On Wed, May 22, 2013 at 12:58 AM, Markus Schaber wrote:
> Hmm, when checking those places, I'm not sure whether modern compilers
> with their auto-vectorizing optimizers really profit from the hand-optimized
> code there, or whether that code might rather trouble the optimizer into
> generating su
Hi,
Von: Ben Reser [mailto:b...@reser.org]
> On Thu, May 16, 2013 at 2:27 PM, Branko Čibej wrote:
> > On 16.05.2013 23:12, Mattias Engdegård wrote:
> >> [[[
> >> * subversion/include/svn_types.h:
> >> Set SVN_UNALIGNED_ACCESS_IS_OK for PowerPC.
> >> ]]]
> >
> > At first I thought, "that can't b
On Thu, May 16, 2013 at 2:27 PM, Branko Čibej wrote:
> On 16.05.2013 23:12, Mattias Engdegård wrote:
>> [[[
>> * subversion/include/svn_types.h:
>> Set SVN_UNALIGNED_ACCESS_IS_OK for PowerPC.
>> ]]]
>
> At first I thought, "that can't be right!" but it turns out that it is
> ... somewhat. Only i
As discussed on IRC.
Committed as r1483575 with trivial formatting changes.
Thanks for the patch!
-- Stefan^2.
On Thu, May 16, 2013 at 11:12 PM, Mattias Engdegård
wrote:
> [[[
> * subversion/include/svn_types.**h:
> Set SVN_UNALIGNED_ACCESS_IS_OK for PowerPC.
> ]]]
>
--
*Join one of our
On 16.05.2013 23:12, Mattias Engdegård wrote:
> [[[
> * subversion/include/svn_types.h:
> Set SVN_UNALIGNED_ACCESS_IS_OK for PowerPC.
> ]]]
At first I thought, "that can't be right!" but it turns out that it is
... somewhat. Only if we're sure that we use that macro for integer
access only. Whic
[[[
* subversion/include/svn_types.h:
Set SVN_UNALIGNED_ACCESS_IS_OK for PowerPC.
]]]
Index: subversion/include/svn_types.h
===
--- subversion/include/svn_types.h (revision 1483563)
+++ subversion/include/svn_types.h (work
6 matches
Mail list logo