Add MPS inference support#131
Open
matthewijordan wants to merge 1 commit intofacebookresearch:mainfrom
Open
Conversation
meta-cla
bot
added
the
CLA Signed
- Add device utilities (get_device, get_amp_dtype, get_autocast_device_type, is_memory_query_supported) - Enable MPS device detection and device-agnostic autocast/AMP - Update all inference scripts to use auto device selection (CUDA > MPS > CPU) - Update Gradio app and data processing scripts for MPS compatibility - Add memory-safe cache clearing for CUDA-only operations
matthewijordan
force-pushed
the
feature/mps-inference-support
branch
from
2c5e824 to
2ca6e15
Compare
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Add MPS (Metal Performance Shaders) inference support for Apple Silicon Macs.
Changes
Device utilities (
mapanything/utils/device.py):get_device(): Auto-detects best available device (CUDA > MPS > CPU)get_device_capabilities(): Query device-specific capabilities (bf16 support)get_amp_dtype(): Device-aware mixed precision dtype selectionget_autocast_device_type(): Device type string for torch.autocastis_memory_query_supported(): Check if memory operations like empty_cache() are supportedempty_cache(): GPU cache clearing with device checkInference updates:
model.py&inference.py: Use device-agnostic autocast/AMPget_device()for auto device selectionConfig updates:
moge/default.yaml: Device set to 'auto' for automatic detectionTesting