We want to do TensorRT int8 inference.
According to documentation, ZOTAC GAMING GeForce RTX 3070 Twin Edge has Tensor Core, but information about int8 inference is still missing:
We want to know, does RTX 3070 support int8 inference? What is the int8 computation power in TOPS?
According to:
The FP32 (float) performance is 20.31 TFLOPS.
How about FP16/int8 performance?
Thanks for your reply.
We want to buy 3070 for int8 inference. But we don’t know if 3070 support int8.
My question is:
1.does 3070 support int8 inference?
2.If it does, what is the computation power in TOPs?