Meng, Hengyu

@aahouzi @sgwhat Can you try https://github.com/ggerganov/llama.cpp/pull/6006?

Hi @fakezeta UHD770 might be limited nether on performance or functionality, as issued in the [README](https://github.com/ggerganov/llama.cpp/blob/master/README-sycl.md#verified)

hi @AlexFierro9 sorry for the late. can you paste the output of ```source /opt/intel/oneapi/setvars.sh``` there should be MKL listed, otherwise you might not install MKL correctly ``` hengyume@mlp-618:~$ source /opt/intel/oneapi/setvars.sh...

https://stackoverflow.com/questions/58992376/no-module-named-cmake Hi @ei-grad can you check cmake as the link above?

hi @sungkim11 , you could refer to [gpu instruction](https://github.com/intel/intel-extension-for-transformers/blob/main/docs/weightonlyquant.md#prepare-dependency-packages) to enable WOQ on intel GPU

> Why is this asking for username/password: > > python -m pip install torch==2.1.2 -f https://developer.intel.com/ipex-whl-stable-xpu > > source /opt/intel/oneapi/setvars.sh > > git clone https://github.com/intel-innersource/frameworks.ai.pytorch.ipex-gpu.git ipex-gpu cd ipex-gpu git checkout...

> Error: > > UserWarning: Attempted to use ninja as the BuildExtension backend but we could not find ninja.. Falling back to using the slow distutils backend. warnings.warn(msg.format('we could not...

you should raise the PR to IPEX too to avoid a huge PR again

BTW why AMX will greatly improve next token latency?

sorry for late response, we just back from chinese national holiday :) since it is windows, I suppose you install the drivers all with Intel Arc Control, right? @characharm @easyfab