On Fri, Jun 26, 2015 at 08:38:56PM +0300, Roman Lebedev wrote:
> On Fri, Jun 26, 2015 at 6:33 PM, Bruce Guenter wrote:
> > Here's what I've found out: GCC does support function multiversioning
> > (https://gcc.gnu.org/wiki/FunctionMultiVersioning) to provide the same
> > function name for differen
On Fri, Jun 26, 2015 at 6:33 PM, Bruce Guenter wrote:
> On Fri, Jun 26, 2015 at 09:58:06AM +0200, jeremy rosen wrote:
> > On Fri, Jun 26, 2015 at 6:57 AM, Bruce Guenter
> wrote:
> >
> > > In fact, it could potentially (long term) eliminate the need for
> > > process_cl() as well, with compilers
On Fri, Jun 26, 2015 at 07:12:26PM +0300, Roman Lebedev wrote:
> On Fri, Jun 26, 2015 at 6:46 PM, Bruce Guenter wrote:
> > How is it broken? Who considers it broken? I had not heard this, though
> > I don't hear much on that front. Ubuntu has it by default in the latest
> > release and I am using
On Fri, Jun 26, 2015 at 6:46 PM, Bruce Guenter wrote:
> On Fri, Jun 26, 2015 at 02:10:32PM +0300, Roman Lebedev wrote:
> > Last time it was discussed, it was agreed that there is just no compilers
> > that
> > support it:
> > *) only GCC 4.9, which is considered half-broken by many people and
>
>
On Fri, Jun 26, 2015 at 02:10:32PM +0300, Roman Lebedev wrote:
> Last time it was discussed, it was agreed that there is just no compilers
> that
> support it:
> *) only GCC 4.9, which is considered half-broken by many people and
How is it broken? Who considers it broken? I had not heard this, tho
On Fri, Jun 26, 2015 at 09:58:06AM +0200, jeremy rosen wrote:
> On Fri, Jun 26, 2015 at 6:57 AM, Bruce Guenter wrote:
>
> > In fact, it could potentially (long term) eliminate the need for
> > process_cl() as well, with compilers gaining support for offloading work
> > to accelerators. As I under
On Fri, Jun 26, 2015 at 7:57 AM, Bruce Guenter wrote:
> On Fri, Jun 26, 2015 at 02:50:45AM +0300, Roman Lebedev wrote:
> > There is also a second way (not counting leaving it as it is),
> > and i think there are valid arguments why it should be chosen: OpenMP
> SIMD.
>
> I agree that would be the
On Fri, Jun 26, 2015 at 6:57 AM, Bruce Guenter wrote:
> On Fri, Jun 26, 2015 at 02:50:45AM +0300, Roman Lebedev wrote:
> > There is also a second way (not counting leaving it as it is),
> > and i think there are valid arguments why it should be chosen: OpenMP
> SIMD.
>
> I agree that would be the
On Fri, Jun 26, 2015 at 02:50:45AM +0300, Roman Lebedev wrote:
> There is also a second way (not counting leaving it as it is),
> and i think there are valid arguments why it should be chosen: OpenMP SIMD.
I agree that would be the ideal, but I have several issues below.
I had looked at that and
There is also a second way (not counting leaving it as it is),
and i think there are valid arguments why it should be chosen: OpenMP SIMD.
1. more versions of process() - more code to keep synced
Right now we already have process() with SSE[3] and process_cl() with opencl
And even now, there is no
hi!
sounds exciting :)
1. maybe near the sse detection in src/common/darktable.c?
2. no preference from my side. i would probably put it into process() with
a branch, maybe some modules can make use of the same code in between
SIMDfied blocks.
3. it's done with dt_alloc_align, most places use 6
11 matches
Mail list logo
