Re: multispectral file access

On 1 September 2014 16:02, Bob Friesenhahn <[email protected]> wrote:
> It would be interesting to see how the GraphicsMagick results change
> for this benchmark with this environment variable setting:
>
>    MAGICK_MMAP_READ=TRUE

Oh that's interesting. Yes, vips did something similar: too much VM
was being chewed up when entire image files were mapped, so it now
just maps "windows" of a few hundred scanlines and scrolls them around
as operations read pixels.

This benchmark just uses libtiff for file read, so vips will be using
whatever that is configured to use, read() by default now I think.

I tried with and without this, but it didn't make a large difference,
on this machine at least:

Without the mmap var:

$ ./benchmark.sh
clearing test area ...
building test image ...
tmp/x.tif: 5000x5000 uchar, 3 bands, srgb, tiffload
program, time (s), peak memory (MB)
gm, 0.49, 260.66796875

And with:

program, time (s), peak memory (MB)
gm, 0.51, 258.2734375

This is Ubuntu 14.04 with a six-core i7.

Measuring RSS: I spent ages on this, it's a very hard thing to do
well. I tried five or six different things, including parsing strace
output. The numbers above are found by running "ps" at 100 Hz and
watching the RES line.

John

------------------------------------------------------------------------------
Slashdot TV.  
Video for Nerds.  Stuff that matters.
http://tv.slashdot.org/