Ahmed Saber

overlay bin for U50

2

I can't find the overlay bin for U50 card under [https://github.com/Xilinx/ml-suite/tree/v1.5/overlaybins](url) however the deployment/dev shells are available from [https://www.xilinx.com/products/boards-and-kits/alveo/u50.html#gettingStarted](url)

I wasn't able to get the git commit for the same XRT version provided prepackaged from [https://www.xilinx.com/products/acceleration-solutions/xilinx-machine-learning-suite.html#gettingStartedU250](https://www.xilinx.com/products/acceleration-solutions/xilinx-machine-learning-suite.html#gettingStartedU250) (2.1.127). And this was an issue as the this version isn't compatible with...

I used the python Xdnn compiler to compile googlenet_v1 caffemodel to generated fpga version but when I benchmark the the generated one I only get 2793 FPS max throughput vs....

apps/yolo/run.sh uses non-existent script ([https://github.com/Xilinx/ml-suite/blob/v1.5/apps/yolo/run.sh#L457](https://github.com/Xilinx/ml-suite/blob/v1.5/apps/yolo/run.sh#L457)) (this doesn't exist in both docker and git repo)

some code/tools aren't available on the github repo (e.g. xfdnn/tools/emu and custom-built caffe/tensorflow under /opt) and only exists in the docker container.

the overlaybins provided from Xilinx [https://www.xilinx.com/products/acceleration-solutions/xilinx-machine-learning-suite.html](https://www.xilinx.com/products/acceleration-solutions/xilinx-machine-learning-suite.html) don't match the prepackaged ones in the docker container. I had to copy the ones inside in the container to get the correct result...

Running the caffe jupyter notebook goes through fine but the python process keeps running in the background communicating with the card. This locks the device and doesn't allow other examples...

if Invoice Rate is 0, it becomes empty in the XML

1

If I want to set the Rate in InvoiceLineAdd = 0, the generated XML request has the node empty which isn't the desired value as QuickBooks will ignore it and...