Separate names with a comma.
Discussion in 'Mac Pro' started by IEatApples, Jan 24, 2006.
Out of 13 test only 1 did the G5 win, fact is this is a faster iMac. I think we have a lot of PPC lovers that just cant let go of the fact that G5 aint all that. Let it go, Intel is faster.
It's a shame that in all the Windows vs Mac hype, we have that PPC vs x86 war within a supposedly perfect Mac community.....
truly a sad transition for Apple.....
Yes, I know, but I'm just saying that the Intel iMac has been "hyped" a bit much. I think that the ones comming this summer will be the "true" stuff.
What a surprise.
What a surprise, they did Rosetta tests with 512 RAM in there.
WASTE OF MY FRIKKEN TIME
But it was vs. a G5 with the same amount of RAM
i have to agree. Conroe and Merom will be the big thing. I think that the purpose of going with the same design on the Mactels was to have as few people buy them until they get all of the bugs out of them. When all of the bugs are out of them, by april maybe, they will release new things with new designs, with an overall beter computing experience. Merom and conroe will be in the new Macs as soon as they can get a hold of them, and they might follow this same pattern. If they make a brand new thing in todays limited supplies, they will be sold out with as many people buying them that can afford them, getting shipping until late march, and people who need a new computer soon would not be able to wait that long would buy the old thing or a windows. Once supply and demand evens out, apple will release all of their cool innovative things.
I do think that these new macs have been over 'hyped' and the tests on apple are exagurated.
Photoshop On my InteliMac is not doing anything (just sitting open in hte dock) and is currently using 950MB RAM - doing nothing.
Anything under rosetta EATS ram. Dual Cores (using them) requires a lot of ram to use them to there fully potential.
What I'd like to know is, what happens when you try to do two CPU intensive tasks at once?
An example, I'm a Java developer, and do project builds many many times a day. I also do maintenance on my PC while waiting for builds to complete, for instance, tarring up my iTunes directory, compressing it and backing it up, using a shell script in the background. There are often tasks I do that mean I can't do anything else, and just have to take a coffee break until they complete, either because the CPU is at 100% or the UI has become totally unresponsive.
On my Dell Latitude 810, with a 2.26 GHz Pentium-M (ie. single core, no hyperthreading) and 2GB RAM, one of my smaller projects takes 15 seconds to build with nothing else much going on. When tarring up my iTunes directory, this goes to 25 seconds, ie. a 66% increase in time.
I've ordered my Intel iMac and it should be here in a couple of weeks (I ordered the 17" iMac with 2GB RAM from Apple - I know, I paid more, but it actually isn't that much more than e.g. Crucial, and I won't have to wait for the RAM to be delivered separately, and I won't have a useless 512KB RAM stick to throw into a drawer and forget about)
I'm going to do some real world benchmarks of my own, e.g. compiling while importing a CD into iTunes ... things like that.
Yes, vs. a PowerMac with Dual-core 2GHz G5
Wow, real memory or virtual?
It was real - seems not to be doing it now so im not sure what was going on but... its only using 200 mb now
This whole industry bites at....
Benchmarking. It is pretty sad.
So, what's wrong with this benchmark?
1. The applications selected have very small gains from threading.
The major aspect of the Core Duo is that it is DUAL CORE. A significant part of the gain is from the TWO processing cores.
2. The 512MB of RAM in each system.
Look, almost everyone will admit that OS X itself runs sluggish under 512MB of RAM. Yeah, it's the default memory amount. But, everyone either upgrades from Apple or upgrades third party. So, instead of testing the two systems in a somewhat even manner, let's hamstring the Intel Mac by throwing it deep into swap for Rosetta performance testing.
Why not give this a try....
1. Load both the G5 and the Intel Mac with 2GB or RAM.
2. Instead of running each of these performance tests independently....how about running the iTunes, iMovie, iPhoto, iDVD, ZIP, iSquint, and BBEdit all at the same time.
I don't know about ya'll, but I tend to have 8-12 applications going at any one time. There are a couple of ways to test.
1. Specific application performance. This would be like photoshop testing. Where the performance of a specific application is critical.
2. Overall System Performance. This is where you run a number of applications all at once, simulating a high stress but realistic operating environment.
3. Component Performance Testing. This is like the SPEC benchmarking. This is where you attempt to isolate a specific component. These tend to be the most useless from the end user perspective. There are just to many component interactions in a system for this isolated data to be that useful.
So...to further make the case for testing with more RAM.....
1. If you are doing application performance testing, at least make sure the evironment is similar to the production evironment. If you are using Photoshop in a professional capacity, are you really going to run 512MB of RAM? I hope not. You will load up with 2GB most likely (especially at current pricing). This means the benchmarks for Photoshop running on 512MB of RAM are useless. Benchmarks at 2GB or RAM would provide a reasonable representation of what you can expect in the real world, which is where it all matters.
The same goes for general system testing. I care how it will perform in a real environment, not some contrived environment that is known to be resource constrained.
Ok first and foremost I'm a mac lover I don't care what chip is powering it. Both current Intel mac models are nothing more then place holders so Apple can say look we have products and the hardware transition can begin. Until all pro apps are Intel compatible on the mac platform they are nothing more then glorified machines to run ilife.
Is anybody really surprised that a machine with two cores is faster then a machine with one running dual CPU aware apps that Apple have themselves said they have been writing Intel versions of for years? The sad fact is that compared to the G4 Apple never really optimised as well as they could have done for the G5. Whatever is said the Intel transition has been coming for years. It is common knowledge that the benchmarks Apple quoted during the Intel introduction in San Fran were based on apps compiled with top level non standardised Intel compilers http://www.tgdaily.com/2006/01/20/the_coming_change_to_mac_software/ , I highly doubt the retail versions of univeral bindary apps use this compiler so are unlikely to show the same gains. It is also unlikely that many 3rd parties will use anything other than the more main stream standardised compilers which currently just do not provide the necessary optimisations for multiple CPU machines.
Rosetta may be a technical wonder but it is next to useless in the real world especially given the imposed RAM limit in the consumer and laptop lines. A much fairer test would have been intel imac versus dual core Powermac if a real performance test was required. Saying the new Intel macs are fast is like saying a car is quick with no fuel to run it, we need software. PPC development for Apple has stopped and has done for months now so Intel's latest and greatest are bound to be quicker certainly compared to the G4.
An interesting interview with AMD vice president http://www.tgdaily.com/2006/01/20/tgdaily_interviews_amd_q1_2006/ states he doesn't believe Intel will be competitve with their products until 2008. Granted some of this will be hype but until Intel move away from the FSB architecture and go for direct connections with onboard memory controllers they will be second best.
As Theo Gray said when the G5 was introduced Apple were competing with Unix based number crunchers. By Apple's own admission the G5 did not excel at integer but it flew at floating point. The G5 actually made Apple competitive again so much so that they found new markets in science and medicine. Moving to Intel so far is a move to compete in the mass consumer market. We can safely say very few Intel macs will find there way in to supercomputer installations. Judging by the current price points of these Intel chips it is likely the profit margins will come down as well although R&D should be considerably cheaper.
I would suggest that an AMD vice president is perhaps just maybe not entirely objective when comparing Intel and AMD chips.
The prediction of theregister is that Intel is ahead in the notebook space right now, that Intel will be ahead in the desktop space in about July/August when Merom appears, and that AMD has a chance to mostly catch up with Intel by the end of the year.
I would agree with that for the most part. Intel's laptop lineup is formidable but AMD's offering is very under rated and a dual core version can't be far off. I think it is pretty amazing that a company the size of Intel has to play catch up at all, guess that is what happens when you make one mistake in this game. The main thing for Intel is to kill the power requirements of their desktop chips only then can they build on the performance, I think AMD will be able to match whatever Merom and future desktop CPU's offer in the short to medium run.
It is of course a mute point as Apple announced a switch to Intel not sadly x86/x64 in general