Optimized memory and string operations for SPARC

Joel Sherrill joel at rtems.org
Wed Dec 6 14:02:32 UTC 2017


On Dec 6, 2017 8:58 AM, "Sebastian Huber" <
sebastian.huber at embedded-brains.de> wrote:

On 06/12/17 14:38, Joel Sherrill wrote:

> I spoke with someone yesterday at the FSW who pointed out that Linux has
> an assembly version of memcpy() for SPARC. Obviously this code is pure GPL
> so we can't use it.
>

Is this assembly version faster, better, whatever?


The person said it was faster especially on unaligned data. Supposed to
present this week so maybe will have more details.

He showed me the code and it looked more or less like the optimized ARM
version recently added to newlib. Set things up for sequences of ldd/std.

Of course, the key to all these optimized version is to get to a tight loop
moving as much data per iteration with some balance between register
pressure and cache size.

Just hoping we could find optimized versions from other sources and merge
it to newlib. Otherwise, it is tedious work.



-- 
Sebastian Huber, embedded brains GmbH

Address : Dornier
<https://maps.google.com/?q=ess+:+Dornier&entry=gmail&source=g>str. 4,
D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber at embedded-brains.de
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20171206/2a66ffd7/attachment-0002.html>


More information about the devel mailing list