You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am attempting to use the xfcvDataMover Tiler & Sticher IP to send image data to AI-engines. I linked the .xo files with my custom design via V++. On hardware, my AIE graph finishes execution however the tiler and stitcher instances hang indefinitely on the .wait() calls.
Have you tested the design in Hw_emu flow? Is it passing or hanging as above ?
If not can you please try to build and run it in hw_emu flow.
Here the data transfer will be from tiler->aie->stitcher. Tiler will send tiles of image data along with the tile metadata depending on the overlap info and aie computes the data and sends it to stitcher and stitcher will remove the overlapped data if any and also metadata. So if aie graph completes execution then atleast tiler should not be in hang state.
I am attempting to use the xfcvDataMover Tiler & Sticher IP to send image data to AI-engines. I linked the .xo files with my custom design via V++. On hardware, my AIE graph finishes execution however the tiler and stitcher instances hang indefinitely on the .wait() calls.
I am following this guide: https://xilinx.github.io/Vitis_Libraries/vision/2022.1/overview-aie.html
Additionally, I am using this design as a reference: https://github.com/Xilinx/Vitis_Libraries/tree/2022.1/vision/L2/tests/aie/Filter2D
The lack of documentation on these IPs have made it difficult to debug. Any insight/guidance to solving this issue?
The text was updated successfully, but these errors were encountered: