Hello!
Which is the stable version of MESA which supports the NPU in imx8mp using the teflon delegate
Hi, I’m the Toradex AI Assistant.
I might make mistakes or occasionally get you wrong. Our support team will step in if needed.
MESA and NPU Support on i.MX8M Plus
I don’t have specific information about MESA versions and their support for NPU acceleration with the VX Delegate (TensorFlow Lite delegate for Vivante NPU) on i.MX8M Plus in my available sources.
However, I can provide some context about NPU acceleration on i.MX8M Plus:
NPU Acceleration on i.MX8M Plus
The i.MX8M Plus includes a dedicated Neural Processing Unit (NPU) that can be leveraged for machine learning inference through:
- VX Delegate: TensorFlow Lite delegate that provides hardware acceleration via the Vivante NPU
- TIM-VX: Low-level interface for NPU operations
- DeepViewRT: NXP’s inference engine
Recommendations
For the most accurate information about MESA version compatibility with NPU acceleration on i.MX8M Plus using Torizon, I’d suggest checking:
- The machine learning documentation in the Torizon documentation
- Release notes for specific Torizon versions that include ML acceleration support
- Hardware-specific documentation for i.MX8M Plus NPU integration
Would you like me to provide a general answer based on my knowledge about NPU acceleration frameworks, or do you have access to specific Torizon documentation that mentions MESA and NPU integration?
Best regards, Toradex AI Assistant.
Did you like my answer? 
If I answered your question, please tick the “Solution” checkbox.
For follow-up questions to me, please use @ToradexAI in your post.
Hi @sandyspark!
AFAIK, we didn’t test MESA to determine which version supports the Verdin iMX8M Plus NPU, and I’m afraid we won’t be very helpful in this matter.
That being said, I can recommend checking our Partner Network. There you’ll find people who have the necessary expertise to help you with your issue.
Best regards.