Code

Add XPU support (duplicate #125)#209

Code

Open

ma595 wants to merge 26 commits intomainCambridge-ICCS/FTorch:mainfrom add-devices-iccsCambridge-ICCS/FTorch:add-devices-iccsCopy head branch name to clipboard

+201-96

Commits

Commits on Dec 16, 2024

• Add MacOS GPU device option

  

  ElliottKasoar

  authored and

  ma595

  committed
  aa714fe

  Copy full SHA for aa714fe

• Add XPU device option

  

  ElliottKasoar

  authored and

  ma595

  committed
  6a96d49

  Copy full SHA for 6a96d49

• Update C++ XPU interface to handle multiple devices indices.

  

  jatkinson1000

  authored and

  ma595

  committed
  84a4e5d

  Copy full SHA for 84a4e5d

Commits on Dec 20, 2024

• Update ftorch.F90 for XPU support

  ma595

  committed
  4c2fc90

  Copy full SHA for 4c2fc90

• Add xpu python modifications to examples 1 and 2

  ma595

  committed
  34316e0

  Copy full SHA for 34316e0

• Add xpu modifications to fortran for example 2, init still not called

  ma595

  committed
  d08ce7e

  Copy full SHA for d08ce7e

Commits on Jan 13, 2025

• Build example 3 if CUDA and MPI enabled

  jwallwork23

  committed
  9d4aa86

  Copy full SHA for 9d4aa86

• Put model on CUDA device in simplenet

  jwallwork23

  committed
  5f4418a

  Copy full SHA for 5f4418a

• Run example 3 if it's been built

  jwallwork23

  committed
  b864b90

  Copy full SHA for b864b90

• Add missing imports for pt2ts

  jwallwork23

  committed
  911989b

  Copy full SHA for 911989b

• More helpful output for simplenet_infer_python

  jwallwork23

  committed
  4591325

  Copy full SHA for 4591325

• Fix numbering in CMakeLists for examples

  jwallwork23

  committed
  b4246c4

  Copy full SHA for b4246c4

• Renaming in MultiGPU example; set up unit testing

  jwallwork23

  committed
  28e3778

  Copy full SHA for 28e3778

• Raise error if no CUDA in example 3

  jwallwork23

  committed
  5fa7801

  Copy full SHA for 5fa7801

• Lint

  jwallwork23

  committed
  8c27dc1

  Copy full SHA for 8c27dc1

• Fix model filename passed to fortran

  jwallwork23

  committed
  68f111a

  Copy full SHA for 68f111a

• Do require mpi4py in Python script

  jwallwork23

  committed
  3f3cd53

  Copy full SHA for 3f3cd53

Commits on Jan 24, 2025

• Merge branch 'main' into add-devices-iccs

  jwallwork23

  committed
  781a69f

  Copy full SHA for 781a69f

• DO NOT MERGE - drop unit tests so we don't need to install pFUnit

  jwallwork23

  committed
  c86fe4a

  Copy full SHA for c86fe4a

• DO NOT MERGE: remove annoying compiler warnings

  TomMelt

  committed
  5c0f007

  Copy full SHA for 5c0f007

• chore: add device type to resnet example

  TomMelt

  committed
  3946312

  Copy full SHA for 3946312

• Fix devices in pt filenames

  jwallwork23

  committed
  765f79a

  Copy full SHA for 765f79a

• Merge remote-tracking branch 'origin/208_multi-gpu-build' into add-devices-iccs

  jwallwork23

  committed
  4504725

  Copy full SHA for 4504725

• Convert example 3 to XPU

  jwallwork23

  committed
  6034cb7

  Copy full SHA for 6034cb7

• DO NOT MERGE - turn off CI

  jwallwork23

  committed
  d5e094b

  Copy full SHA for d5e094b

• Lint

  jwallwork23

  committed
  1ba10f5

  Copy full SHA for 1ba10f5