NVIDIA Developer Forums

Peak Performance INT1, INT4, INT8, INT16, INT32 for RTX3090 Tensorcore

Developer Tools CUDA Developer Tools

Daniel_Wong January 12, 2021, 1:22am 1

Hi,

is there any reference for the peak performance of INT1, INT4, INT8, INT16, INT32 for RTX3090 on Tensorcore?

Thanks!

Topic		Replies	Views	Activity
Why does Int8 quantization occupy more GPU graphics memory than float16, TensorRT quantization TensorRT	1	525	June 6, 2023
Maximum Performance of ResNet50 model for NVIDIA T4 in TensorRT using trtexec Deep Learning (Training & Inference)	1	439	October 5, 2020
RTX 3090 Peak Performance GPU-Accelerated Libraries cutensor	1	7485	December 14, 2021
How does INT4 work in the Tesla T4? GPU-Accelerated Libraries tensorrt	0	607	April 13, 2020
Question about tensor cores performance CUDA Programming and Performance	3	650	October 12, 2021
Minimum compute capability for INT8 optimization? TensorRT	2	729	January 14, 2021
TensorRT in INT4 precision mode TensorRT	1	1111	February 25, 2019
High inference time while running UNet with INT8 precision TensorRT tensorrt	5	986	February 10, 2021
How to choose size of tensor core? CUDA Programming and Performance	1	1298	October 6, 2023
Question about the tensorrt precision transformation TensorRT	4	470	July 12, 2021