Guancheng Fu

From your `lscpu` output, we can get the following information: 1. `Hyper-threading` is enabled, which can do harm to performance for computation-intensive task 2. One socket has 20 physical cores,...

Hi, I am working on this and will try to reproduce the issue first. May I ask what environment are you using? Can you list the output of `lscpu` or...

The transformer version `4.34.1` works normally for `codellama` as mentioned in our [examples](https://github.com/intel-analytics/BigDL/tree/main/python/llm/example/CPU/HF-Transformers-AutoModels/Model/codellama#example-predict-tokens-using-generate-api). For transformer version `4.36`, @jenniew can you have a look on this and fix this issue, thanks...

Are you using the recommended environment variable settings? Try `bigdl-llm-init -t`

> > Are you using the recommended environment variable settings? Try `bigdl-llm-init -t` > > Yes, I installed bigdl llm according to the instructions in README. However, I am not...

> > > > Are you using the recommended environment variable settings? Try `bigdl-llm-init -t` > > > > > > > > > Yes, I installed bigdl llm according...

Can you attach the hardware config using `lscpu`? You can try `sudo numactl -C 0-31 python3 mysql_qa_bench.py`. The problem is possibly caused by memory socket bind.

> > Can you attach the hardware config using `lscpu`? > > You can try `sudo numactl -C 0-31 python3 mysql_qa_bench.py`. The problem is possibly caused by memory socket bind....

CC @glorysdj @Oscilloscope98

After some investigation, the cause of the problem is somehow wired. It seems that the implementation of the `native_sdp` will generate the wrong output. However, this `native_sdp` is roughly the...