Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
But, notice that most of them are dual. I say second is not bad.
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
I'm sure the Opteron 250 would exceed even the highest numbers. I doubt a Xeon 3.2 would have made that much more of a difference. Quad Xeons? This isn't a pissing contest..this is a rather informal test showing up the relative speed of the Powermac 2.5 in a test that runs better on PCs to begin with(Cinebench, while optimized, still favors PCs).
So again this is very encouraging news if the benchmarks are to be trusted.
Funny thing is PC Apologists also resort to lame whining about benchmarks. First they will launch into hyperbole over something like how fast the Opteron is and when that falls apart they move on to more apologetic statements like.
Quote:
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
Apple needs multiple processors to compete against multiple processors. Wow, great find!
I hope IBM gets their act together and pushes some 3.2 GHz G5s in January. Because right now, there's really no incentive for high-end pros to buy G5s instead of PCs. (Yes, it is a Mac, but that doesn't outweigh "it's a good 10% slower than a top-of-the-line PC for the same price")
Cinebench 2003 G5 beta is a quick dirty compilation of the source code with the G5 tags enabled. It took Maxon 3 weeks to get a 30% performance increase over the standard cinebench. They've also stated that they were only getting started.
Cinebench doesn't use any Altivec or other goodies specific to the 970 which other chips (x86) don't have. It [render] is mainly a no thrills 64 bit FPU test.
Also the Intel compilers are far more performance orientated than the beta compilers Maxon used for the G5 Cinebench, a situation that will only improve to the G5's favour as better g5 compilers become available. There isn't anywhere as much headroom in the Intel compiler for future performance increases. Unfortunately Maxon seem to be waiting for a Metrowerks (read Motorola!) G5 compiler which might be some time coming.
With this in mind a Dual G5 stacks up extremely well against a dual 3ghz Xeon.
Unoptimized code on beta compilers of a new architecture equals parity to highly optimized code on established compiler in a proven architecture running 1GHZ faster
Give it some other cross platform tests that compare Altivec to SSE and the G5 seriously smacks the ass of a Xeon at any speed. Video, music, Multimedia etc.
Unfortunately Maxon seem to be waiting for a Metrowerks (read Motorola!) G5 compiler which might be some time coming.
Which compiler were they using? How does it compare to gcc 3.x and IBM's xlc for G5 compilation? And surely Motorola don't make anything like the 970, so why would a Moto G5 compiler be any good for the 970?
/wonders how recent my work's version of xlc is...
Which compiler were they using? How does it compare to gcc 3.x and IBM's xlc for G5 compilation? And surely Motorola don't make anything like the 970, so why would a Moto G5 compiler be any good for the 970?
/wonders how recent my work's version of xlc is...
They are using Metrowerks at the moment. I don't know what compiler they used for Cinebench, (I'd imagine it was GCC, since that was the only compiler available for the G5 when Cinebench was released over a year ago) but the main Cinema 4d application uses Metrowerks for sure. They have stated that they have to use Metrowerks, because changing to another compiler will break all of the plug-ins, or all third party developers will have to change and recompile.
How does it compare? I dont know, but extensive support for a G5 class processor isn't very promising coming from Moto is it?
Comments
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
Originally posted by Existence
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
But, notice that most of them are dual. I say second is not bad.
Originally posted by Existence
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
I'm sure the Opteron 250 would exceed even the highest numbers. I doubt a Xeon 3.2 would have made that much more of a difference. Quad Xeons? This isn't a pissing contest..this is a rather informal test showing up the relative speed of the Powermac 2.5 in a test that runs better on PCs to begin with(Cinebench, while optimized, still favors PCs).
So again this is very encouraging news if the benchmarks are to be trusted.
Funny thing is PC Apologists also resort to lame whining about benchmarks. First they will launch into hyperbole over something like how fast the Opteron is and when that falls apart they move on to more apologetic statements like.
Too bad they didn't test a 2.4GHz Opteron 250 or 3.2GHz MP or Quad Xeons on 800MHz FSB. Now that would be interesting instead of these tests that compare the G5 to year-old CPUs.
Originally posted by Existence
Also notice how Apple needs multiple processors to compete. Unfortunately for Apple, most applications do not benefit appreciably from dual processors.
Apple needs multiple processors to compete against multiple processors. Wow, great find!
Cinebench doesn't use any Altivec or other goodies specific to the 970 which other chips (x86) don't have. It [render] is mainly a no thrills 64 bit FPU test.
Also the Intel compilers are far more performance orientated than the beta compilers Maxon used for the G5 Cinebench, a situation that will only improve to the G5's favour as better g5 compilers become available. There isn't anywhere as much headroom in the Intel compiler for future performance increases. Unfortunately Maxon seem to be waiting for a Metrowerks (read Motorola!) G5 compiler which might be some time coming.
With this in mind a Dual G5 stacks up extremely well against a dual 3ghz Xeon.
Unoptimized code on beta compilers of a new architecture equals parity to highly optimized code on established compiler in a proven architecture running 1GHZ faster
Give it some other cross platform tests that compare Altivec to SSE and the G5 seriously smacks the ass of a Xeon at any speed. Video, music, Multimedia etc.
Here http://www.3dfluff.com/mash/cbdual.php is an extensive list of Cinebench results
Unfortunately Maxon seem to be waiting for a Metrowerks (read Motorola!) G5 compiler which might be some time coming.
Which compiler were they using? How does it compare to gcc 3.x and IBM's xlc for G5 compilation? And surely Motorola don't make anything like the 970, so why would a Moto G5 compiler be any good for the 970?
/wonders how recent my work's version of xlc is...
Originally posted by Stoo
Which compiler were they using? How does it compare to gcc 3.x and IBM's xlc for G5 compilation? And surely Motorola don't make anything like the 970, so why would a Moto G5 compiler be any good for the 970?
/wonders how recent my work's version of xlc is...
They are using Metrowerks at the moment. I don't know what compiler they used for Cinebench, (I'd imagine it was GCC, since that was the only compiler available for the G5 when Cinebench was released over a year ago) but the main Cinema 4d application uses Metrowerks for sure. They have stated that they have to use Metrowerks, because changing to another compiler will break all of the plug-ins, or all third party developers will have to change and recompile.
How does it compare? I dont know, but extensive support for a G5 class processor isn't very promising coming from Moto is it?