Blog
/blog3 posts
- We designed a decode chip for DeepSeek-V4.1-FlashAn open hardware study. Half of each decode step is wait time. We show what a chip does about it.
- We designed a core for Inkling 2 SmallAn open hardware study. We compare reloadable weights on the die with B300 GPUs.
- Four TPUs in one yearGoogle just showed where AI silicon is going.