How can I enable openblas?

[newTomas]

When using a long prompt, processing takes a very long time, openblas helps a lot and does not require a video card. But how to enable it in llama-node?

[hlhr202]

i m currently working on cublas. for openblas probably will try it later

[newTomas]

Are they hard to add at all? Maybe I can help?

[hlhr202]

you can try modifying build.rs for llama-sys, which uses cmake flags to build features for different architectures.

[hlhr202]

emm, you can use "feature/cuda" branch if you wish to pr for this. I v changed a lot build script there.