3 ms·

Well that sounds encouraging, Intel QuickSync has made efficient video encoding really accessible. Would love something similar for LLM inference Is this paper

Well that sounds encouraging, Intel QuickSync has made efficient video encoding really accessible. Would love something similar for LLM inference Is this paper the same work as goes in to Intel's BigDL-LLM code? That's been out for a few months now but I haven't seen it in use yet. https://medium.com/intel-tech/bigdl-llm-easily-optimize-your-large-language-model-on-intel-arc-gpus-5ba33a06729e https://medium.com/intel-tech/bigdl-llm-easily-optimize-your...

brucethemoose2 3y ago
> I haven't seen it in use yet Neither have I, but this is interesting and I am bookmarking it, thanks. The Gen AI space is littered with very interesting high performance runtimes that precisely no one uses. There are too many for integrators to even keep up with, much less integrate!