From 592296cd45f458f7d4e2c7a370e1c755376cde3b Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Tue, 4 Mar 2025 15:54:46 +0800 Subject: [PATCH] Add some plans --- README.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/README.md b/README.md index 16b87c5..1eb5e29 100644 --- a/README.md +++ b/README.md @@ -280,6 +280,13 @@ For two micro-batch overlapping, you can refer to the following figure. With our ![low-latency](figures/low-latency.png) +## Roadmap + +- [ ] A100 support (intranode only) +- [ ] Support BF16 for the low-latency dispatch kernel +- [ ] Support NVLink protocol for intranode low-latency kernels +- [ ] SM-free normal kernels + ## Notices #### Easier potential overall design