How to configure the OpenVINO EP to use all socketed CPU's in ONNX?

[SingerFling]

Using the standard inferencing option for ONNX all the cores across two socketed Xeon's are used.  However, when using the OpenVINO EP option only the cores on one of the Xeon's is used.  How can I configure the OpenVINO EP / ONNX to use all physically socketed CPU's?

This target environment is using the latest releases of ONNX and OpenVINO and C++ on Windows.

Thanks in advance for any insight.

[Wan_Intel]

Hi SingerFling,

Thanks for reaching out to us.

For your information, num_streams is set to the value of 1 by default, performance for latency is used during build time and it will be used for inference. You may set the num_streams with any unsigned positive number other than 0. For more information, please refer to the Summary of options in OpenVINO™ Execution Provider.

Regards,

Wan

[Wan_Intel]

Hi SingerFling,

Thanks for your question.

If you need additional information from Intel, please submit a new question as this thread will no longer be monitored.

Regards,

Wan