conv2d + formatting fix in conv2d operator.

  Show description for 9fbdd87
  

  Aelphy

  authored and

  xnnpack-bot

  committed
  9fbdd87

  Copy full SHA for 9fbdd87

• Remove unused bf16 minmax params

  Show description for bc0cb7b
  

  alankelly

  authored and

  xnnpack-bot

  committed
  bc0cb7b

  Copy full SHA for bc0cb7b

• Don't apply the clang stack alignment flags if compiling on fuchsia or ios.

  Show description for 311e2ee
  

  gonnet

  authored and

  xnnpack-bot

  committed
  311e2ee

  Copy full SHA for 311e2ee

• Rollback of "XNNPACK: update googlebenchmark to v1.9.1-22-g4a805f9"

  Show description for 29e4204
  

  gonnet

  authored and

  xnnpack-bot

  committed
  29e4204

  Copy full SHA for 29e4204

Commits on Feb 2, 2025

• Add SDE 9.48 Diamond Rapids, Panther Lake and ClearWater Forest

  Show description for 8c469bf
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  8c469bf

  Copy full SHA for 8c469bf

Commits on Feb 1, 2025

• Add a few more flags to control slinky behavior

  Show description for f389840
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  f389840

  Copy full SHA for f389840

Commits on Jan 31, 2025

• Add dfsan stubs for assembly kernels to fix linker errors in dfsan builds

  Show description for ca8f209
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  ca8f209

  Copy full SHA for ca8f209

• update googlebenchmark to v1.9.1-22-g4a805f9

  Show description for 2aa56d8
  xnnpack-bot

  committed
  2aa56d8

  Copy full SHA for 2aa56d8

• Fix bug in avx512vnni input register allocation.

  Show description for 03729c4
  

  gonnet

  authored and

  xnnpack-bot

  committed
  03729c4

  Copy full SHA for 03729c4

• Update SDE from 9.44 to 9.48 for github testing on Intel

  Show description for a4da1c1
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a4da1c1

  Copy full SHA for a4da1c1

Commits on Jan 30, 2025

• Disable stack alignment options on iOS

  Show description for dcfbadd
  xnnpack-bot

  committed
  dcfbadd

  Copy full SHA for dcfbadd

• Add an elementwise op sequence benchmark to bench/models

  Show description for 614491a
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  614491a

  Copy full SHA for 614491a

• Fix debug build with CMake for rndnu16

  Show description for b8d5d5a
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  b8d5d5a

  Copy full SHA for b8d5d5a

• Disable x64 bf16->f32 GEMM microkernels due to crash

  Show description for 36ab224
  

  alankelly

  authored and

  xnnpack-bot

  committed
  36ab224

  Copy full SHA for 36ab224

• Reapply the patch with stack alignment options.

  Show description for 2776f3b
  xnnpack-bot

  committed
  2776f3b

  Copy full SHA for 2776f3b

Commits on Jan 29, 2025

• Set the stack alignment to the size of the underlying vector type for x86 microkernel builds.

  Show description for a315385
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  a315385

  Copy full SHA for a315385

• Enable pf32 and pf16 left-hand sides for batch-matrix-multiply.

  Show description for 2cb1b3a
  

  gonnet

  authored and

  xnnpack-bot

  committed
  2cb1b3a

  Copy full SHA for 2cb1b3a

• Uncomment accidently commented tests

  Show description for d326ad1
  

  alankelly

  authored and

  xnnpack-bot

  committed
  d326ad1

  Copy full SHA for d326ad1

• Protect against OOB reads during packing

  Show description for 681c768
  

  alankelly

  authored and

  xnnpack-bot

  committed
  681c768

  Copy full SHA for 681c768

• bf16 f32 FC operator

  Show description for 59055be
  

  alankelly

  authored and

  xnnpack-bot

  committed
  59055be

  Copy full SHA for 59055be

Commits on Jan 28, 2025

• bf16->f32 avx512bf16 GEMM microkernels

  Show description for d7f398e
  

  alankelly

  authored and

  xnnpack-bot

  committed
  d7f398e

  Copy full SHA for d7f398e

• Log the KleidiAI version being used.

  Show description for 0c3e255
  

  gonnet

  authored and

  xnnpack-bot

  committed
  0c3e255

  Copy full SHA for 0c3e255

• Set the stack alignment to the size of the underlying vector type for x86 microkernel builds.

  Show description for 41dd468
  

  gonnet

  authored and

  xnnpack-bot

  committed
  41dd468

  Copy full SHA for 41dd468

• Remove more vld1_dup_* usage (fixes google#7727)

  Show description for f8e3c97
  

  dsharletg

  authored and

  xnnpack-bot

  committed
  f8e3c97

  Copy full SHA for f8e3c97

• Visual C set /vlen=256 for avx256 microkernel builds

  Show description for ffcaa54
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  ffcaa54

  Copy full SHA for ffcaa54

Pagination

Previous

Next