This chip we designed (with my mentee Taehyunh Kim (https://github.com/thkim2031)) from scratch, following digital designs macros are harden in it:
- CLA
- BFLOAT-FMA (Bfloat-Multiplication, Single Precision Accmulation)
- 2x2 Output Stationary Systolic Array Based ML Accelerator
- 8 bit Signed Booth-Radix8 Multiplier