- Does TensorRT 1.0 support GTX 1080?
- Does Tensor Rt 1.0 have int8 support or does it have just FP32, FP16 support?
If it has only FP16 and FP32 support, then it makes sense why it’s not supported on GTX 1080.GTX 1080 does not have FP16 native support.
- I’m using this tool for inference. GTX 1080 has int8 support. I think one GTX 1080 should be sufficient for int8 inference.
- If tensorRT 1.0 does not support int8, then I should probably wait for TensorRT 2.0 approval.
Your comments will be helpful.