httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dean Gaudet <>
Subject Re: mmap() versus read() (fwd)
Date Fri, 06 Mar 1998 09:56:46 GMT

Everything runs on linux 2.0.x really well until you get a working set
larger than RAM.  Then you do far too much i/o... more than you would if
you were using read() instead of mmap().

(Which exactly describes this week because of a .mov
mirror that's just larger than RAM.) 


---------- Forwarded message ----------
From: Alan Cox <>
Subject: Re: mmap() versus read()
To: Dean Gaudet <>
Date: Fri, 6 Mar 1998 08:52:56 +0000 (GMT)

> So is this expected?  Is there a knob I can twist?  Is this just 2.0.32
> buffer problems and would be fixed by 2.0.33?

mmap doesnt do readahead on faults in 2.0.x, that increases the I/O load
when paging mmap data versus read() quite a lot

View raw message