Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

28 add gpu embedded backend #30

Open
wants to merge 59 commits into
base: main
Choose a base branch
from
Open

28 add gpu embedded backend #30

wants to merge 59 commits into from

Conversation

lorenzovarese
Copy link
Collaborator

@lorenzovarese lorenzovarese commented Sep 4, 2024 •
edited by tehrengruber
Loading

#28

  • Add support to fields with CuArray
  • Specialize Custom Broadcast for GPU
  • Add small benchmark suite for gpu (with memory bandwidth)

Relevant notes:
I have merged the changes related to the 16-setup-benchmark branch to ensure compatibility with the other changes done while I was working on the benchmarks for CPU.
The actual code that add support to GPU could be found at this specific commit.

Note: This PR depends on #16

lorenzovarese and others added 30 commits July 18, 2024 16:05
@lorenzovarese
Fix mesh definitions in benchmarks
5c46d04
@lorenzovarese
Attempt to fix the benchmark PR
1eea839
@lorenzovarese
Fix pip install gt4py
02ee8b7
@lorenzovarese
Fix PyCall installation in the benchmark_pr.yml
2ae9b77
@lorenzovarese
Fix the PyCall invoke
8686326
@lorenzovarese
Add reference to julia env in benchmark_pr config
e978b6c
@lorenzovarese
Add cache for python, and fix the pycall (again :/)
ea7d32d
@lorenzovarese
Activate the env in the benchmark CI
3d12f85
@lorenzovarese
Include the Cell and K definitions in the benchmark
2ae25f2
@lorenzovarese
Add readme to run benchmark example
5412e44
@lorenzovarese
Add commented benchmark for field operations
978e6b5
@lorenzovarese
Merge remote-tracking branch 'origin' into 16-setup-benchmarking-infr…
49687ce 
…astructure
@lorenzovarese
fix benchmarks
8046538
@lorenzovarese
Add benchmark comparison between Julia's broadcast addition and the f…
8fea0ca 
…ield operator one
@lorenzovarese
Update the Benchmark suite. Provide comparison between broadcast on a…
8d0296b 
…rray, on fields data and with field operator.
@lorenzovarese
Improve naming and type checking
b9d4e2e
@lorenzovarese
Add draft of neighbour_sum benchmark
2ab1421
@lorenzovarese
Add the benchmarks for sine and cosine field operators
3062bdf
@lorenzovarese
Add benchmarks for remapping
323c269
@lorenzovarese
Add draft mpdata
9c138c2
@lorenzovarese
Merge remote-tracking branch 'origin' into 16-setup-benchmarking-infr…
e3ba8ec 
…astructure
@lorenzovarese
Clear benchmarks.jl and add remapping
700a545
@lorenzovarese
Add neighbor sum benchmark to the suite
3590280
@lorenzovarese
Fix dependencies in benchmarks
0be7ec1
@lorenzovarese
Add verbose flag to avoid printing
62276b0
@lorenzovarese
Quick fix to the unary/binary operation support
1bbb1e6
@lorenzovarese
Use the ExampleMeshes in Atlas miniapp (with workaround on the offset…
17a55f8 
…_provider)
@lorenzovarese
Add benchmark for mp_data
fcb44cd
@lorenzovarese
Fix slicing operation in the advection_miniapp
22d1257
@lorenzovarese
Fix benchmarking size
30d92c1
@lorenzovarese
Add multi-threads optimization on broadcasting operation
9048666
@lorenzovarese
Change benchmark SUITE for compatibility with AirSpeedVelocity
e2ce601
@lorenzovarese
Add multi-threads optimization
45cf97a
@lorenzovarese
Merge branch '16-setup-benchmarking-infrastructure' of https://github…
6fb4870 
….com/GridTools/GridTools.jl into 16-setup-benchmarking-infrastructure
@lorenzovarese
Restoring SIMD loop in broadcast
8b8a68f
@lorenzovarese
Add benchmark readme on how to run benchmarks on separate revisions
be385b7
@lorenzovarese
Create an AtlasMeshes module and resolve issues with atlas4py import
b9ebf8e
@lorenzovarese
Fix embedded test with the new K dimension definition in example meshes
085877d
@lorenzovarese
Separate the simulation loop from the Advection Setup of the miniapp
89572e1
@lorenzovarese
Small changes in benchmark documentation
4d71e0b
@lorenzovarese
Fix the names retrieval of the modules automatically generated by Air…
26fe900 
…SpeedVelocity when running the advection benchmark
@lorenzovarese
Ignore plot files by AirSpeedVelocity
7cac41c
@lorenzovarese
Add Polyester to the dependencies
6cb5585
@lorenzovarese
Increase the size of the Atlas Mesh for benchmarking purposes
96f416f
@lorenzovarese
Add script to automate the benchmark comparison between the last two …
426f936 
…commits
@lorenzovarese
Add utilis for benchmark/profiling in the interactive REPL
c8a08bb
@lorenzovarese
Move autorun in the utils folder
d964221
@lorenzovarese
Update autorun script
182dd6d
@lorenzovarese
Fix the autorun script to use hashes instead of tags
b1f539e
@lorenzovarese
Add gpu backend support for basic broadcast operation
5b0f1dc
@lorenzovarese
Merge remote-tracking branch 'origin/16-setup-benchmarking-infrastruc…
85f36fe 
…ture' into 28-add-gpu-embedded-backend
@lorenzovarese lorenzovarese added the enhancement New feature or request label Sep 4, 2024
@lorenzovarese lorenzovarese self-assigned this Sep 4, 2024
@lorenzovarese lorenzovarese linked an issue Sep 4, 2024 that may be closed by this pull request
Add gpu embedded backend #28
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tehrengruber tehrengruber Awaiting requested review from tehrengruber

Assignees

@lorenzovarese lorenzovarese

Labels
enhancement New feature or request
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Add gpu embedded backend
1 participant
@lorenzovarese