On 14/09/2020 20:23, Dave Yeo wrote:
IIRC, MMX's registers were aliased with the FPU registers, making
mixing MMX and floating point inefficient. Seems that SSE2 would be
better as it can also do floating point depending on the compiler flags.
All 64 bit processors support SSE2 IIRC, while som
On 14/09/2020 09:09, John Emmas wrote:
So rather than USE_X86_MMX, should I be #defining something like
USE_SSE2 - or USE_SSE3 maybe ??
Another idea occurred to me...
Is there anything in pixman which can allow it to generate VS project
files? (e.g. meson can sometimes be used for this
On 13/09/2020 12:12, John Emmas wrote:
Understandably, USE_X86_MMX doesn't work for a 64-bit build. I
assumed there might be a 64-bit alternative (USE_X64_MMX) but that
doesn't seem to exist :-(
So do I just need to stop #defining USE_X86_MMX now? Or is there some
alternati
Hi there - hopefully this'll be a simple one...
I've been building pixman-1 (from git) for many years, using MSVC - but
until now I'd only ever built as 32-bit. This morning I tried switching
to a 64-bit build (with VS2019) and I've hit a small problem which
basically comes down to me using t
On 13/07/2020 15:57, Alan Coopersmith wrote:
Which URL did you try? g...@gitlab.freedesktop.org:cairo/cairo.git or
https://gitlab.freedesktop.org/cairo/cairo.git ?
I tried the one you gave me:- https://gitlab.freedesktop.org/cairo/cairo
- but I've just added ".git" on the end and it's worki
On 13/07/2020 15:25, Alan Coopersmith wrote:
You should be able to use the URL's shown under the "clone" button on
https://gitlab.freedesktop.org/pixman/pixman
https://gitlab.freedesktop.org/cairo/cairo
Many thanks Alan,
Haven't tried pixman yet but I just tried that for cairo but it gives
This might be a temporary issue but I can't seem to do a git pull any
more from either pixman or cairo. In both cases, TortoiseGit seems to
fail at the following stage:-
Connecting to anongit.freedesktop.org
Is this just something temporary or have the repos moved somewhere
else? Thanks
On 10/06/2019 17:33, Matt Turner wrote:
Thanks. Pushed.
___
Yes, pixman is happily compiling with MSVC again this morning. Thanks guys.
John
___
Pixman mailing list
Pixman@lists.freedesktop.org
https://l
On 06/06/2019 16:17, Basile Clement wrote:
This is supposed to be binary numbers. I forgot this was a nonstandard extension ; they
should be changed to corresponding hex values: 0b001 -> 0x1 ; 0b010 -> 0x2 ;
0b100 -> 0x4.
Thanks Basile, that works now. Would you mind pushing this upstream
MSVC is complaining about a new function called 'dither_factor_bayer_8'
(in pixman/pixman-bits-image.c). The relevant code looks like this:-
m = ((y & 0b001) << 5) | ((x & 0b001) << 4) |
((y & 0b010) << 2) | ((x & 0b010) << 1) |
((y & 0b100) >> 1) | ((x & 0b100) >> 2);
It's complai
On 18 Nov 2013, at 17:00, Søren Sandmann wrote:
>
> If that commit is making libffi make calls with a 16-byte aligned stack
> pointer, that's a good change since GCC compiled functions seems to be
> expecting that in many cases.
>
> Whether GCC *should* be expecting that is debatable (and I thin
I'm not sure if there's a connection here or if it's just a coincidence
but in a completely separate library (libffi) I noticed a very recent
commit by Mark Kettenis (6th Nov 2013) which he describes as "Align the
stack pointer to 16-bytes". Unfortunately, his change is in an assembly
language
On 05/10/2013 19:32, John Emmas wrote:
On 5 Oct 2013, at 19:00, Siarhei Siamashka wrote:
Andrea Canciani has already investigated the problem and submitted the
fixes here:
http://lists.freedesktop.org/archives/pixman/2013-September/002954.html
Many thanks for the super fast response
On 5 Oct 2013, at 19:00, Siarhei Siamashka wrote:
>
> Andrea Canciani has already investigated the problem and submitted the
> fixes here:
>
>http://lists.freedesktop.org/archives/pixman/2013-September/002954.html
>
Many thanks for the super fast response guys. I'm at a different PC now b
tried force-building
as C++ but that just gave me a whole different set of errors, so I'd be
grateful for any assistance anyone can offer. The current code is
simply not buildable with MSVC any more. Hope someone can help
John Emmas
___
15 matches
Mail list logo
