[pypy-issue] Issue #2235: [Man bites dog] SSE2 backend slower in some benchmarks (pypy/pypy)

Pete Vine issues-reply at bitbucket.org
Sat Feb 6 13:42:04 EST 2016


New issue 2235: [Man bites dog] SSE2 backend slower in some benchmarks
https://bitbucket.org/pypy/pypy/issues/2235/man-bites-dog-sse2-backend-slower-in-some

Pete Vine:

Pitting the latest official `x86` Linux nightly against the old backend reveals a few surprising results: (just the ones SSE2 was `%` slower)

`Ai` 12
`bm_cham/bm_dulwich` 12
`go` 64
`json` 27
`pyflate` 12
`slowspitfire/spitfire` 35/38
`pidigits` 33 
 
It could even be called a regression as the official `4.0.1` release was doing much [better](https://bitbucket.org/pypy/pypy/issues/2197/sqlite-benchmark-fails-in-401#comment-25137937) the first I tried this comparison.




More information about the pypy-issue mailing list