[Pixman] [cairo] Floating point API in Pixman

Siarhei Siamashka siarhei.siamashka at gmail.com
Tue Aug 24 17:37:10 PDT 2010


On Thursday 12 August 2010 19:59:29 Koen Kooi wrote:
> On 12-08-10 12:24, Soeren Sandmann wrote:
> > Andrea Canciani <ranma42 at gmail.com> writes:
> >> If we can't assume that we have a (fast) FPU,
> > 
> > There is a fairly common misconception that ARM CPUs never have
> > hardware floating point; that may be part of why some people are
> > worried.
> > 
> > However, until we hear from someone who can credibly claim that he is
> > going to ship pixman to a non-trivial number of people on non-FPU
> > hardware, I think assuming hardware floating point is fine.
> 
> At TI we're still selling millions of ARM9 based SoCs without an FPU on
> the arm side. A fair percentage will end up running linux with a GUI.
> The market for cheap ARM chips is huge, not everyone wants to pay $30 or
> more for the latest A8/A9/eagle.

This is only my personal opinion and I surely don't want to sound rude, but 
there is no free lunch. Those who are saving money by using less expensive 
chips should understand that software development also requires more efforts, 
time (and as a result - money) if some of the useful hardware features can't be 
used. Or when multiple alternative implementations have to be maintained.

Do those who are making decisions about what CPU core is going to be used in 
new SoCs take into account these hidden software related costs?

Honestly, I do not see a lot of open source activity developing optimizations 
for older ARM cores. Moreover, maybe the code is tested for correctness, but I 
offhand even don't remember seeing any benchmarks reports or complaints about 
performance from low end ARM hardware users. Could it be that such hardware is 
used in the areas where performance is totally irrelevant?

Anyway, speaking as a purely software guy, I definitely would prefer to work 
with more capable hardware :) Maybe it's time to consider something like ARM 
Cortex-A5 with both NEON and VFP as a solution for low end *application 
profile* processors? Also what about at least ARM11 with VFP unit? TI used to 
have it in OMAP2. Was it much more expensive than ARM9?

I don't mean to be too negative. If there are lots of pixman users on less 
capable ARM hardware and they are just silent for whatever reason (maybe they 
are just perfectly satisfied?), then a bit more feedback would be welcome in 
order not to forget that they exist :)

-- 
Best regards,
Siarhei Siamashka
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/pixman/attachments/20100825/6ec5f1a0/attachment.pgp>


More information about the Pixman mailing list