Support of X86 quantization engine as in pytorch?

[yanghou2000]

According to this pytorch quantization guide, pytorch now supports a faster backend for int8 called X86 as a replace for FBGEMM. I find that in the tch-rs jit-quantized example tch-rs only supports either FBGEMM or QNNPACK.

I wonder if X86 will be supported soon, or is there a workaround so that tchrs can use X86 backend?