Over 1k tok/s on an RTX 5090 with Qwen3 0.6B(blog.alpindale.net)
blog.alpindale.net
Over 1k tok/s on an RTX 5090 with Qwen3 0.6B
https://blog.alpindale.net/posts/5090_decode_optimization/
1 comments
GitHub seems to be down right now, but the kernel is here: https://github.com/AlpinDale/qwen_megakernel (1,008 tok/s now, a bit improved after I made the blogpost)