[cairo] [PATCH] Rely less on the FPU, take 3

[Jonathan Morton]

It turns out, of course, that memcmp() isn't the most efficient
mechanism to compare small, aligned, known-size blocks of memory.

Attached is yet another version of the patch, which is verified to
produce small but consistent increases in benchmark numbers on a
Cortex-A8.  Note that the compare function unrolls and inlines to
branch-free code on ARM, and should behave similarly on PPC and x86.

What I don't yet know is whether this version is as fast as the original
on desktop platforms, given that these have properly integrated FPUs.

-- 
------
From: Jonathan Morton
      jonathan.morton at movial.com

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Rely-less-on-DP-FPU-for-common-matrix-funcs.patch
Type: text/x-patch
Size: 2014 bytes
Desc: not available
URL: <http://lists.cairographics.org/archives/cairo/attachments/20100602/24ddb03f/attachment.bin>