I complied a lenet model for an nv_small and nv_large configuration with specifying the precision to be int8 without a problem. It created a fast-math.nvdla file. However when I execute the nvdla_runtime on the qemu platform it is able to start the emulator and generates about 10 or so log lines because crashing and returning to the usr/local/nvdla folder. I use opendla_2.ko module before running nvdla_runtime. There are no errors outputted before qemu crashes. Why is this happening? I'm able to run the nv_full fp16configuration to completion and get correct results at the output. It seems that nv_full int8 configuration also works fine. Any help would be greatly appreciated. Thank you!
Hello,
I complied a lenet model for an nv_small and nv_large configuration with specifying the precision to be int8 without a problem. It created a fast-math.nvdla file. However when I execute the nvdla_runtime on the qemu platform it is able to start the emulator and generates about 10 or so log lines because crashing and returning to the usr/local/nvdla folder. I use opendla_2.ko module before running nvdla_runtime. There are no errors outputted before qemu crashes. Why is this happening? I'm able to run the nv_full fp16configuration to completion and get correct results at the output. It seems that nv_full int8 configuration also works fine. Any help would be greatly appreciated. Thank you!