Dual 2.5 GHz--great benchmarks (big jump from DP (2.0)

Posted:
in Current Mac Hardware edited January 2014
if anyone hasn't found their way to macrumors yet, they have benchmarks on the DP 2.5. the most notable improvements are in quicktime encoding . Now if I could just do my OWN benchmarks on it...

Comments

  • Reply 1 of 18
    existenceexistence Posts: 991member
    You best stay away from the Apple drones at MacRumors. The CineBench results are misleading because the new 2.5GHz G5 uses a newer G5-optimized version of CineBench while the older 2GHz G5 don't. If one uses the newer version on the 2GHz G5, one gets results one expects.



    More info.



    Opterons, dual Xeons etc.. still outperform the G5.
  • Reply 2 of 18
    marcukmarcuk Posts: 4,442member
    Cinebench is a very good benchmark for general 64bit FPU performance, and it puts a 2.5 G5 on par with a dual 3.0GHZ Xeon. However Cinebench makes no use of many of the advantages of a G5, namely Altivec, which has a massive bearing on anything programmed for it.



    I look at Cinebench as a worst case scenario for the mac performance. Its hard to imagine any real world app where a dual 2.5 G5 would only scrape equal to a dual 3.0 xeon. Anything optimized for a G5, or altivec would murder a Xeon at any speed in most cases.



    Also there is some debate on wether OSX has good OpenGL drivers, or wether Cinema4d/bench has a poor openGL implementation.



    http://www.postforum.com/forums/read...06137&t=106137
  • Reply 3 of 18
    existenceexistence Posts: 991member
    Quote:

    Originally posted by MarcUK

    Cinebench is a very good benchmark for general 64bit FPU performance, and it puts a 2.5 G5 on par with a dual 3.0GHZ Xeon. However Cinebench makes no use of many of the advantages of a G5, namely Altivec, which has a massive bearing on anything programmed for it.



    I look at Cinebench as a worst case scenario for the mac performance. Its hard to imagine any real world app where a dual 2.5 G5 would only scrape equal to a dual 3.0 xeon. Anything optimized for a G5, or altivec would murder a Xeon at any speed in most cases.



    Also there is some debate on wether OSX has good OpenGL drivers, or wether Cinema4d/bench has a poor openGL implementation.



    http://www.postforum.com/forums/read...06137&t=106137




    CineBench as the worst case? You obviously haven't seen some other scores. The performance delta is even greater for games.



    Altivec is irrelevant since Altivec is only single precision.
  • Reply 4 of 18
    xsmixsmi Posts: 119member
    Explain your last sentence please.
  • Reply 5 of 18
    zapchudzapchud Posts: 844member
    Yes, Altivec is irrelevant for the Cinema 4D potential, since the C4D rendering engine computes with 64 bit floating point values (double precision). Altivec doesn't support more than 32 bit floats, which makes it unsuitable for C4D.
  • Reply 6 of 18
    auroraaurora Posts: 1,142member
    You also have to adjust for the fact that the 2.5 had a 6800 and the 2.0 had a (yawn) 9600. i still find it amazing that Apple is using $50 fx 5200 in 2.0 and 1.8- pretty poor choice of video for a workstation class computer. but i will say this last year they went from dual 1.42 stale G4s to 2.0 G5s and now 2.5 G5s. we have gotten more progress from Ibm in a year then 5 years of Moto. Thats a very good thing.
  • Reply 7 of 18
    ugh...

    i'm getting nauseous about this whole "switching" shennanigan........... alienware..so..fast
  • Reply 8 of 18
    concordconcord Posts: 312member
    Quote:

    Originally posted by SwitchingSoon:

    ugh...

    i'm getting nauseous about this whole "switching" shennanigan........... alienware..so..fast



    No no no no.... you want the nForce 4... dual x16 slots... Athlon64 or FX.... cheaper.... faster... 8)



    Anyway, that's what I'm eyeing for my new home box if it's as good as they say.



    The Cinebench scores are about what I'd expect - they've always been the most favorable Mac 3D benchmark.





    Cheers!



    C.
  • Reply 9 of 18
    hmurchisonhmurchison Posts: 12,116member
    Quote:

    Originally posted by SwitchingSoon

    ugh...

    i'm getting nauseous about this whole "switching" shennanigan........... alienware..so..fast




    I could build a PC that blows Alienware out the water for less money. The only thing special about the Alienware computers are the cases. Everything else componentwise is readily available. However honestly I wouldn't even think about a PC unless I was doing 3D work. I'm not waiting until 2007 for Longhorn before I get some decent OS features. I guess it hasn't really sunk into many people how significant Tiger is going to be or how fast the Dual 2.5 is. Cinebench doesn't really tell me much and like MarcUK said it's a worst case scenario as 3D is a current weakness of Apple's that should be rectified with future support for decent cards as well as the ratification of OpenGL 2.0(which has good legacy support).



    Apple didn't increase the pipes to get to 2.5 so the IPC is going to scale linearly. For general purpose computing by mid 2005 we'll have a system that simply cannot be fully appreciate by numbers alone. Benchmarks will never be the same because how do you benchmark a computer which is doing heavy processing in the GPU? This won't show up on a benchmark but in actual use you will notice the fluidity and realtime functions.



    The 2.5 are going to be fast. They will keep up with the Opterons in applications that aren't dependent on memory latency(ondie mem controllers are hard to beat here)and due to the OS advancemens work better in the "real world" for your general purpose computing.
  • Reply 10 of 18
    marcukmarcuk Posts: 4,442member
    Quote:

    Originally posted by Existence

    CineBench as the worst case? You obviously haven't seen some other scores. The performance delta is even greater for games.



    Altivec is irrelevant since Altivec is only single precision.




    Yeah, like I wouldn't buy a dual 2.5 because games matter! duh
  • Reply 11 of 18
    bungebunge Posts: 7,329member
    Quote:

    Originally posted by Existence

    You best stay away from the Apple drones at MacRumors. The CineBench results are misleading because the new 2.5GHz G5 uses a newer G5-optimized version of CineBench while the older 2GHz G5 don't. If one uses the newer version on the 2GHz G5, one gets results one expects.



    More info.



    Opterons, dual Xeons etc.. still outperform the G5.




    We can't use scores on software optimized for an Apple G5, but we can use scores based on games optimized for Windows. Make up your mind.
  • Reply 12 of 18
    Cross-platform benchmarks are just plain non-sense (the concept of benchmark in general is hardly brilliant, anyway).



    For me, here is the only way to compare two platforms :

    * Take 300 people : 100 pure beginners, 100 platform A geeks and 100 platform B geeks

    * Give to all of them an extensive list of tasks to achieve

    * Put each one of them in front of a computer :

    - 50 beginners on platform A

    - 50 beginners on platform B

    - 50 platform A geeks on platform A

    - 50 platform B geeks on platform B

    - 50 platform A geeks on platform B

    - 50 platform B geeks on platform A

    * Chronometer them from the moment the push the "on" button to the moment they shut the computer down

    * Analyze the results
  • Reply 13 of 18
    Quote:

    Originally posted by The One to Rescue

    Cross-platform benchmarks are just plain non-sense (the concept of benchmark in general is hardly brilliant, anyway).



    For me, here is the only way to compare two platforms :

    * Take 300 people : 100 pure beginners, 100 platform A geeks and 100 platform B geeks

    * Give to all of them an extensive list of tasks to achieve

    * Put each one of them in front of a computer :

    - 50 beginners on platform A

    - 50 beginners on platform B

    - 50 platform A geeks on platform A

    - 50 platform B geeks on platform B

    - 50 platform A geeks on platform B

    - 50 platform B geeks on platform A

    * Chronometer them from the moment the push the "on" button to the moment they shut the computer down

    * Analyze the results




    Yeah software is a big part of the equation !



    .
  • Reply 14 of 18
    unixmacunixmac Posts: 65member
    Quote:

    Originally posted by Existence

    CineBench as the worst case? You obviously haven't seen some other scores. The performance delta is even greater for games.



    Altivec is irrelevant since Altivec is only single precision.




    A software industry insider once told me something very telling and important about Mac software vs hardware.. and I paraphrase-..



    "due to the small market share of the Mac, the business case for optimizing software on a Mac vs simple port overs and quick re-writes from the PC versions is just not there for the most part. Apple and IBM make the best hardware today with the G5, its system bandwidth, and Unix based OS, but I doubt if software will ever be optimized to reach all of that potential in the same way the it is on the PC, it would simple end up costing too much!"



    So the way I see it, we as Mac users are always going to have to accept slightly less performance out of less than optimized apps and games, in return for that great Mac experience which has a value in and of its self.
  • Reply 15 of 18
    hmurchisonhmurchison Posts: 12,116member
    Quote:

    So the way I see it, we as Mac users are always going to have to accept slightly less performance out of less than optimized apps and games, in return for that great Mac experience which has a value in and of its self.



    The only way we can combat this and affect change is to support the developers who "are" optimizing for Mac OS.



    There are plenty of Mac only developers but as long as we keep supporting the lowest common denominator apps from larger companies there is no incentive for them to optimize for Macs.



    I'll tell you I'll be supporting developers like



    Omnigroup

    Rogue Amoeba

    Stone Design

    Ambrosia Software



    and any other companie determined to make my experience as good as possible. They don't get enough credit nor attention from us and they very well should.
  • Reply 16 of 18
    unixmacunixmac Posts: 65member
    Quote:

    Originally posted by hmurchison

    The only way we can combat this and affect change is to support the developers who "are" optimizing for Mac OS.



    There are plenty of Mac only developers but as long as we keep supporting the lowest common denominator apps from larger companies there is no incentive for them to optimize for Macs.



    I'll tell you I'll be supporting developers like



    Omnigroup

    Rogue Amoeba

    Stone Design

    Ambrosia Software



    and any other companie determined to make my experience as good as possible. They don't get enough credit nor attention from us and they very well should.




    Good point.. You saw Steve J in the Keynote telling developers about new technologies in 10.4 such as core image, etc.. and then basically begging Adobe to incorporate it into the next photoshop... he shouldn't have to even say it.. they should all be jumping on it life flies on stink.
  • Reply 17 of 18
    Quote:

    Originally posted by hmurchison

    The only way we can combat this and affect change is to support the developers who "are" optimizing for Mac OS.



    There are plenty of Mac only developers but as long as we keep supporting the lowest common denominator apps from larger companies there is no incentive for them to optimize for Macs.



    I'll tell you I'll be supporting developers like



    Omnigroup

    Rogue Amoeba

    Stone Design

    Ambrosia Software



    and any other companie determined to make my experience as good as possible. They don't get enough credit nor attention from us and they very well should.






    I would like to add http://www.synthetik.com/ makers of studio artist, one of he coolest apps I've ever had a chance to play with. And you'll only find it on a Mac.
  • Reply 18 of 18
    hmurchisonhmurchison Posts: 12,116member
    Quote:

    Originally posted by Carson O'Genic

    I would like to add http://www.synthetik.com/ makers of studio artist, one of he coolest apps I've ever had a chance to play with. And you'll only find it on a Mac.



    Yes I wonder if they are looking at Core Image with some interest. I'm probably going to pick up SA because it works well with video.
Sign In or Register to comment.