github-actions[bot]
commented
2 months ago Benchmarks summary
Performance benchmarks
| aha-mont64 | crc32 | minver | nettle-sha256 | nsichneu | slre | statemate | ud |
|---|---|---|---|---|---|---|---|
| 🔺 0.406 (+0.002) | 🔺 0.463 (+0.006) | 🔺 0.312 (+0.001) | 🔺 0.646 (+0.002) | 🔺 0.345 (+0.000) | 🔺 0.258 (+0.002) | 🔺 0.307 (+0.003) | 🔺 0.416 (+0.018) |
You can view all the metrics here.
Synthesis benchmarks (basic)
| Device utilisation: (ECP5) | LUTs used as DFF: (ECP5) | LUTs used as carry: (ECP5) | LUTs used as ram: (ECP5) | Max clock frequency (Fmax) |
|---|---|---|---|---|
| 🔻 21017 (-1584) | 🔺 5980 (+420) | 🔻 798 (-4) | 🔻 536 (-468) | 🔺 51 (+3) |
Synthesis benchmarks (full)
| Device utilisation: (ECP5) | LUTs used as DFF: (ECP5) | LUTs used as carry: (ECP5) | LUTs used as ram: (ECP5) | Max clock frequency (Fmax) |
|---|---|---|---|---|
| 🔻 31713 (-1938) | 🔺 9248 (+446) | 🔺 1970 (+32) | 🔻 716 (-468) | 🔺 43 (+1) |
tilk
This PR changes some 2-FIFOs (mostly in the scheduler) into
Pipes. Pipes have more combinational connections than FIFOs, but use less resources. The purpose is to validate the usefulness of pipes for gluing the core together.TODO: replace more 2-FIFOs with pipes and check benchmarks again.