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

Add some notes for flags to README #300

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Add some notes for flags to README #300

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f404fe0
Add some notes for flags to README
mingmingtasd Dec 18, 2024
8f90a33
address belem's comments
mingmingtasd Dec 20, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 8 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ WebNN Samples is currently hosted on GitHub Pages, with model files stored on Am

To get started with WebNN on Intel AI PCs you will need:
* Window 11, version 21H2 or newer
* It's recommended to install the latest [Intel® Arc™ & Iris® Xe Graphics](https://www.intel.com/content/www/us/en/download/785597/intel-arc-iris-xe-graphics-windows.html) on Windows for improved WebNN compatibility and performance
* It's recommended to install the latest [Intel® Arc™ & Iris® Xe Graphics on Windows](https://www.intel.com/content/www/us/en/download/785597/intel-arc-iris-xe-graphics-windows.html) for improved WebNN compatibility and performance

1. Download and install the latest [Chrome Canary](https://www.google.com/chrome/canary/) or [Edge Canary](https://www.microsoft.com/en-us/edge/download/insider?form=MA13FJ)
2. Navigate to `about://flags` in browser address bar
Expand All @@ -56,6 +56,11 @@ To get started with WebNN on Intel AI PCs you will need:
#### Running WebNN on CPU or GPU
1. Launch Chrome Canary or Edge Canary

* **Notes**:

1. Running WebNN on Qualcomm GPUs is disabled by default. You can use the `--disable_webnn_for_gpu=0` flag to bypass. However, the functionality is not guaranteed.
Copy link
Contributor

@ibelem ibelem Dec 20, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it's better for Qualcomm to add such notes, not from Intel. @mingmingtasd We don't know if Qualcomm guarantees the functionality or not at all.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I remember we want a README that includes descriptions of all the flags. I can remove the note for
disable_webnn_for_gpu for Qualcomm, then that means if people run on Qualcomm and see GPU blocked issue, it's not our responsibility to tell them: you should use --disable_webnn_for_gpu=0 flag. WDYT?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But agree to remove the sentence "However, the functionality is not guaranteed." for Qualcomm, we should not judge other vendors

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done by the new commit, PTAL, thanks! @ibelem

2. If you run WebNN on Intel GPUs with old driver versions between 31.0.101.5518 and 32.0.101.6083, the performance may not be good enough, because the DirectML execution of metacommands on these drivers is disabled by default due to some stability bugs. You can use `--disable_dml_meta_commands_for_gpu=0` to bypass. However, the functionality is not guaranteed.

#### Running WebNN on NPU
At present, the [image classification](https://webmachinelearning.github.io/webnn-samples/image_classification/) and [object detection](https://webmachinelearning.github.io/webnn-samples/object_detection/) samples support NPU.

Expand Down Expand Up @@ -84,7 +89,8 @@ At present, the [image classification](https://webmachinelearning.github.io/webn

* **Notes**:

Running WebNN on NPU requires a higher version of DirectML.dll than the one in the Windows system. Using the `--use-redist-dml` flag will allow Google Chrome Canary to load the downloaded DirectML.dll with a sufficiently high version.
1. Running WebNN on NPU requires a higher version of DirectML.dll than the one in the Windows system. Using the `--use-redist-dml` flag will allow Google Chrome Canary to load the downloaded DirectML.dll with a sufficiently high version.
2. Since the NPU drivers may be blocked by default if they are not the latest [Intel® Core™ Ultra NPU Driver on Windows](https://www.intel.com/content/www/us/en/download/794734/intel-npu-driver-windows.html), you can use the `--disable_webnn_for_npu=0` flag to bypass. However, the functionality is not guaranteed.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we really need However, the functionality is not guaranteed.? This is known to users by default before officially shipping WebNN.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I just intend to be responsible for these new added notes, to notice that if you follow these notes, you may meet functionality issues.

Agree that the However, the functionality is not guaranteed is negative, I am OK to remove them if you think that's really not necessary.


## Support and Feedback
If you encounter any issues or have feedback on the WebNN Samples, please open an issue on the repository. We appreciate your input and will strive to address any problems as quickly as possible.
Expand Down
Loading