Dual G5 versus Dual Opteron
  • Order of the Butterfly
    Order of the Butterfly
    liquidbit
    Posts: 407 from 2003/10/12
    Good news for the G5 processor !!
    Genesis what are the plans for the Pegasos III?
    Am I asking too much?
    If you bring it, I 'll definitely buy it !

    http://www.barefeats.com/g5op.html

    Edited by FALCON1 to correct URL link :-)
    ..there will be only one left.
  • »16.12.03 - 21:07
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    KimmoK
    Posts: 102 from 2003/5/19
    And never trust any single benchmark as the final word about the performance...

    From AMDzone:
    "Pointless G5 vs Opteron Benchmarks
    Reported by: Chris Tom <mailto:chris@amdzone.com> At: 8:21 AM Source: The Inquirer <http://www.theinquirer.net>
    Barefeats is at it again. This time they compare <http://www.barefeats.com/g5op.html> a dual XI Opteron 246 against a dual 2GHZ G5 PowerMac. They do not get performance equal to my 246 tests. They do not acknowledge the Opteron 248. They do not give out full system specs. They do not realize beta Windows XP 64 for AMD64 has been out for months. They continue to have no idea how going from 128mb of video memory to 256mb changes gaming performance, and can not fathom that the integrated memory controller is the real reason that the games and other 3D marks are much higher. I mean come on, more video memory is going to make Quake 3 Arena faster? They also does not run SMP Quake 3 Arena. Barefeats is in no way qualified to run or comment on any benchmarks as they have demonstrated almost no knowledge of CPU or video card hardware. Do not take any of their numbers seriously. Their Cinebench marks are not as high as mine, and I worry that will also be the case with the other marks. Scott at Tech Report <http://www.techreport.com> had told me he was trying to get a dual 2GHz G5 box to test, but of course Apple does not have the guts to let a real hardware site test one of their boxes."
    :-x :-P 8-)
  • »17.12.03 - 06:58
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    liquidbit
    Posts: 407 from 2003/10/12
    I dont really care about the stupid behaviour of apple, I care about the numbers.

    but if you dont find that site serious, maybe this one would answer your questions?
    http://www.top500.org/list/2003/11/
    ..there will be only one left.
  • »17.12.03 - 07:18
    Profile
  • Butterfly
    Butterfly
    tomjoad
    Posts: 99 from 2003/2/24
    Quote:

    I dont really care about the stupid behaviour of apple, I care about the numbers.


    Who cares about numbers? I wouldn't trust any benchmark, no matter whether it's graphics cards, processors or entire machines.

    Besides, what's the point benchmarking a Mac against a PC? It's totally different architectures with totally different OSes. Show me just three people who will choose the OS they use based on benchmark numbers. Ok, it may be interesting to see whether both are about equal or Apple way behind. That's all. No Apple fan will go Windows because of benchmark numbers. Nor vice versa.

    I've got a Dual G5 here, and it works well and is more than fast enough and that's all I'm interested in. Just the AppleCare "expert" support sucks like hell. No experts there, just monkeys trained to talk away existing problems.

    As for Dual G5 in Pegasos, what exactly would you want to do with this? It would most certainly be more expensive than a G3 machine and there's very little software out there that can make use of that speed. Heck, there is not even a single program out there supporting Altivec, let alone specially modified programs/MorphOS that could exploit this processor. I doubt GCC 2.95 has or will get any G5-related optimisations. I'd rather go with a cheaper machine that's fully sufficient for most software out there - for now.

    Until there are firm deals with authors of software that would make G5-optimised software, I doubt a G5 will make much sense in Pegasos other than for the children ("I've got a better processor than you").

    Don't get me wrong. I'm not saying that I don't want to see a G5 Pegasos. But I'm already having a really hard time trying to push the processor load of my Dual G5 Mac >10%. For current Amiga apps, I consider it overkill.

    Just the processor alone isn't enough also. I've been working with SoundFX for quite some time on my Amiga. I'm currently working with the trial version of Bias Peak 4 on my Mac. It's a real pleasure to deal with sound files that are several GBytes in size where I was pretty much annoyed in SFX with 700 MByte files already. Faster processor alone won't help much, Peak simply has smarter data structures (I suppose they use some kind of piece table vs. SFX probably using a single chunk of memory although I haven't used the latter for some time and do not exactly recall its behaviour in certain situations).

    Ideally, that's why Genesi should do it like Apple and develop their own apps. Now I'm not sure whether it's smart that Apple themselves are competing with some software companies for their own platform. For the Pegasos, it would be my preferred solution, because I don't see much happening here otherwise in terms of smart and powerful software. I already told some Genesi employees several times, but noone's listening. Giving machines to developers is a Very Good Thing, but it's not a replacement for Genesi-internally led projects that are not cancelled at the author's whims and undergo some quality assurance. Probably all asleep. Goodnight.
  • »17.12.03 - 10:50
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    minator
    Posts: 370 from 2003/3/28
    >And never trust any single benchmark as
    >the final word about the performance...

    ...or comments from the AMDZone.

    The last barefeats benchmarks showed the G5 doing pretty applingly so I don't think anyone can complain these are biased.

    They show the G5 pretty much sucks on games (no surprise there as thats hardly the Mac's main area). The other tests are more interesting with the G5 holding it's own and sometimes showing a considerble lead, evidently this rather annoys some Opteron fans.

    I hadn't seen any benchmarks against the Opteron so it's good to see some at last.

    For a new processor it's a damn good showing. The 2nd gen 970 should give the Opteron a good run for it's money, especially with maturing compilers etc.
  • »17.12.03 - 10:53
    Profile Visit Website
  • Order of the Butterfly
    Order of the Butterfly
    liquidbit
    Posts: 407 from 2003/10/12
    Well tomjoad, I understand your concerne but they are people who use their systems for research instead of a office use and they looking a very low budget super-computer.
    The compare ofcourse is not so objective as I would like to see but atleast I 'm able to see the difference with an existing software.
    ..there will be only one left.
  • »17.12.03 - 11:54
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    KimmoK
    Posts: 102 from 2003/5/19
    Well, I definitely did not mean to say that barefeats are cheating and AMD zone posters are right. The AMD zone poster just pointed out some valid points.

    I'm pretty sure AMDzone would not be capable of providing any less biased G5 vs Opteron information.


    In general, also I often refer to barefeats when I need some x86 vs PPC information.
    Barefeats might be the best attempt at trying to compare x86 and PPC performance in several real world applications (where it matters).

    For some uses CPU performance means a lot. For the rest, it's just "nice to know" information.

    I find it pleasing that PPC CPUs are now capable of delivering more compareable performance than before.

    For my needs they definitely seem to be "fast enough".
    :-x :-P 8-)
  • »18.12.03 - 14:47
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    Wishmaster
    Posts: 342 from 2003/6/29
    Finally the nice PowerPC is catching the ugly x86.
    IBM go!
    Pegasos PPC
  • »20.12.03 - 01:57
    Profile