Alan Zhai

> @xyegithub > https://discuss.pytorch.org/t/compiling-cpp-cuda-extension-composed-of-multiple-files/30685/4 > use #include in *.cu rather than replace '#include ' with '#include ' in lltm_cuda_kernel.cpp. It works. Thanks for your suggestions. ^^

@vpirogov Thanks. `vX.Y` works. It looks like an implicit website for customers. Would you be willing to add an explicit switch button? It should be very user-friendly.

@vpirogov That's great. Listen to and evaluate developers' suggestions. Polish your projects.

@mschilling0 any updates? It is from l0 or driver, I think.

@pratikbariintel Hi, the current initialization is `zeInit` https://github.com/intel/pti-gpu/blob/00e4bbc736a64811195b94243b83d30383309396/tools/sysmon/main.cc#L1252. I try to replace `zesDeviceGetProperties` with `zeDeviceGetProperties` and then `zeDeviceGetProperties` can get correct property. So the next question is why `zesDeviceGetProperties` and...

@pratikbariintel Got it. I mis-understand the api according the spec. ```cpp /// @brief Handle of device object typedef ze_device_handle_t zes_device_handle_t; ``` https://github.com/oneapi-src/level-zero/blob/3c938e21d827af014971d69dfd66759c2444e4d0/include/zes_api.h#L34C13-L34C48

`zesDeviceEnumEngineGroups` still cannot achieve correct engines on B580 via `zes_device_handle_t`. @AshwinKumarKulkarni @pratikbariintel

These infrastructure-related APIs should be ready before the new hardware is released. @pratikbariintel @AshwinKumarKulkarni

@shu1chen Is INT4 weight decompression feature supported on CPU platforms?

@shu1chen Thanks for your comments. The example works. But I still have some questions. 1. src & dst with `f32` data type will be supported for `u4` / `s4` decompression?...