I verified that the NPU and XRT stack are healthy on this system. xrt-smi examine detects the Krackan NPU and xrt-smi validate --run latency passes successfully. However, the Whisper demo in RyzenAI-SW v1.7.0 clean checkout still fails after selecting VitisAIExecutionProvider with : no such option flexible-l2-buffer-layout. This suggests the issue is specific to the Whisper demo configuration or pass/toolchain integration, not the NPU runtime itself.
I verified that the NPU and XRT stack are healthy on this system. xrt-smi examine detects the Krackan NPU and xrt-smi validate --run latency passes successfully. However, the Whisper demo in RyzenAI-SW v1.7.0 clean checkout still fails after selecting VitisAIExecutionProvider with : no such option flexible-l2-buffer-layout. This suggests the issue is specific to the Whisper demo configuration or pass/toolchain integration, not the NPU runtime itself.