Berkeley HardFloat
productionParameterized IEEE 754 floating-point units (FMA, conversions) in Chisel.
Install with
lfpga
lfpga add berkeley-hardfloat
Indexed from a public source and curated by LibFPGA. Own this project? Claim it by adding a manifest file to your repo.
Verification
We have not run this core through the LibFPGA toolchain yet, so it carries no earned badges. Many of these projects have strong verification of their own. Claim the listing to run our checks.
[](https://libfpga.com/cores/berkeley-hardfloat)
- 385 GitHub stars
About
The reference open-source floating-point library from UC Berkeley, written in Chisel. It provides parameterized fused multiply-add, add/multiply, and float/integer conversion blocks built around an internal recoded FP format, and is reused in RISC-V cores such as Rocket Chip and BOOM.