Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I think that these tests are poor regardless of the results. Testing is all based on evidence and I see none, just what they say are the results.

When you run a test you normally document the process for the test conditions. You don't just say Photoshop CS2 - MP aware actions, but which ones - why didn't they use the Photoshop test.

"For FCP 5, we rendered a 20 second HD clip we had imported and dropped into a sequence."

Does this mean they imported a 20 second clip into a sequence and had to render the clip before it would play with the rest of the sequence.

They basically used the render tools in the sequence menu. Why measure something like that.

I would of thought that exporting a sequence of 20 seconds to create a HD clip might have been better and timed that.
 
elmimmo said:
They are comparing a 2 generations old G5 (Dual 2,5) versus a new Intel (Quad 2,6) which is not even the fastest out there. What kind of comparison is that?

If you want to know what is the fastest Mac, the comparison is no good. If you want to know whether you should upgrade your machine, the comparison makes a lot of sense. First, the 2.66 GHz Quad has the best price/performance ratio. If you start with the 2.0 GHz, you get 666 MHz more for $300, then you get another 333 MHz for a mere $800. So if you want to upgrade, the 2.66 is _the_ machine to buy. Second, there will be much less difference between a Quad G5 and a Quad Xeon. On performance critical Rosetta applications (like Photoshop) the Quad G5 will be stronger. In that case, it doesn't matter how much stronger - you won't upgrade, that is all that matters. But if you have a dual G5, then the question whether to upgrade or not is really interesting.

And we need to know whether apps use four cores or not. In many cases, changing from two threads to four threads is very easy (that is if all the threads to the same work; it is much harder if the threads do different work), but the app uses only two threads because most machines had only two CPUs. As an example, early versions of Handbrake didn't gain anything from Quad G5s; the CPUs were 50% idle all the time. People complained, and it was changed. The same thing will happen again, especially since _all_ Mac Pros have four cores.
 
Edit: Please ignore this post, I can't count!!!

If you buy a Xeon 5160 (3.0GHz) at the moment they are £570. Apple are charging £530 to upgrade from Xeon 5150 (2.66GHz) to the Xeon 5160. Bearing in mind that you can probably sell the original 2.66Gz chip for around £300, it would be cheaper to buy the lower spec Mac Pro and upgrade yourself.

Forgive the £ for those that think in $.
 
Well the Quad is still the fastest when it comes to CS2. But man, once those are intel native that thing is going to fly at lightspeed.
 
Glassbathroom said:
If you buy a Xeon 5160 (3.0GHz) at the moment they are £570. Apple are charging £530 to upgrade from Xeon 5150 (2.66GHz) to the Xeon 5160. Bearing in mind that you can probably sell the original 2.66Gz chip for around £300, it would be cheaper to buy the lower spec Mac Pro and upgrade yourself.

Forgive the £ for those that think in $.

Aren't there 2 chips though?
 
Glassbathroom said:
Edit: Please ignore this post, I can't count!!!

If you buy a Xeon 5160 (3.0GHz) at the moment they are £570. Apple are charging £530 to upgrade from Xeon 5150 (2.66GHz) to the Xeon 5160. Bearing in mind that you can probably sell the original 2.66Gz chip for around £300, it would be cheaper to buy the lower spec Mac Pro and upgrade yourself.

It's not something I would do myself, but some enterprising dealer could easily do that. In US prices, difference between two 3.00 GHz and two 2.66 GHz chips is roughly $300, and you could sell the 2.66 GHz ones at a premium because they had "extra burn-in testing" 🙂

Seriously, the problem is getting money for the 2.66 chips.
 
Core Image change anything with CS3?

Won't Adobe use Core Image when the Universal Binaries come out? If both Quads had the same high powered graphics card, the benchmarks may show them to be the same with Core Image tasks.
 
bretm said:
You're right. I'm extremely unimpressed that the fastest xeon only days old is actually slower mhz for mhz than a G5 that is pushing 4 year old technology. Really sad.

But overall it's not. Whenever you change chips, you'll probably always find a benchmark that favors the old one. Just because one app isn't faster doesn't mean the new chip is slower.

bretm said:
But it's not faster. Slower actually than the G5 at some apps. What's everyone looking at anyway? I'm pretty unimpressed. Other than Adobe's usage of cache (AE is a cache lover and will use all of it, hence the faster performance).

But the actual xeon processors are only as fast as the G5 processors. Look at the average specs... the 2.66 machines are only a teeny bit faster than the G5s except in a few apps like filemaker. But not in the biggies like Final Cut Pro where it actually appears that mhz for mhz the G5 is a faster machine hands down!

What are you talking about? The xeon is faster in every native benchmark, the only exception is one render where the slower xeon tied the G5. If you do indeed look at the average specs, the xeons blow away the G5.

jicon said:
Looks like the Xeons got killed by the G5 in Word in their tests.

Because it's running under rosetta, ram has nothing to do with it.

Henri Gaudier said:
It's odd, seeing as Mac's are still the choice for many musicians that some kind of specs are never given that would be of interest to musicians. The released figures don't do much for me. I'd like to know the polyphony improvements say for Kontakt under both systems in Digital Performer 5.

There have been Logic benchmarks elsewhere, and they're pretty impressive. 1.4-1.5x improvements, pretty nice considering how fast the quad is already for audio plugins.
 
BornAgainMac said:
Won't Adobe use Core Image when the Universal Binaries come out? If both Quads had the same high powered graphics card, the benchmarks may show them to be the same with Core Image tasks.
Hah! Adobe can't even be bothered to make a Cocoa-native version of Photoshop on the Mac. They won't use Core Image because it's an OS X-only technology which can't be ported to Windows without them having to (essentially) write their own framework to mimic its functionality.
 
BornAgainMac said:
Won't Adobe use Core Image when the Universal Binaries come out? If both Quads had the same high powered graphics card, the benchmarks may show them to be the same with Core Image tasks.
doubt it. because having core image would mean a totally seperate windows version. developing 2 totally different codebases would take forever.
 
BornAgainMac said:
I don't like Adobe anymore. 😡
I dunno, I mean, I guess they could use core image, but really. CS3 will probably have PS, Illustrator, ID, FW, Flash, DW, CF, Contribute, Bridge and Acrobat. That's 10 apps. Would you want to develop 20 apps, just so that Mac users could have Core Image? Until competitors come along with Core Image support, don't expect Adobe to have it.
 
I have to say, I actually expected the woodcrest results to be better. It really shows that the G5 was years ahead of the competition. 😎
 
simie said:
I think that these tests are poor regardless of the results. Testing is all based on evidence and I see none, just what they say are the results.

When you run a test you normally document the process for the test conditions. You don't just say Photoshop CS2 - MP aware actions, but which ones - why didn't they use the Photoshop test.

"For FCP 5, we rendered a 20 second HD clip we had imported and dropped into a sequence."

Does this mean they imported a 20 second clip into a sequence and had to render the clip before it would play with the rest of the sequence.

They basically used the render tools in the sequence menu. Why measure something like that.

I would of thought that exporting a sequence of 20 seconds to create a HD clip might have been better and timed that.
It doesn't matter what the tests are if you are doing it for comparison. As long as it is done the same on both machines, who cares?
 
mmmcheese said:
(sideshow bob)The Power PC...The!!!(/sideshow bob)

(silent bob)F*ck Power PC(/silent bob)


Note: OK, that reference is probably super obscure. Kevin Smith (aka Silent Bob) said in a commentary for one of his early LaserDiscs "F*ck DVD", obviously before the format took off the way it did).
 
EricNau said:
It doesn't matter what the tests are if you are doing it for comparison. As long as it is done the same on both machines, who cares?

That is wrong.

Lets say I wrote some Altivec code to make some function faster on a 400 MHz G4, because on that machine it made a noticable difference. After porting to Intel, with the slowest machine (1.66 GHz Core Solo) being at least six times faster, I didn't bother. If you measure that code, you won't find too much difference in speed. It is the code that matters that matters.
 
FF_productions said:
Check it out!

http://barefeats.com/quad06.html


The 3 ghz Mac Pro is neck and neck with the G5 Quad in the Adobe benchmarks, sick considering the fact it's running under rosetta!!

when cs 3 comes out, which will be sometime in the spring of 2007 according to macworld magazine, the mac pro will be "hands down" the best machine across the board on "all" benchmarks concerning adobe software

let's hope we get cs 3 sooner rather than later in 2007 because i would hate to wait until late march

pc world, september issue, mentioned amd's plan for a quad core processor in 2007 and if that happens, some pc box will be faster than our best xeon powered machines...that is, he he, unless we get that quad core K8L amd with their 4x4 motherboard architecture which would enable a desktop to run two quads for a total of 8 amd cores (but the price of such a machine will debut at a very high price and probably won't directly compete with the mac pro)

but for now, apple has the best pro desktop machine dollar for dollar that i have seen and with cs 3 next year, it will be a designer's dream machine better than anything out there in its price range...at least for a few months 😉
 
iGary said:
I would have thought that the Final Cut Pro benchmark would have really blown away the G5 - not so much, right?

Awesome on FileMaker and I can't wait to see how this stuff runs Adobe PS Natively.

The vague manner in which they described the test, it seems like this may have been more of an I/O problem than a processing one. Can't say for sure.
 
jicon said:
When playing a game on a PC, you have DirectX to take full advantage of the hardware, and your processor is usually tagged consuming any and all cycles it can for the game. On a Mac, multithreading, and sharing the processor among apps seems to be the flow of the computing experience.
You should really do deeper analysis/research before making generally incorrect statements like the above.
 
63dot said:
pc world, september issue, mentioned amd's plan for a quad core processor in 2007 and if that happens, some pc box will be faster than our best xeon powered machines...that is, he he, unless we get that quad core K8L amd with their 4x4 motherboard architecture which would enable a desktop to run two quads for a total of 8 amd cores (but the price of such a machine will debut at a very high price and probably won't directly compete with the mac pro)


Um....that's why intel has quad core chips coming out...starting in *2006*
On the Xeon side, Clovertown, on the consumer side, kentsfield. Sometime in the first half of 2007 I believe we'll see Tigerton, which will be an even more formidable quad core xeon, capable of more than 2 processor configurations- so if apple gets a 3 socket logic board, or a 4 socket one, we could have 12 or 16 cores.
 
Lots of stuff on Anandtech about the poor memory performance on the Intel chipset.

Looks like the Xeons got killed by the G5 in Word in their tests.
Might be an interesting machine when/if the motherboard chipset/ memory performance issue is looked in to.

I think part 3 of their review will be telling, paring the machine up to XP machines in a variety of tests.

Also from the Anandtech review (the reviewers conclusion actually):

The Mac Pro is pretty much everything the PowerMac G5 should have been. It's cooler, quieter, faster, has more expansion and it gives you more for your value than the older systems ever could.

From a performance standpoint, running OS X, the Mac Pro is truly Apple's fastest system by a long shot. Some of the performance advantages over the PowerMac G5 aren't enormous, but then you look at situations like iPhoto, Xcode or Final Cut Pro where the G5 is just put to shame. Rosetta performance is just about as good as it gets, the only real solution to that problem is for Adobe and Microsoft to hurry up and release updated software. Unfortunately since Apple isn't really a favorite of either company, it's not like greater than usual amounts of resources are being thrown at releasing new products specifically for the Mac platform.


The Core 2 also seems to be far faster than the G5 in the FCP comparison of this review. The 2.66 Ghz Pro (with the dual cores disabled to ensure balance) performed 60% better than the dual G5 system. Even the 2 GHz Core 2 model outperformed the G5 by a wide margin (and judging from the fact that the dual and quad configs had the same performance, I would wager that having quad cores makes no difference in this benchmark).

But it's not faster. Slower actually than the G5 at some apps. What's everyone looking at anyway? I'm pretty unimpressed. Other than Adobe's usage of cache (AE is a cache lover and will use all of it, hence the faster performance).

The only apps it is slower in are non-universal apps that haven't been optimized, once these become universal, you can expect it to be FAR faster than the G5 (seeing how it evens out or beats the G5 already).

But the actual xeon processors are only as fast as the G5 processors. Look at the average specs... the 2.66 machines are only a teeny bit faster than the G5s except in a few apps like filemaker. But not in the biggies like Final Cut Pro where it actually appears that mhz for mhz the G5 is a faster machine hands down!

That would be a very biased and inaccurate analysis of the comparison. Firstly, only a few benchmarks were used, its useless to discount all the benchmarks where the Core 2 are far faster and than put value only on the ONE benchmark where it is a bit faster (and even that particular benchmark is doubtful in accuracy given the fact that other reviews place the Core 2 as FAR faster in FCP) without any real reason. Secondly, the MHz to MHz comparison is useless as frankly, the core 2 architecture can scale far higher than the G5 as far as clock speeds go. Even the 3 GHz specification is barely pushing it and it shows in the fact of the significantly reduced power draw and heat dissipation in comparison to the G5.

I have to say, I actually expected the woodcrest results to be better. It really shows that the G5 was years ahead of the competition.

Besides FCP, Woodcrest beat the G5 by 30-80+% on every universal benchmark, thats a VERY wide margin when comparing ANY architecture over the past 2-3 years (just look at how much the performance of the G5 has developed over the years, upgrading to woodcrest would be comparable to upgrading to a 3.3-4 GHz G5). Barefeats own past comparisons showed that the 2.6 GHz Opteron was faster than the G5 in the majority of the benchmarks. I think at most, it can be argued the G5 was even with its competition.

Does anyone seriously believe games today will show any significant improvement on a Mac Pro?

There will be MASSIVE improvements over the G5 powermacs certainly. 😀

It was just the performance was dam quick I just wasn't sure if there was an Intel version out or not, either way that is killer performance.

Yep, or right now, you could also run a PC version of CS2. As far as relative performance goes, a 2.6 GHz Opteron system beats the 2.5 GHz G5 quad by about ~16% in the barefeats photoshop comparison, a 3 GHz Core 2 beats a 2.8 GHz Athlon 64 by 20+%. Hardware-wise, there really is no comparison.

Summary:
I think everyone has to remember that Apple took a HUGE PR risk by switching to intel and that it would be foolish to think that they didn't have a VERY GOOD REASON for doing it. As much as RISC is loved here, there really is no compelling reason to think that the G5 architecture stands much of a chance in this comparison. Furthermore, it is foolish to assume that a "up-to-date" G5 would fare any better, firstly because IBM has never stopped developing the G5 (its primary usage was IBM blade servers after all) and secondly because the Core 2 architecture as it stands now isn't being pushed to perform at its maximum levels. In the end, arrogance and pride has never helped anyone, its time to let go.
 
2GHz Mac Pro Is 25% Slower And 8.75% Cheaper Than The 2.66GHz - 160GB HD Model

Danksi said:
My main interest is in FCP the FCP results.

On a fixed budget, does anyone know the advantage/disadvantage of going for the 2.0Ghz with 1900XT over 2.6Ghz with the std video card?
The 2GHz Mac Pro is 25% slower while the price is only 8.75% lower when you also lower the order $90 by making the HD a 160 instead of the 250 stock. 300GB SATA/300 drives are only $80 now. So I think it isn't really worth ordering the 2GHz model for that much less power for that little less money. For most of Final Cut Pro work you will not need the fancy video card. Motion is tne only thing that ATI card will help with. If you won't be using Motion a lot, the stock Mac Pro card will be fine.

The 3GHz model is just the opposite - 12.78% more power for 33% more money. Time is money has to be the reason for ordering the 3GHz model.
 
Silentwave said:
so if apple gets a 3 socket logic board, or a 4 socket one, we could have 12 or 16 cores.

now we are talking...processors get me so horny 🙂

i used to go bug my friend who worked in the field, in his past life, soldering very small widgets and thingies on motherboards and processors in the 80s and early 90s...he burned out and became a private investigator for way less money than an electrical engineer in the valley...but way more exciting since he gets to carry a gun (can anybody say midlife crisis?)

actually, my love of processors was not that great...i dropped out of a phd program in computer engineering specializing in mass networking equipment processors and chipsets...but those are in a totally different price range...and there are some exciting ideas in the world of processing using water molecules and string theory, but that's way out there right now

anyway, for my normal daily uses here at home, i am eyeing the 17" inch imac and that would actually be the best machine for me, dollar for dollar, and a truly fine machine to replace my five year old power mac
 
Dual Clovertown Option Will Be On Mac Pro BTO Page By January

63dot said:
when cs 3 comes out, which will be sometime in the spring of 2007 according to macworld magazine, the mac pro will be "hands down" the best machine across the board on "all" benchmarks concerning adobe software

let's hope we get cs 3 sooner rather than later in 2007 because i would hate to wait until late march

pc world, september issue, mentioned amd's plan for a quad core processor in 2007 and if that happens, some pc box will be faster than our best xeon powered machines...that is, he he, unless we get that quad core K8L amd with their 4x4 motherboard architecture which would enable a desktop to run two quads for a total of 8 amd cores (but the price of such a machine will debut at a very high price and probably won't directly compete with the mac pro)

but for now, apple has the best pro desktop machine dollar for dollar that i have seen and with cs 3 next year, it will be a designer's dream machine better than anything out there in its price range...at least for a few months 😉
I think Apple will add a Dual Clovertown processor option to the Mac Pro BTO page as soon as they can get them. I'm thinking it will be about a $1k option - Minus Two Woodcrests Plus Two Clovertowns = about $1k I thiink.
Silentwave said:
Um....that's why intel has quad core chips coming out...starting in *2006*
On the Xeon side, Clovertown, on the consumer side, Kentsfield. Sometime in the first half of 2007 I believe we'll see Tigerton, which will be an even more formidable quad core xeon, capable of more than 2 processor configurations- so if apple gets a 3 socket logic board, or a 4 socket one, we could have 12 or 16 cores.
I'll settle for Dual Clovertown or perhaps a Quad Tigerton - if it's only $2k more - when Leopard ships on board next Spring. 🙂

So I'm thinking the Dual Clovertown OctoCore will cost about $4,000 plus ram and the Sixteen Core Tigerton Setup about $6,000 plus ram. Is that a fair guess?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.