
Runs from yesterday aggregate decode hit 40.9/53.1/ 64.6 tok/s for C1-4. Warm TTFT measures 0.353 s at 8K and 0.361 s at 20K. KV pool scales to ~2.53M tokens FP4-local support is the best around too. Must try @jmacftw :)
You can find our implementation on @AtlasInference and all the test code and receipts here. github.com/enntity/sparkglm?…








