More Intel iMac Benchmarks (Native and Rosetta)

[andiwm2003]

Note also that MacCentral loaded each with only 512 MB of ram, which likely has a bigger impact on the Intel iMac running programs in Rosetta - much more likely to need to page out/in to VM. We all know that OS X loves ram, so to run one program on a processor running natively, while the other runs the program in emulation, neither of which has much ram, heightens the difference. That's not to say it isn't real-world, because many people may only buy the stock iMac. It would be insteresting to see if running the test on both systems having 2 gig of ram has an impact.

i also feel that rosetta maybe needs more than 512 mb ram. i can't really understand why they did the review with only 512 mb ram on both machines. for one most users will have 1 GB of ram or more anyway. and my pb got much faster after upgrading to 1.25 GB of ram. in addition it runs a lot more stable.
with ram at the lower limit i doubt the test results are really representative for both machines. well lets see what barefeats.com comes up with.

 

[AidenShaw]

Do do [sic] the math: the Quad G5 has 4x64 bits while a Quad core duo would only have 4x32 bit. So for complicated computations which are based on 64 bit, the G5 will win.

"Doo doo" describes the logic here....

The Yonah does 64-bit arithmetic - it has 64-bit floating point (like every other 32-bit CPU chip in desktops), and it has 64-bit integer support in SSE.

But that's more or less irrelevant - there are very few uses for 64-bit integers where performance is important!

Adding the bit widths is no more meaningful than trying to say that two 2GHz cores give you a 4 GHz CPU.

OSX is a 32-bit operating system, even on the G5. (Cocoa, Carbon, GUI apps can only be 32-bit.) Although the chip can do 64-bit, Apple is only using it in 32-bit mode.

A terminal app on 10.4 can use 64-bit addressing, but almost no applications actually do that - so few that Apple shipped a 10.4 update that completely disabled 64-bit. It took a while before anyone even noticed....
_____________

But, Apple should have waited and gone 64-bit only for the Intel chips.... Intel chips are faster running 64-bit (20% is a general rule of thumb).

 

[shawnce]

OSX is a 32-bit operating system, even on the G5. (Cocoa, Carbon, GUI apps can only be 32-bit.) Although the chip can do 64-bit, Apple is only using it in 32-bit mode.

Mac OS X frameworks, etc. are using 64 bit integer capabilities of the G5.

 

[MrCrowbar]

[...]
OSX is a 32-bit operating system, even on the G5. (Cocoa, Carbon, GUI apps can only be 32-bit.) Although the chip can do 64-bit, Apple is only using it in 32-bit mode.

A terminal app on 10.4 can use 64-bit addressing, but almost no applications actually do that - so few that Apple shipped a 10.4 update that completely disabled 64-bit. It took a while before anyone even noticed....
[...]

Well, I know, you can't say double bits = double power. But I've done some assembler programming for a 64 bit mashine and know the advantages of larger words. Didn't know OSX actually disabled that... Hope they let the intels go 64 bits.

 

[bankshot]

one test coming in slightly slower (.91x) than the G5 iMac.

That one test was iPhoto (export to files). This is most likely a very disk-heavy operation, negating any advantage in the Intel CPU. According to this post at Ars Technica, the newer iMacs (recent G5 and Intel) use Western Digital disks, while older G5 iMacs use Maxtor disks. Apparently the Maxtors are faster. I'll bet anything that's what happened here: the G5 had a Maxtor while the Intel had a Western Digital. Intel gets beat on a disk-heavy test.

 

[shawnce]

Didn't know OSX actually disabled that... Hope they let the intels go 64 bits.

It doesn't disable it for the G5... really nothing to disable on the PPC. You can use the G5's 64 bit integer capabilities in any application as of Mac OS X 10.2.8 (and related tool chain). It is just that 64 bit addressing support isn't universal in Mac OS X 10.4 (confined only to libSystem using processes, no higher-level framework... i.e. no UI).

 

[MacNemesis]

...Though the early benchmarks on Rosetta would explain my experience with PhotoShop at MWSF - it felt like I was trying to run the CS2 suite on a Pismo. Thank goodness Adobe's promised universal updates by March.

Not to be a doubting Thomas, but care to post a link substantiating Adobe's promise? I don't recall seeing such an official promise.

 

[bankshot]

Apple shipped a 10.4 update that completely disabled 64-bit. It took a while before anyone even noticed....

Really? I don't recall hearing about this. Source? I'm curious.

 

[RichCoder]

MacWorld making stuff up

When did Steve Jobs ever say that applications would be 2-times faster? This MacWorld article implies that he said exactly that. Steve actually said that was the processor performance and did not reflect other hardware such as disk, etc.

If MacWorld ran the same benchmarks that Apple did and got much slower results then they would have a story.

-rich

 

[shawnce]

Really? I don't recall hearing about this. Source? I'm curious.

A Mac OS X update that Apple shipped at one point didn't include the 64b version of libSystem so applications that linked against would fail to run.

 

[kainjow]

Really? I don't recall hearing about this. Source? I'm curious.

http://www.macminute.com/2005/08/18/apple-security-update/

 

[boombashi]

Right. So the real winner is the MacBookPro that goes from G4 directly to core duo...

You are soooo right!

From the Macworld article:

"The speed of applications running under Rosetta will be something to keep in mind, especially when it comes to the forthcoming release of the MacBook Pro. The users of that professional-level laptop are far more likely to demand serious speed from their applications; if there’s no Universal version of Photoshop available at the time, professional photographers may balk at the idea of running Photoshop at a fraction of its speed."

So what's this guys point if Rosetta runs half the speed of a G5 on a MacBookPro. If the current iMac G5 is twice as fast as the iMac Core Duo running Rosetta, and the new MacBook Pro (similar specs) is 4x faster than the G4 PowerBook, wouldn't that lead you to believe that performance should be much more tollerable (possibly close to the same as the recent G4) considering it's predecessor? Of course lets not forget Universal compiled programs!

Ignore the numbers and be happy that Rosetta performs at close to G4 speeds, I know I am. Apple will improve on Rosetta AND OS X 10.4.4 in the near future, expect an update soon. Sounds like iLife will need some updates as well, but that's typical from Apple's iLife apps.

***I don't know if any of you actually "used" Classic through OS X 10.0, but if you did you know what it takes to make something emulate to native speed. I haven't used Classic for several years, but I remember it improved with leaps and bounds in 10.1

[Edit] Not to mention you had to boot the freaking operating system to run Classic - what a pain. This transition feels like nothing compared to that, and we all survived

 

[nagromme]

Remember the reason why Rosetta running slower than native is perfecly OK as a transitional measure:

Because most people buying an Intel Mac are not sidegrading from a G5 Mac... they upgrading from a G4!

Anybody who is working now on a recent G5 and then switches to Intel with non-multiprocessing Rosetta apps and is "disappointed" by the speed is in the minority. Most of those people are keeping their G5 until a later date when it makes more sense to get a new machine.

Likewise switchers are probably not switching from a very recent fast PC. Normally you buy a new machine when your old one gets, well... old (But for people who do switch from a fast PC, hopefully they aren't running out in droves to buy non-Universal speed-intensive apps like Photoshop. Best to wait a bit on that.)

 

[MrCrowbar]

Photoshop runs ok on the G4 Powerbooks. So it will through Rosetta on the MBP. Then comes the universal version (hope it's a free update, else I will download it, I swear. Cmon, a licence for CS2 is a licence for CS2!) and we will be "thrilled with it".

Let's all wait for April when they show off the full line of MBP (I mean the small and the big one).

 

[nagromme]

Photoshop runs ok on the G4 Powerbooks. So it will through Rosetta on the MBP. Then comes the universal version (hope it's a free update, else I will download it, I swear. Cmon, a licence for CS2 is a licence for CS2!) and we will be "thrilled with it".

What has Adobe said on this? I thought we weren't expecting Universal Pshop until CS3.

 

[RichP]

Everyone quit whining about intel. The majority of computer sales is, and will be, laptop sales. Intel chips are getting more efficient and faster. The G5 and G4 are dead for useful portables. Case closed.

 

[digitalbiker]

Nice to see that once everything gets Universal Binaries, that the apps will really (or at least, should) fly. I'm guessing that the pressure is really going to be on developers this year to re-compile. Rosetta is only (hopefully) temporary for a lot of major apps, and it might turn in to a PPC "classic" environment for apps that are never re-compiled in the future - presumably at which time the overall speed of the chips on the Intel roadmap will compensate for the emulation performance hit.

Though the early benchmarks on Rosetta would explain my experience with PhotoShop at MWSF - it felt like I was trying to run the CS2 suite on a Pismo. Thank goodness Adobe's promised universal updates by March.

Where did you hear that Adobe would have Universal Binaries by March? The last thing I heard was from the CEO who said not to expect Universal Binaries until late 2006 or early 2007.

 

[aussie_geek]

This is another good reason to hang on to what you have now and update your Mac's when you are forced to do so. Anyone with a G5 of any kind should hang out for at least 6 months AFTER the whole transition has been completed. That means start looking for a new mac in mid 2007.

This way, there will be a good chance that most apps will be universal. Some Intel only ones will start popping up too.

aussie_geek

 

[JackAxe]

The Yonah does 64-bit arithmetic - it has 64-bit floating point (like every other 32-bit CPU chip in desktops), and it has 64-bit integer support in SSE.

Ok, if you're only refering to SSE 3, then yes, but the Yohan when it comes to Integer and FPU is only 32-bit. Intel's 64-bit version is the Merom, which will be out later this year. I guess you forgort that AltiVec is 128-bit. It will be sad to see this go, if Intel doesn't come up with a comparable alternative. Hopefully the Conroe will compensate enough in other areas to off balance this downgrade, which it should without a problem.

But that's more or less irrelevant - there are very few uses for 64-bit integers where performance is important!

Yep, in the consumer arena. But for the peeps that do need it, 32-bit falls way short. I do lots of heavy rendering through Maya, to the point that it keeps my G5 and 2 other PCs busy for hours at a time. 64-bit rendering could easily mean the difference between hours and minutes.

OSX is a 32-bit operating system, even on the G5. (Cocoa, Carbon, GUI apps can only be 32-bit.) Although the chip can do 64-bit, Apple is only using it in 32-bit mode.

Leaving the GUI at 32-bit is a good thing, since not everything needs 64-bit and switching it over now would break practically everything. I would like to have it as an option in the near future though.

A terminal app on 10.4 can use 64-bit addressing, but almost no applications actually do that - so few that Apple shipped a 10.4 update that completely disabled 64-bit. It took a while before anyone even noticed....

Funny, the apps I use daily are all benifiting from 64-bit. CS2 and Apple's pro apps all support 64-bit addressing. Even my other apps, which are still 32-bit, run much better now, since they can each have their own 2 gig address space, if available. After Effects showed the biggest gaing by far, since it no longer goes into a swap frenzy when it hits its celing. (I look forward to AE 7, I'm pretty sure it will suport 64-bit addressing, if it does, I'm adding 4 more gigs.)

If Apple did disable 64-bit with one of its updates, I would like ot see a link stating that they had? I upgraded to Tiger day one and my system still saw and used all of my 5-gigs, with every update. So I'm not sure what you're refereing to, unless they temporarily disabled 64-bit addressing for applications, which was new to Tiger.

Anyways, it's almost time to upgrade my Ti-Book. If I ended up with a 32-bit MacBook, I would still be "extremely" happy.

<]=)

 

[SiliconAddict]

And quieter, too. Remember folks, lack of fan noise is more important than pure speed.

Umm NO. No^50

 

[swingerofbirch]

Count me among the nonplussed masses.

I shall wait for the SuperMac. Or perchance to dream, the SuperMacBook Pro Extreme Ultra Homemakover Edition. Yes, that is what I need.

 

[Ulfhednar]

512mbs of memory??

Heck, why didn't they just swap out that chip with 128mbs. Let's really have some fun!

. . . let's see . . . who do I know that runs OSX Tiger with 512mbs of memory . . . hmmmmm . . . ummmmm . . . gee, all I can think of is those idiots at Macworld! What a great real-world test! Great job guys! I'm so excited! I can't what to open up my MacIntel and rip out all that memory I ordered it with. What in the world was I thinking?

 

[SiliconAddict]

Don't forgot about VMware, which is quite popular on current x86 machines.

VMWare workstation kicks the SNOT out of VPC. I have both on my now defunct desktop and I can say without hesitation the feature set on the VMWW is just plain better then VPC. Might I suggest a campaign to blanket VMWare with requests?

 

[AidenShaw]

Really? I don't recall hearing about this. Source? I'm curious.

No problem!

http://www.macworld.com/news/2005/08/17/64bit/index.php?pf=1

"A security update released on Monday by Apple rendered 64-bit optimized applications for Mac OS X unusable.
...
“Due to an error on the part of Apple, this update prevents any 64-bit-native application from running,” said Wolfram in the note to customers. “In particular, this means that Mathematica 5.2 will not run on any G5 system if it has installed this Security Update.”"

Mac OS X frameworks, etc. are using 64 bit integer capabilities of the G5.

Using 64-bit integers is minor - SSE has 64-bit integers.

"32-bit mode" means 32-bit addressing.

Floating point has been 64-bit on every desktop chip in the last 20 years - but nobody has called those 64-bit chips. Doing 64-bit integers is no different.

It's only 64-bit addressing that most people use to define a 64-bit system - and Apple's 64-bit implementation is very lame.

Ok, if you're only refering to SSE 3, then yes, but the Yohan when it comes to Integer and FPU is only 32-bit.

The FPU is 64-bit, and has been on every Pentium. Where does this crap come from?

And it's not SSE3 - SSE (before SSE2 and way before SSE3) had 64-bit integer support (http://arstechnica.com/articles/paedia/cpu/pentium-2.ars/3) It was improved in SSE2 and SSE3, but it was present in SSE. (And to a limited extent in MMX, but too limited for me to claim that it really supported 64-bit integers.)

I guess you forgort that AltiVec is 128-bit.

So is SSE. Your point?

64-bit rendering could easily mean the difference between hours and minutes.

I was unable to find any posted benchmarks comparing Maya 32-bit to Maya 64-bit on the same hardware.

I would be very surprised to see a 60-fold improvement (hours to minutes), however.

Leaving the GUI at 32-bit is a good thing, since not everything needs 64-bit and switching it over now would break practically everything.

You should try Windows 64-bit then, all the old Windows 32-bit applications run just fine.

The 32-bit applications run as fast as on a 32-bit system, and 64-bit applications run even faster (typically 20% faster than 32-bit applications on the same hardware).

CS2 and Apple's pro apps all support 64-bit addressing.

Umm, CS2 is a GUI app - it's completely 32-bit. Same with Apple's apps.

(Don't you think that someone would have noticed that an Apple security update broke Photoshop and all of Apple's own apps?)

This claim is 100% BS, maybe 200% if you use 64-bit.

Even my other apps, which are still 32-bit, run much better now, since they can each have their own 2 gig address space, if available.

Just like any other 32-bit virtual memory OS. (Windows 32-bit supports up to 64 GiB of RAM - 32 applications can each have their own 2 GiB of physical memory.)

If Apple did disable 64-bit with one of its updates, I would like ot see a link stating that they had? I upgraded to Tiger day one and my system still saw and used all of my 5-gigs, with every update. So I'm not sure what you're refereing to, unless they temporarily disabled 64-bit addressing for applications, which was new to Tiger.

Look above for the links - they definitely screwed up and posted an updater that killed all 64-bit apps.

But, like a Windows 32-bit system that can support 64 GiB, your 32-bit OSX system could still support 5 GiB of RAM.

A 32-bit 10.3 system, which nobody would claim had any 64-bit addressing support, could also support your 5 GiB. 'nuf said about needing 64-bit to support more than 4 GiB in a system.

 

[bhirt]

Anybody else get the feeling that M$ either has abandoned, or will be abandoning VPC in the near future? I have this eerie feeling that VPC 7.0.2 is the last version we'll see on PPC-based Macs, and perhaps may go the way of Internet Explorer and Windows Media Player...

Hopefully the rumors are true that vmware will come out for OS X. If that's true, who cares about VPC. vmware is an incredible product that i use every day, and I'd much rather give them money than M$. If i could get it on OS X i could finally ditch my linux server.