Re: anyone notice speed of 2.17.95 on Windows ?

[David Kastrup]

"Keith OHara" <address@hidden> writes:

> On Tue, 10 Dec 2013 09:57:20 -0800, Keith OHara <address@hidden> wrote:
>
>> On Tue, 10 Dec 2013 01:59:09 -0800, Phil Holmes <address@hidden> wrote:
>>
>>> When I benchmarked with and without
>>> skylines, I found there was only a noticeable difference with a lot of
>>> markup or similar: "normal" music had almost no effect.  As a result, I
>>> concluded with skylining was the correct default.
>
>> I'll try your Mikado on my WinXP system,
>
> Chorus Number 5 takes 17.2s with 2.16.2, and takes 31.1s with 2.17.95
>
> This ratio of times is worse than Phil saw when he timed the full Act1
> using Win64Vista.  So maybe the larger slowdown is localized to WinXP.
> Disabling all the vertical-skylines-from-stencil-integral settings
> brings the time down to 16.2s with version 2.17.95.
>
> The last time we had a doubling of time required on Windows relative
> to Linux, issue 1926, it was repeated calls to find_by_name() that go
> through Pango to the font server.
>
> Here the outlines seem to be properly cached; it looks like each text
> glyph is looked up just one extra time.

Once every time it is written, or once per document?  And what does
"just one extra time" mean?  One extra per which occasion?

> The call the to FreeType glyph-outline code might cause the extra
> time, or maybe Guile is less efficient on WinXP, as every nook and
> cranny in every letter is represented in the skylines objects in
> Scheme.

If there are differences in speed, I can't imagine them to be in
anything but possibly garbage collection/allocation, and it seems
unlikely that they would not then have been apparent with other uses.

-- 
David Kastrup