On Thu, 25 Jun 2026 02:57:34 GMT, Shawn Emery <[email protected]> wrote:
> I ran ML-KEM and ML-DSA benchmarks on a Macbook Pro M3 36GB Memory, see > attached. Everything is improved with the set of changes except for > ML-KEM-512 (encapsulation:-0.92%/decapsulation:-2.25%) and ML-DSA-44 (key > generation:-1.03%). > [8384353BenchmarksM3Pro.pdf](https://github.com/user-attachments/files/29319713/8384353BenchmarksM3Pro.pdf) Reran the specific benchmarks that showed regression on a quieter system (MacbookAir M3 8GB Memory) and averaged the results, which showed negligible differences of the afflicted benchmarks as follows: ML-KEM-512 (encapsulation:-0.38%/decapsulation:-0.21%) and ML-DSA-44 (key generation: -0.18%). In other words, the current fix resolves the regression in performance by either negligible differences or better performance, where the high observed +6.55% (ML-KEM-768, which would be one of the algorithms that would gain the most from this delta). ------------- PR Comment: https://git.openjdk.org/jdk/pull/31648#issuecomment-4796329690
