...bits virtual power management: ts fid vid ttp Benchmarking current code from the shootout with bounds checking...SpectralNorm N-Body OCaml 14.353s 9.469s F# 9.374s 6.933s The following compiler options...: x86_64 GNU/Linux 2.6.18 SMP F#: Windows XP Pro 2002 SP2 You're...had been assuming that by "complete" you meant "benchmarked on every permutation of setup" which is...

... And next time anybody submits benchmarking numbers here, I'd kindly...are given in table 1"), (f) the analysis ("applying the transformations...and seldom is, even "professional" benchmarking). Regards -- Markus Perhaps this part...and accurate details of your benchmark testing methodology, the test scripts...m not so fixated on performance anyway. (I think, though, OCaml...

... are given in table 1"), (f) the analysis ("applying the transformations ... blinding flash of insight that benchmarking should be treated as every...(and seldom is, even "professional" benchmarking). Regards -- Markus Perhaps this part...and accurate details of your benchmark testing methodology, the test scripts...complete source code for the benchmark suite/harness that is developed...

... the thread to find out what the current state of affairs is and patch the together with discriptions and measurements from earlier posts? You're only pretending to be dense now, don't you? Really, really, disappointing that is. I'm not so much interested in performance measurements at the moment that I need to play those games. Sorry. Just ignore my request. Regards -- Markus

... when you wrote "these are bast-case results for OCaml and C++" what you meant was that these are the best-case results when compiler options which would improve performance significantly are ignored? In OCaml, you should remove bounds checks explicitly at specific points rather than globally altering the language semantics via a compiler flag. So why do you ...

... have shown. The original statement gave measurements showing F# spectralnorm was 37%% faster than ocamlopt spectralnorm, and now we ... programs and put effort into optimizing the OCaml but not the F#, yes. It seems like you've told us the important ...and ocamlopt compilers would be a much stronger basis for F# advocacy. The -unsafe option alters the semantics of the language ...

...when you wrote "these are bast-case results for OCaml and C++" what you meant was that these are the best-case results when compiler options which would improve performance significantly are ignored? In OCaml, you should remove bounds checks explicitly at specific points rather than globally altering the language semantics via a compiler flag. So why do you suppose that ...

... timing measurements That will always be true. You could improve performance further by tweaking the Linux kernel. Yes it will always be ..., as we have shown. The original statement gave measurements showing F# spectralnorm was 37%% faster than ocamlopt spectralnorm, and now we know ... the best from the g++ and ocamlopt compilers would be a much stronger basis for F# advocacy.

... and say the 4.5%% difference between F# and ocamlopt spectral-norm is "well ...as "best case" when we can improve performance so significantly. My original assumption was that ... results when compiler options which would improve performance significantly are ignored? In OCaml, you ...slippery slope that most of the shootout benchmarks are trivially reducible. For example, just ...

...it even helped. And next time anybody submits benchmarking numbers here, I'd kindly ask to see ... the raw results are given in table 1"), (f) the analysis ("applying the transformations (I) and...suddenly had this blinding flash of insight that benchmarking should be treated as every other piece of labwork (and seldom is, even "professional" benchmarking). Regards -- Markus