Re: armv7hl requirements

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, May 5, 2011 at 7:35 AM, Gordan Bobic <gordan@xxxxxxxxxx> wrote:
> On 05/04/2011 07:46 PM, Jon Masters wrote:
>> Folks,
>>
>> I'd like to kick off a discussion about flags for ARMv7. My proposal
>> here is that we treat v7hl as an entirely different architecture, and
>> don't try any multi-arch kind of hacks (there isn't the established user
>> base for Fedora ARM to justify doing any of those things at the moment).
>
> Agreed, as long as armv5tel remains supported.
>
>> Things I think we should consider as a minimum:
>>
>> *). Little endian (obviously, but worth stating) (l)
>> *). Cortex-A8 or higher fully compliant core(s)

Defiantly compile with -march=armv7-a to enable the Cortex-A8 errata
work arounds.  Tuning is another story.  Code tuned for an A8 will
work quite well on an A9, but code tuned for an A9 may do poorly on an
A8.

>> *). ARM VFP3 hardware floating point (h)
>> *). ARM NEON Architecture
>> *). Thumb2 interworking

I don't think you have to anything explicit here.  Thumb-2 and ARM
code interoperate just fine.

>> *). Your suggestion here?

*) Hard float.  Never worse than soft float, sometimes much better.

> I think NEON is one step too far because Tegra doesn't support it

I'm quite fond of my AC100.  I'd recommend VFPv3-D16 and then using
IFUNC or hwcaps to pull in NEON optimised libraries for things like
ffmpeg.

>> I think we should build for ARM (as opposed to Thumb2) but we should
>> support interworking with Thumb2 code through the toolchain options. We
>> should then later consider implementing some Thumb2 optimization. It's
>> more armv7thl, but the (t) is implied since it's ARMv7 anyway.
>
> What reason is there to not use Thumb2 on ARMv7?

Thumb-2 code is a bit slower but significantly smaller.  Benchmarks in
Thumb-2 mode run at around 93 % of the speed of the same code in ARM
mode and take around ~75 % of the space.  For some applications
Thumb-2 should be faster as more of the hot loops will fit in the I
cache. I haven't seen that in any benchmarks so far.

See also Dave's email at:
 http://lists.linaro.org/pipermail/linaro-dev/2011-April/004106.html

for the difference between an ARM and Thumb-2 kernel.

I recommend Thumb-2.  It's also were we're putting our efforts in in Linaro.

-- Michael
_______________________________________________
arm mailing list
arm@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/arm


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Linux ARM (Vger)]     [Linux ARM]     [ARM Kernel]     [Fedora User Discussion]     [Older Fedora Users Discussion]     [Fedora Advisory Board]     [Fedora Security]     [Fedora Maintainers]     [Fedora Devel Java]     [Fedora Legacy]     [Fedora Desktop]     [ATA RAID]     [Fedora Marketing]     [Fedora Mentors]     [Fedora Package Announce]     [Fedora Package Review]     [Fedora Music]     [Fedora Packaging]     [Centos]     [Fedora SELinux]     [Coolkey]     [Yum Users]     [Tux]     [Yosemite News]     [Linux Apps]     [KDE Users]     [Fedora Tools]     [Fedora Art]     [Fedora Docs]     [Asterisk PBX]

Powered by Linux