Code

28 add gpu embedded backend#30

Code

Open

lorenzovarese wants to merge 59 commits intomainGridTools/GridTools.jl:mainfrom 28-add-gpu-embedded-backendGridTools/GridTools.jl:28-add-gpu-embedded-backendCopy head branch name to clipboard

+1,641-150

Commits

Commits on Jul 18, 2024

• Fix mesh definitions in benchmarks

  lorenzovarese

  committed
  5c46d04

  Copy full SHA for 5c46d04

• Attempt to fix the benchmark PR

  lorenzovarese

  committed
  1eea839

  Copy full SHA for 1eea839

• Fix pip install gt4py

  lorenzovarese

  committed
  02ee8b7

  Copy full SHA for 02ee8b7

• Fix PyCall installation in the benchmark_pr.yml

  lorenzovarese

  committed
  2ae9b77

  Copy full SHA for 2ae9b77

• Fix the PyCall invoke

  lorenzovarese

  committed
  8686326

  Copy full SHA for 8686326

• Add reference to julia env in benchmark_pr config

  lorenzovarese

  committed
  e978b6c

  Copy full SHA for e978b6c

• Add cache for python, and fix the pycall (again :/)

  lorenzovarese

  committed
  ea7d32d

  Copy full SHA for ea7d32d

Commits on Jul 19, 2024

• Activate the env in the benchmark CI

  lorenzovarese

  committed
  3d12f85

  Copy full SHA for 3d12f85

• Include the Cell and K definitions in the benchmark

  lorenzovarese

  committed
  2ae25f2

  Copy full SHA for 2ae25f2

Commits on Jul 22, 2024

• Add readme to run benchmark example

  lorenzovarese

  committed
  5412e44

  Copy full SHA for 5412e44

Commits on Jul 24, 2024

• Add commented benchmark for field operations

  lorenzovarese

  committed
  978e6b5

  Copy full SHA for 978e6b5

• Merge remote-tracking branch 'origin' into 16-setup-benchmarking-infrastructure

  lorenzovarese

  committed
  49687ce

  Copy full SHA for 49687ce

Commits on Jul 26, 2024

• fix benchmarks

  lorenzovarese

  committed
  8046538

  Copy full SHA for 8046538

Commits on Jul 31, 2024

• Add benchmark comparison between Julia's broadcast addition and the field operator one

  lorenzovarese

  committed
  8fea0ca

  Copy full SHA for 8fea0ca

Commits on Aug 2, 2024

• Update the Benchmark suite. Provide comparison between broadcast on array, on fields data and with field operator.

  lorenzovarese

  committed
  8d0296b

  Copy full SHA for 8d0296b

• Improve naming and type checking

  lorenzovarese

  committed
  b9d4e2e

  Copy full SHA for b9d4e2e

Commits on Aug 13, 2024

• Add draft of neighbour_sum benchmark

  lorenzovarese

  committed
  2ab1421

  Copy full SHA for 2ab1421

• Add the benchmarks for sine and cosine field operators

  lorenzovarese

  committed
  3062bdf

  Copy full SHA for 3062bdf

Commits on Aug 14, 2024

• Add benchmarks for remapping

  lorenzovarese

  committed
  323c269

  Copy full SHA for 323c269

• Add draft mpdata

  lorenzovarese

  committed
  9c138c2

  Copy full SHA for 9c138c2

Commits on Aug 16, 2024

• Merge remote-tracking branch 'origin' into 16-setup-benchmarking-infrastructure

  lorenzovarese

  committed
  e3ba8ec

  Copy full SHA for e3ba8ec

• Clear benchmarks.jl and add remapping

  lorenzovarese

  committed
  700a545

  Copy full SHA for 700a545

• Add neighbor sum benchmark to the suite

  lorenzovarese

  committed
  3590280

  Copy full SHA for 3590280

• Fix dependencies in benchmarks

  lorenzovarese

  committed
  0be7ec1

  Copy full SHA for 0be7ec1

• Add verbose flag to avoid printing

  lorenzovarese

  committed
  62276b0

  Copy full SHA for 62276b0

• Quick fix to the unary/binary operation support

  lorenzovarese

  committed
  1bbb1e6

  Copy full SHA for 1bbb1e6

• Use the ExampleMeshes in Atlas miniapp (with workaround on the offset_provider)

  lorenzovarese

  committed
  17a55f8

  Copy full SHA for 17a55f8

• Add benchmark for mp_data

  lorenzovarese

  committed
  fcb44cd

  Copy full SHA for fcb44cd

Commits on Aug 19, 2024

• Fix slicing operation in the advection_miniapp

  lorenzovarese

  committed
  22d1257

  Copy full SHA for 22d1257

• Fix benchmarking size

  lorenzovarese

  committed
  30d92c1

  Copy full SHA for 30d92c1

• Remove deprecated benchmark files

  lorenzovarese

  committed
  9928e14

  Copy full SHA for 9928e14

• Fix advection benchmarks and place them in a separate script

  lorenzovarese

  committed
  177f8ba

  Copy full SHA for 177f8ba

Commits on Aug 20, 2024

• Fix K dimension in advection meshes

  lorenzovarese

  committed
  391dc0a

  Copy full SHA for 391dc0a

• Add multi-threads optimization on broadcasting operation

  lorenzovarese

  committed
  9048666

  Copy full SHA for 9048666

• Change benchmark SUITE for compatibility with AirSpeedVelocity

  lorenzovarese

  committed
  e2ce601

  Copy full SHA for e2ce601

• Add multi-threads optimization

  lorenzovarese

  committed
  45cf97a

  Copy full SHA for 45cf97a

• Merge branch '16-setup-benchmarking-infrastructure' of https://github.com/GridTools/GridTools.jl into 16-setup-benchmarking-infrastructure

  lorenzovarese

  committed
  6fb4870

  Copy full SHA for 6fb4870

• Restoring SIMD loop in broadcast

  lorenzovarese

  committed
  8b8a68f

  Copy full SHA for 8b8a68f

• Add benchmark readme on how to run benchmarks on separate revisions

  lorenzovarese

  committed
  be385b7

  Copy full SHA for be385b7

• Create an AtlasMeshes module and resolve issues with atlas4py import

  lorenzovarese

  committed
  b9ebf8e

  Copy full SHA for b9ebf8e

• Fix embedded test with the new K dimension definition in example meshes

  lorenzovarese

  committed
  085877d

  Copy full SHA for 085877d

• Separate the simulation loop from the Advection Setup of the miniapp

  lorenzovarese

  committed
  89572e1

  Copy full SHA for 89572e1

• Small changes in benchmark documentation

  lorenzovarese

  committed
  4d71e0b

  Copy full SHA for 4d71e0b

• Fix the names retrieval of the modules automatically generated by AirSpeedVelocity when running the advection benchmark

  lorenzovarese

  committed
  26fe900

  Copy full SHA for 26fe900

Commits on Aug 21, 2024

• Ignore plot files by AirSpeedVelocity

  lorenzovarese

  committed
  7cac41c

  Copy full SHA for 7cac41c

• Add Polyester to the dependencies

  lorenzovarese

  committed
  6cb5585

  Copy full SHA for 6cb5585

Commits on Aug 22, 2024

• Increase the size of the Atlas Mesh for benchmarking purposes

  lorenzovarese

  committed
  96f416f

  Copy full SHA for 96f416f

• Add script to automate the benchmark comparison between the last two commits

  lorenzovarese

  committed
  426f936

  Copy full SHA for 426f936

• Add utilis for benchmark/profiling in the interactive REPL

  lorenzovarese

  committed
  c8a08bb

  Copy full SHA for c8a08bb

• Move autorun in the utils folder

  lorenzovarese

  committed
  d964221

  Copy full SHA for d964221

• Update autorun script

  lorenzovarese

  committed
  182dd6d

  Copy full SHA for 182dd6d

Commits on Aug 23, 2024

• Fix the autorun script to use hashes instead of tags

  lorenzovarese

  committed
  b1f539e

  Copy full SHA for b1f539e

Commits on Sep 4, 2024

• Add gpu backend support for basic broadcast operation

  lorenzovarese

  committed
  5b0f1dc

  Copy full SHA for 5b0f1dc

• Merge remote-tracking branch 'origin/16-setup-benchmarking-infrastructure' into 28-add-gpu-embedded-backend

  lorenzovarese

  committed
  85f36fe

  Copy full SHA for 85f36fe

• Add benchmarking suite for gpu

  lorenzovarese

  committed
  5f6d164

  Copy full SHA for 5f6d164

Commits on Sep 5, 2024

• Fix gpu benchmark for memory bandwidth computation

  lorenzovarese

  committed
  47dbe38

  Copy full SHA for 47dbe38

• Improve the printing of the gpu benchmark results

  lorenzovarese

  committed
  b1fe0b2

  Copy full SHA for b1fe0b2

Commits on Sep 13, 2024

• Merge remote-tracking branch 'origin' into 28-add-gpu-embedded-backend

  lorenzovarese

  committed
  8b2e6f5

  Copy full SHA for 8b2e6f5

• Merge remote-tracking branch 'origin' into 28-add-gpu-embedded-backend

  lorenzovarese

  committed
  063478a

  Copy full SHA for 063478a