What's the support state for Intel CPU with AMX and Nvidia GPU with tensor cores?

[hangfu-guo]

Hi, appreciate this benchmark tool!

Just wondering if the results utilize the latest AI acceleration techs in different products. Since I saw the FLOPS/OPS could be much lower than what we can see from the NV websites. (For examble, INT8 TOPS).

Can someone please clarify what the results are based on?

And i know barely nothing about OpenCL, but does the results of FLOPS/OPS align with the theoretical AI capabilities?