introduce "--webnn-ort-ov-gpu-precision" to specify gpu precision #162
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
huningxin
approved these changes
Feb 26, 2025
| // the flag to use FP32 or ACCURACY inference precision to get better accuracy, | ||
| // but it may result in decreased performance. | ||
| // Usage1: --webnn-ort-ov-gpu-precision=FP32 | ||
| // Usage2: --webnn-ort-ov-gpu-precision=ACCURACY |
There was a problem hiding this comment.
You may want to explain what the underlying setting to OV execution mode when specifying "FP32" or "ACCURACY". Adding a pointer to OV documentation would be helpful.
| @@ -30,6 +30,18 @@ namespace webnn::ort { | |||
|
|
|||
| namespace { | |||
|
|
|||
| // These keys and values must align with the implementation of the ORT OpenVINO | |||
There was a problem hiding this comment.
Could you please point out the valid keys and values available in OV EP?
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Outdated suggestions cannot be applied.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.
Change the
--webnn-ort-use-ov-gpu-fp32flag to--webnn-ort-ov-gpu-precisionYou can specify the flag to use
FP32orACCURACYinference precision:Usage1:
--webnn-ort-ov-gpu-precision=FP32Usage2:
--webnn-ort-ov-gpu-precision=ACCURACY--webnn-ort-ov-gpu-precision=ACCURACYwill let OV EP GPU use ACCURACY execution mode. As my observation, this mode will not decrease the SD Turbo's inference time a lot, but can increase the pass rate of many ops' wpt tests (resolve the accuracy failure). We can test more with this flag.see https://github.com/microsoft/onnxruntime/blob/main/onnxruntime/core/providers/openvino/backends/basic_backend.cc#L167
PTAL, thanks!
cc/ @BruceDai @lisa0314 @miaobin