Skip to content

Conversation

@ax3l
Copy link
Member

@ax3l ax3l commented Mar 20, 2025

amrex.the_arena_init_size=0 is default on CPU but not GPU. In order to run tests in parallel on the same GPU, we need to change the default so the first test does not allocate 3/4th of the memory of the GPU (to speed up later allocations in a pre-allocated heap).

`amrex.the_arena_init_size=0` is default on CPU but not GPU.
In order to run tests in parallel on the same GPU, we need
to change the default so the first test does not allocate
3/4th of the memory of the GPU (to speed up later allocations
in a pre-allocated heap).
@ax3l ax3l added backend: cuda Specific to CUDA execution (GPUs) component: tests tests and examples backend: sycl Specific to DPC++/SYCL execution (CPUs/GPUs) backend: hip Specific to ROCm execution (GPUs) labels Mar 20, 2025
@EZoni EZoni mentioned this pull request May 6, 2025
5 tasks
@EZoni
Copy link
Member

EZoni commented May 6, 2025

Merging development to rerun CI after the fixes in #436 and #441.

Companion PR in WarpX for reference: BLAST-WarpX/warpx#5785.

@EZoni EZoni enabled auto-merge (squash) May 6, 2025 01:03
@EZoni EZoni self-assigned this May 6, 2025
@EZoni EZoni merged commit 1e212aa into AMReX-Codes:development May 6, 2025
17 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

backend: cuda Specific to CUDA execution (GPUs) backend: hip Specific to ROCm execution (GPUs) backend: sycl Specific to DPC++/SYCL execution (CPUs/GPUs) component: tests tests and examples

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants