[BUG FIX][FEATURE] Rework backend & device selection logic #568
+39
−27
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Fixes Vulkan on non-cuda GPUs & adds Pytorch 2.5+ Intel XPU support Fixes Genesis-Embodied-AI#274 Merges the logic of get_gpu_device() and get_cpu_device() to handle device selection in a single place. If a specific backend is requested, this request is honored. The vulkan backend will use intel's pytorch xps device if available, reducing cross-device data transfer. The `gpu` backend will - default to `cuda` when available, - resolve to `mps` on macOS - or fallback to `vulkan`.
bxtbold
reviewed
Jan 15, 2025
Switched from match / case to if / else syntax
bxtbold
approved these changes
Jan 17, 2025
|
@zswang666 could you review and approve this if all look good? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #274 - Vulkan backend on non-cuda GPUs
Resolves #567 - Adds Pytorch 2.5+ Intel XPU support
Merges the logic of get_gpu_device() and get_cpu_device() to handle device selection in a single place.
If a specific backend is requested, this request is honored.
The vulkan backend will use intel's pytorch xps device if available, reducing cross-device data transfer.
The
gpubackend willcudawhen available,mpson macOSvulkan.