Re: Interesting pentium-memcpy results

David C Niemi (niemi@wauug.erols.com)
Tue, 29 Jul 1997 11:15:00 -0400 (EDT)

>I just compared 2.1.46 vs. 2.1.46+pentium memcpy patch, and interestingly
>enough found that the UNIX byte benchmarks tended to _drop_ a fair bit,
>with the exception of process creation and execl throughput. (Note that I
>only ran the basic 'system' tests - TCP bandwidth etc. to be determined
>when I find the newer benchmarks)
...

There are a great number of flaws in the original Byte Unix benchmarks,
such that many of those tests really aren't measuring anything meaningful.
Would you mind rerunning your comparison under the fixed version and
reporting back on which tests are noticeably different? I'd recommend
running them in single-user mode for maximum reproducibility.

ftp://wauug.erols.com/pub/people/david-niemi/bench/unixbench-4.0.1.tgz

The "newer" Byte benchmarks are very Windows-centric and were quite a pain
in the neck to compile under Unix, last time I tried them (about 2 years
ago).

Thanks

David
Niemi@erols.com 703-810-5538 Reston, Virginia, USA
--- Most operating systems, sometimes even DOS, separate different
--- types of files into different directories. The Windows philosophy:
--- Throw everything into C:\WINDOWS and let God sort it out.