[lkml]   [1999]   [Apr]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: bad lmbench numbers for mmap

On Fri, 16 Apr 1999, David Miller wrote:
> It may be that in a previous version of lat_mmap it did only read
> taps, but Larry for one reason or another changed those into writes.

I know there was a version that didn't any accesses at all. And I
_thought_ the new one did reads, but I was obviously wrong. I can't
imagine why it would ever have been changed - I don't have the old emails
where I discussed the original problem with Larry saved..

> I do agree that doing writes is probably not making the benchmark
> report what the name would indicate. However I do see that a version
> that does writes is useful (think of a linker implementation that
> mmap's the final object file it creates, or some mmap'd database of
> some sort). But such a benchmark should have a different name
> (lat_mmapwrite, or somesuch).

Indeed. The benchmark is potentially useful (and we _will_ perform well on
it eventually), it's just that it's not nearly as useful as the obvious
"read only" version that apparently doesn't exist at all.

However, it doesn't appear that Larry will change this any time soon, for
completely unrelated reasons ;)


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:51    [W:0.058 / U:3.772 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site