mod_mmap_static, top_fuel.patch?

[Dan Kegel <da...@alumni.caltech.edu>]

Just looked at the mod_mmap_static source and
http://www.arctic.org/~dgaudet/apache/1.3/top_fuel.patch
(mentioned in http://www.apache.org/docs/misc/perf-tuning.html).
Top_feul claims that using them together can reduce the
number of syscalls per request from 18 to 9, which sounds
pretty good.

Is top_fuel.patch still applicable to the current 1.3.6?

How big a win is patching with top_fuel.patch and
using mod_mmap_static on the full document tree in SPECWeb96?

Would this have made any difference in Mindcraft's testing?
- Dan

Re: mod_mmap_static, top_fuel.patch?

[Dean Gaudet <dg...@arctic.org>]

reducing those syscalls really doesn't gain much. 

I'm sure someone could bring the top_fuel patch forward if they wanted it. 
I haven't bothered to keep it up to date. 

Using mod_mmap_static is more interesting. 

No it wouldn't have made much of a difference in mindcraft's testing --
the system still would have choked scheduling after some point.

Dean

On Sat, 8 May 1999, Dan Kegel wrote:

> Just looked at the mod_mmap_static source and
> http://www.arctic.org/~dgaudet/apache/1.3/top_fuel.patch
> (mentioned in http://www.apache.org/docs/misc/perf-tuning.html).
> Top_feul claims that using them together can reduce the
> number of syscalls per request from 18 to 9, which sounds
> pretty good.
> 
> Is top_fuel.patch still applicable to the current 1.3.6?
> 
> How big a win is patching with top_fuel.patch and
> using mod_mmap_static on the full document tree in SPECWeb96?
> 
> Would this have made any difference in Mindcraft's testing?
> - Dan
>