finally, Macworld has posted some benchmarks analysis (2 pages)...
http://www.macworld.com/2006/01/features/imaclabtest1/index.php
http://www.macworld.com/2006/01/features/imaclabtest1/index.php
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Macworld Lab's First Tests: iMac with Intel Core Duo processor
- Thread starter irrªtiºnal
- Start date
- Sort by reaction score
This doesn't really tell us anything that we shouldn't have known already. The Intel-based iMacs boot faster and they are faster running native programs and slower while emulating through Rosetta.
This only confirms what we already knew. If you ruly on non-native programs, you would do better sticking with the G5 iMacs.
This only confirms what we already knew. If you ruly on non-native programs, you would do better sticking with the G5 iMacs.
MBHockey said:
Why not? Tiger 10.4.4 is Intel native, so why do we have to wait for Leopard for it to perform well?
Other non-native apps, yes, but not the OS.
Interesting how the native tasks aren't all that much faster.
The 512 RAM used on MW tests is an issue though I think too. Apparently Rosetta is a RAM hog, and the 512 only allows 256/core. Not really enough..
Try with 1GB/2GB. (On both systems). I think the percentage increase would go up.
The 512 RAM used on MW tests is an issue though I think too. Apparently Rosetta is a RAM hog, and the 512 only allows 256/core. Not really enough..
Try with 1GB/2GB. (On both systems). I think the percentage increase would go up.
Yeah, exactly. It's the FIRST intel native release.
It just is inevitable. A huge platform switch like this. It will require time and optimization before we see the true improvements.
Think of the switch from OS 9 to OS X. OS X was a mess at the start, but turned out to be much better than its predecessor.
It just takes time.
It just is inevitable. A huge platform switch like this. It will require time and optimization before we see the true improvements.
Think of the switch from OS 9 to OS X. OS X was a mess at the start, but turned out to be much better than its predecessor.
It just takes time.
i wouldn;t know why, but i am tempted to say that while by native, Apple meant some sort of chip-specific optimazation, by universal Apple must have done some sort of Google "translate this program" and then appended it to the end of G5 native code... emulated code is not surprisingly slow, but 50% performance of a fully G5 Pro app is too much (or, should i say, too little?).
really there is dissapointment when finding this performance "gains"... funny thing is that Core Duo saves energy as hell, when one does not really care about it in a desktop...
really there is dissapointment when finding this performance "gains"... funny thing is that Core Duo saves energy as hell, when one does not really care about it in a desktop...
I think you're gonna need more RAM than you're used to getting in order to see the benefits of dual core. Most of us are still used to single core CPUs with 1 GB of RAM. Comparing a single G5 chip with an Intel Core Duo isn't really fair. Also, just because the Intel iMac performs only slightly better than the G5 iMac when both are loaded with 512 MB doesn't mean that it wouldn't completely whip the G5 iMac when both are loaded with 2 GB
As is generally the case, it seems the hardware is ready, but the software is not.
It's unknown outside of Apple whether 10.4.4 was compiled with the gcc compiler or an alpha/beta version of Intel's optimizing compiler. I tend to think gcc was used (it would make sense to use a stable, tested compiler for something as important as the operating system).
If Leopard will be compiled with a release version of Intel's optimizing compiler, we may see these benchmarks improve on this basis alone.
It's unknown outside of Apple whether 10.4.4 was compiled with the gcc compiler or an alpha/beta version of Intel's optimizing compiler. I tend to think gcc was used (it would make sense to use a stable, tested compiler for something as important as the operating system).
If Leopard will be compiled with a release version of Intel's optimizing compiler, we may see these benchmarks improve on this basis alone.
dwd3885 said:
Lots of the tests are for applications that I don't expect to take advantage of a dual processor system on their own, so in many cases your PowerMac Dual Core will not be much faster than the iMac G5 either. They should start Handbrake encoding a DVD, and _then_ run all their benchmarks at the same time, and you will see the dual processor system smoking the single processor system.
Some of these tests make you wonder what's going on. Benchmarking BBEdit??? That is ridiculous.
Personally I would still have no hesitation at all going for a Dual G5 PowerMac. As for the iMac I'd probably go Intel over G5 at this point but only because the graphics chipset is so good. Remember Universal binary will still run properly on the legacy hardware -- it isn't going to be like the way PowerPC optimized stuff ran on the 68000 series at all.
On the notebook end of things however, going for the MacBook over a G4 PowerBook is a no-brainer.
On the notebook end of things however, going for the MacBook over a G4 PowerBook is a no-brainer.