[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Fwd: current development
From: |
Joseph Heled |
Subject: |
Fwd: current development |
Date: |
Wed, 4 Dec 2019 11:08:22 +1300 |
---------- Forwarded message ---------
From: Joseph Heled <address@hidden>
Date: Wed, 4 Dec 2019 at 11:07
Subject: Re: current development
To: Russ Allbery <address@hidden>
Yes, I assumed GNUBG is not ready yet, but was not sure if you can
actually compile a (say) version with SSE2 and AVX support which will
run as long as the binary does not use those instructions??
On Wed, 4 Dec 2019 at 11:04, Russ Allbery <address@hidden> wrote:
>
> Joseph Heled <address@hidden> writes:
>
> > I fully appreciate your plight as a package maintainer, but I think that
> > users which use rollouts, a quite important GNUBG feature, care a *lot*
> > about speed.
>
> > Perhaps the GNUBG code can be structured in a way that it can be
> > compiled with all those "advance features", yet they are used *only* in
> > rollouts and so everything else works on older systems?
>
> > Is that possible or am I just showing my ignorance?
>
> I don't see an obvious reason why it wouldn't be possible, although this
> is currently not what gnubg does.
>
> #if defined(USE_SIMD_INSTRUCTIONS)
> result = SIMD_Supported();
> switch (result) {
> case -1:
> outputerrf(_("Can't check for SIMD support\n"));
> break;
> case -2:
> outputerrf(_("No cpuid check available\n"));
> break;
> case 0:
> /* No SIMD support */
> break;
> case 1:
> /* SIMD support */
> simderror = FALSE;
> break;
> default:
> outputerrf(_("Unknown error while doing SIMD support test\n"));
> }
>
> if (simderror) {
> #if defined(USE_AVX)
> outputerrf(_
> ("\nThis version of GNU Backgammon is compiled with
> AVX support but this machine does not support AVX\n"));
> #else
> outputerrf(_
> ("\nThis version of GNU Backgammon is compiled with
> SSE support but this machine does not support SSE\n"));
> #endif
> exit(EXIT_FAILURE);
> }
> #endif
>
> In other words, it checks at runtime if built with AVX to confirm that AVX
> is supported and, if not, it terminates the program. I could define
> DISABLE_SIMD_TEST; then, I'm not sure what behavior it would have.
>
> --
> Russ Allbery (address@hidden) <https://www.eyrie.org/~eagle/>
- Re: current development, (continued)
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Øystein Schønning-Johansen, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Øystein Schønning-Johansen, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Øystein Schønning-Johansen, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Russ Allbery, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Russ Allbery, 2019/12/03
- Message not available
- Fwd: current development,
Joseph Heled <=
- Re: current development, Ralph Corderoy, 2019/12/04
- Re: current development, Philippe Michel, 2019/12/03
- Re: current development, Russ Allbery, 2019/12/03
- Re: current development, Philippe Michel, 2019/12/04
- Re: current development, Philippe Michel, 2019/12/03
- Re: current development, Russ Allbery, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03
- Re: current development, Joseph Heled, 2019/12/03