Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit history

Commits on Aug 2, 2024

• add emscripten hello world implementation (working), update matrix show() to use snprintf instead of sprintf

  austinvhuang

  committed
  7fe8d8b

  Copy full SHA for 7fe8d8b

Commits on Jul 21, 2024

• change toGPU to allow ahead-of-time creation of the staging buffer, fix out-of-date docstrings

  austinvhuang

  committed
  d6a02d5

  Copy full SHA for d6a02d5

Commits on Jul 15, 2024

• fix docstring formatting for doxygen consumption

  austinvhuang

  committed
  356adc7

  Copy full SHA for 356adc7

Commits on Jul 12, 2024

• replace wgpuQueueWriteBuffer for params in examples to an overload of toGPU. Update readme based on internal blog post content

  austinvhuang

  committed
  33cb63a

  Copy full SHA for 33cb63a

Commits on Jul 11, 2024

• shadertui live reloading demo initial implementation

  austinvhuang

  committed
  fb56cf0

  Copy full SHA for fb56cf0

Commits on Jul 10, 2024

• add -stdlib=libc++ to build flags, minor cleanup

  austinvhuang

  committed
  f983bcf

  Copy full SHA for f983bcf

Commits on Jul 9, 2024

• 1D blocktiling matmul works, set as the default

  austinvhuang

  committed
  dda3934

  Copy full SHA for dda3934

• matmul cleanup (kernel v3 partly working for small tile sizes), minor readme edits

  austinvhuang

  committed
  a11c98d

  Copy full SHA for a11c98d

Commits on Jul 4, 2024

• Specify xyz of number of workgroups for createKernel (ala cuda) instead of specifying total # threads (and automatically dividing each dimension by workgroup size). tiled matmul shader now produces…

  Show description for f72c857
  austinvhuang

  committed
  f72c857

  Copy full SHA for f72c857

Commits on Jul 1, 2024

• debugging mm example variations

  austinvhuang

  committed
  3781b21

  Copy full SHA for 3781b21

Commits on Jun 29, 2024

• add arrayutil overloads, debugging mm

  austinvhuang

  committed
  829ddc5

  Copy full SHA for 829ddc5

Commits on Jun 28, 2024

• migrate from google style casing to webgpu camelcasing, qkv projection (wip)

  austinvhuang

  committed
  65ae859

  Copy full SHA for 65ae859

• use wgpu style convention for functions

  austinvhuang

  committed
  fae185f

  Copy full SHA for fae185f

Commits on Jun 27, 2024

• Add inline to avoid duplicated-symbols-error

  junjihashimoto

  committed
  3affc31

  Copy full SHA for 3affc31

Commits on Jun 23, 2024

• minor cleaup + release prep

  austinvhuang

  committed
  34c17a9

  Copy full SHA for 34c17a9

Commits on Jun 18, 2024

• fix build (disable error-on-warnings), WIP multikernel support for nThreads specificaiton

  austinvhuang

  committed
  adf4ce3

  Copy full SHA for adf4ce3

Commits on Jun 17, 2024

• handle 3D nthreads -> workgroup mappings for CreateKernel, fix GELU bug for inputs > 10.0, get rid of inferred # threads from output from output dimensions (too easy to introduce bugs), refine rend…

  Show description for 0735132
  austinvhuang

  committed
  0735132

  Copy full SHA for 0735132

Commits on Jun 11, 2024

• Tensor() factory function -> CreateTensor(), update README

  austinvhuang

  committed
  ae6f7ce

  Copy full SHA for ae6f7ce

Commits on Jun 10, 2024

• multikernel test with 2 shaders, move shaders to nn/shaders.h

  austinvhuang

  committed
  c63d9b2

  Copy full SHA for c63d9b2

Commits on Jun 2, 2024

• initial source code migration

  austinvhuang

  committed
  55210c8

  Copy full SHA for 55210c8

Loading