[Pixman] performance of bilinear_interpolation

[xunxun]

On Wed, Oct 9, 2013 at 1:52 PM, Maksim Lin <maks at manichord.com> wrote:

> scaled_bilinear_scanline_sse2_x888_8888_SRC
>

pixman-sse2.c needs USE_SSE2=1

Do you make sure Firefox for Android for x86's libpixman's module defines
USE_SSE2.

I saw the makefile:
---------------------------------------------
ifdef GNU_CC
ifeq (ppc,$(findstring ppc,$(OS_TEST)))
USE_VMX=1
VMX_CFLAGS=-maltivec
endif
ifeq (86,$(findstring 86,$(OS_TEST)))
USE_MMX=1
MMX_CFLAGS=-mmmx -Winline
ifeq (64,$(findstring 64,$(OS_TEST)))
USE_SSE2=1
endif
ifdef HAVE_GCC_ALIGN_ARG_POINTER
USE_SSE2=1
endif
ifdef USE_SSE2
SSE2_CFLAGS=-msse -msse2 -Winline
endif
ifneq ($(MOZ_WIDGET_TOOLKIT),os2)
MMX_CFLAGS+=--param inline-unit-growth=10000 --param
large-function-growth=10000
endif
endif
---------------------------------------------------------
I don't know your $(OS_TEST), if it has 64, USE_SSE2 is defined.

And except it, GCC USE_SSE2 depends on HAVE_GCC_ALIGN_ARG_POINTER, too.
Does your toolchain support

__attribute__ <https://mxr.mozilla.org/mozilla-central/ident?i=__attribute__>
((__force_align_arg_pointer__))  ?




-- 
Best Regards,
xunxun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/pixman/attachments/20131009/7239c8be/attachment.html>