SSE5 has a great chance to succeed.

By: Michael S (, June 22, 2008 1:42 am
Room: Moderated Discussions
mpx ( on 6/22/08 wrote:
>First of all - there's no sense to compare SSE5 and 3dnow. The latter is a special-purpose,
>limited-functionality extension, unusable for anything except graphics.

Why do you say so? 3dnow is better than x87 for every application that does tones of single-precision FP calculations. That covers all FP signal processing, hi fi audio and even some parts of HPC (weather forecast?). May be, the performance improvement of 3DNow! over x87 are not big in absolute terms but certainly bigger than improvement that SSE5 brings over previous generation of SSE.

>3dnow was
>also different from the other products on the market then.

I don't understand how SSE5 is not in the same boat.

>SSE5 on the other hand is a general purpose instructions set, viable for technical
>and scientific applications. What's more these instructios are of a known, proven
>quality, as they are similar to what's available in other (eg. PowerPC) ISAs.

By chance, "generic" PowerPC has no double-precision SIMD. There are PowerPC _extensions_ that have it (BlueGene, Cell SPE) but they are not compatible with each other.

>The software side is also important. By not increasing register size AMD ensured
>the support for SSE5 on older operating systems, like Windows XP or Windows 2003
>Server. It's most probable that Vista-pushing Microsoft won't patch older OSes to
>support register extensions. And considering slow Vista adoption, this may matter.


>And the market has changed. Perhaps now it will be enough for new instruction sets
>to support immediate layers, like:.NET, Java, graphics drivers (like ATI :-), numeric
>libraries (BLAS etc.), gaming engines/frameworks and of course compilers to make
>enough impact to popularize instructions on the market. And it'll be easy to get
>manufacturers support for just a bunch of new instructions, while larger vectors
>may force them to change algorithms and data structures.

SSE5 has no non-destructive FMUL (nor does it have non-destructive FADD but non-destructive FMUL is more important). For me on that merit alone SSE5 deserved to dye a painful death.

>Besides AMD is trying hard to popularize GPGPU computing, that are exactly the
>type of applications that benefits from very large vectors on the CPU, so doing
>too good FPU may be counterproductive. If they are really serious about integrating
>GPUs with CPUs it may even make more sense to invest in speeding it up, than in
>uncertain technologies like very wide vector units.
>Because this vector computing is a gamble. We don't know if it'll work out. Perhaps
>the future lies in hordes of scalar micro-cores? And if vectors are OK - why not
>go to 8192 bits at once? Or a hybrid approach with some cores having wide vectors,
>some not. By keeping things as they are, just improving them a bit, AMD is avoiding risks they now cannot afford.

Why should I care about the benefits for AMD and their internal politics? As x86 ISA user I don't want extension that fragments the market nor I want the universally-accepted extension that significantly pollutes an opcode space for the marginal performance gain. Actually, given the choice I'd prefer the former as the lesser evil.
< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
Intel AVX kills AMD SSE5Agner2008/06/17 08:14 AM
  Intel AVX kills AMD SSE5a reader2008/06/17 09:03 AM
  Bulldozer?David Kanter2008/06/19 04:23 PM
    Bulldozer?EduardoS2008/06/19 06:11 PM
    Bulldozer?Max2008/06/19 06:16 PM
      Bulldozer?Goose2008/06/21 02:23 AM
        Bulldozer?David Kanter2008/06/21 07:37 AM
          Bulldozer?someone2008/06/21 07:55 AM
            Bulldozer?David Kanter2008/06/21 08:07 AM
              Bulldozer?S. Rao2008/06/21 11:08 AM
                RegfilesPeter2008/06/21 11:49 AM
                Bulldozer?Linus Torvalds2008/06/21 12:23 PM
                  Bulldozer?S. Rao2008/06/21 04:50 PM
                  unified physical register file nullifies x86 advantageMichael S2008/06/22 12:24 AM
                    unified physical register file nullifies x86 advantageDavid Kanter2008/06/22 09:35 AM
                      unified physical register file nullifies x86 advantagehobold2008/06/22 01:03 PM
                    Reg file vs. forwarding networkDavid Kanter2008/06/22 10:36 AM
                      Reg file vs. forwarding networkhobold2008/06/22 12:39 PM
                        Reg file vs. forwarding networkPeter2008/06/22 02:48 PM
                          Reg file vs. forwarding networkDavid Kanter2008/06/22 08:54 PM
                            Reg file vs. forwarding networkPeter2008/06/23 03:44 AM
                              Reg file vs. forwarding networksavantu2008/06/23 04:41 AM
                                Reg file vs. forwarding networkPeter2008/06/23 07:35 AM
                        Reg file vs. forwarding networkAnders Jensen2008/06/23 11:05 AM
                        Reg file vs. forwarding networkleft nutz2008/06/27 07:31 AM
  Intel AVX kills AMD SSE5nobat2008/06/21 11:23 AM
    Intel AVX kills AMD SSE5Agner2008/06/21 10:01 PM
      So...Dean Kent2008/06/22 07:35 AM
  SSE5 has a great chance to succeed.mpx2008/06/22 12:25 AM
    SSE5 has a great chance to succeed.Michael S2008/06/22 01:42 AM
      SSE5 has a great chance for fiascoAgner2008/06/22 03:32 AM
        SSE5 has a great chance for fiascoIan Ameline2008/06/22 08:37 AM
        SSE5 has a great chance for fiascoanonymous2008/06/22 09:02 AM
          SSE5 has a great chance for fiascohobold2008/06/22 12:59 PM
            SSE5 has a great chance for fiascoHoward Chu2008/06/22 04:38 PM
      SSE5 has a great chance to succeed.hobold2008/06/22 12:52 PM
        SSE5 has a great chance to succeed.Michael S2008/06/22 01:46 PM
          SSE5 has a great chance to succeed.Hannes2008/06/24 08:49 AM
            SSE5 has a great chance to succeed.anonymous2008/06/24 10:46 AM
          SSE5 has a great chance to succeed.Ian Ollmann2008/06/24 10:12 PM
Reply to this Topic
Body: No Text
How do you spell tangerine? 🍊