[giantmike]

You will not need to buy new software due to panther. Panther will have some more 64-bit code, and some G5 optimizations, but it will not be fully 64-bit. In fact, that is the main selling point of OS X on the IBM 970. The 970 can run 64-bit and 32-bit applications (the OS is a big application) at native (full) speeds. Therefore, you will not need seperate 32-bit programs and 64-bit programs.

This is different from the Windows side of things, where the 64-bit Windows XP actually only runs 64-bit programs.

Also, 64-bit programs will not be faster on the G5 just because they are 64-bit. They will be able to access memory in larger chucks, which could potentially help where you consistantly use files that are greater than 4 GB in size. The biggest speed improvements will come from optimizations, not a move to 64-bit applications.

[juanpacolopez]

Originally posted by giantmike:
This is different from the Windows side of things, where the 64-bit Windows XP actually only runs 64-bit programs.

Also, 64-bit programs will not be faster on the G5 just because they are 64-bit. They will be able to access memory in larger chucks, which could potentially help where you consistantly use files that are greater than 4 GB in size. The biggest speed improvements will come from optimizations, not a move to 64-bit applications.

This is simply not true.

Firstly, 64-bit XP on an Opteron will run 32 bit apps no problem.

Second, and more importantly; Apps redesigned/recompiled to use a 64 bit data path WILL see a rather large performance gain in some cases. It doesn't mean 64 bit apps will be twice as fast or anything so simple...

Look at it this way, "64 bit" refers to the "width" of the data path within the processor. Using 64 bit code, coupled with the excellent (and in the case of the 2Ghz monster ASTOUNDING) bus on the G5 will show huge gains in performance for tasks that require performing calculations over and over on large chunks of data repeatedly (such as video/photo/sound editing and yes, even 3d games).

The G5 DOES run 32-bit code at full speed... meaning that it allows processes to operate in 32-bit mode natively. That doesn't mean that a process taking full advantage of the 64-bit data path won't see a speed improvement, it just means that running 32-bit code isn't an emulated process that makes it SLOWER than it would be on 32-bit processor.

Once we see a full 64-bit version of OSX (will 10.3.0 be "it" or will it be a later service release?.... ADC select folks care to chime in?) and apps redesigned/(mostly)recompiled to take advantage of the "64-bitness" we'll see great things indeed.

The much-ballyhooed eweek article floating around agrees with me I'd link to it but it's already been linked what, 4 times in this thread? heh.

[awcopus]

Correct me if I'm wrong, but isn't this going to come down to how well software is written to take advantage of the G5 processor and architecture?

Clearly Panther will kick ass. But will Adobe (Photoshop, After Effects), Macromedia (Director), Avid...

Case in point: Adobe After Effects is specifically SUB-OPTIMAL on a duallie Mac compared to the PC. Other apps, like Combustion, which are written to take full advantage of the dual G4 architecture actually rendered at rates comparable to what one would get on a high-end PC (3+ GHz).

Seems to me that Apple and IBM have done their job, but others have to write code diligently for us to experience it.

[kupan787]

Originally posted by juanpacolopez:
Once we see a full 64-bit version of OSX (will 10.3.0 be "it" or will it be a later service release?.... ADC select folks care to chime in?)

Panther is not a full 64 bit OS, or else it wouldn't be abel to run on a G4. This is a good thing that it is not.

Panther, however, can take full advantage of the G5. So it is a bit of a hybrid, parts of it are 64 bit that need to be, and parts that don't need to be, aren't.

In many cases, a 32 bit app ported to 64 bit would run SLOWER. Why? Because it has to push twice as much data around (64bits vs 32bits), and if it only needs to push around 32bits (not dealing with numbers over 4 billion for example), it is doing extra work it doens't need (I hope I got that right, I have heard it enough I am tryign to repeat.) 32 bit apps will be around for some time to come, as most programs simply do not need all of what 64 bitness has to offer.

[moki]

Originally posted by Eug Wanker:
So you compile with GCC 3.3 instead of CodeWarrior? If that's the case, then you might be able to use XLC, correct? Would a speed boost be of use for your games or utilities?

It depends on the project. For anything that is Cocoa (such as WireTap), yes, we use gcc. Snapz Pro X also uses gcc, because it is a mach-o nib-based Carbon application. Most of our games are CodeWarrior.

As for xlc, I have downloaded and installed it, and printed out the docs, but haven't gotten any further with it.

[moki]

Originally posted by awcopus:
Correct me if I'm wrong, but isn't this going to come down to how well software is written to take advantage of the G5 processor and architecture?

Seems to me that Apple and IBM have done their job, but others have to write code diligently for us to experience it.

That's essentially right, sure. But even non-G5 optimized apps are going to do quite well on the G5, especially when running on Panther.

[Brazuca]

Originally posted by moki:
It depends on the project. For anything that is Cocoa (such as WireTap), yes, we use gcc. Snapz Pro X also uses gcc, because it is a mach-o nib-based Carbon application. Most of our games are CodeWarrior.

As for xlc, I have downloaded and installed it, and printed out the docs, but haven't gotten any further with it.

Did you see this Ars thread?

People are seeing significant improvements over gcc.

[juanpacolopez]

Originally posted by kupan787:
Panther is not a full 64 bit OS, or else it wouldn't be abel to run on a G4. This is a good thing that it is not.

Panther, however, can take full advantage of the G5. So it is a bit of a hybrid, parts of it are 64 bit that need to be, and parts that don't need to be, aren't.

In many cases, a 32 bit app ported to 64 bit would run SLOWER. Why? Because it has to push twice as much data around (64bits vs 32bits), and if it only needs to push around 32bits (not dealing with numbers over 4 billion for example), it is doing extra work it doens't need (I hope I got that right, I have heard it enough I am tryign to repeat.) 32 bit apps will be around for some time to come, as most programs simply do not need all of what 64 bitness has to offer.

Firstly, when I asked about Panther I was working under the assumption that Apple will develop/release a specialized version of whatever 64-bit OS (Panther or otherwise) specifically for the G5. Kind of like there's a 32 and 64-bit Windows XP.

Also, just because a processor is 64-bit doesn't mean it HAS to "work on" a single piece of data that is exactly 64-bits. My understanding of the dispatch unit in the G5 is that it takes the full 64-bit data path (which can be divided, for example, into two sets of data/instructions 32-bits long each) and splits it off according to:

1) priority (essentially it's good at "guessing" what instructions are more important, to an extent)
2) branch prediction (if there is any)
3) type of operation (int, fp, vec)

From there the data/instructions are diverted to the individual execution units. In the case of the vector unit (altivec) there's even ANOTHER dispatch unit that functions in a similar way, with an issue queue of vector instructions building up/being dispersed as needed. This is part of the reason that the G5's Altivec is considered inferior... though I'm not entirely convinced it is, I just think it will take a change in paradigm for people to optimize for the chip properly.

As an example of how a 64-bit processor could hugely speed up REAL WORLD performance NOW let's turn to FCP.

When you render video in FCP, the actual encoding of the video is essentially applying the same compression algorithm to a VERY large amount of data. It does this in chunks, dispatching instructions to the CPU to perform the same types of operations over and over, on different locations in memory/stack. If each of these dispatches could contain a full 64-bits (instructions + data + cache) or even two sets of 32-bit instructions issued simultaneously via mutliple independent threads (the essential idea behind Hyperthreading, just w/ a 32-bit CPU and a pseudo "dual" core instead of a 64-bit CPU) the work would get done MUCH more quickly.

The point is, with enough work on making apps take advantage of the full data path available (either by massively parallel threading or using full 64-bit instructions, which would be useful perhaps for scientific calculations where the data set isn't so large but the sheer number of operations is) apps WILL see a speed improvement.

EDIT: After reading my post again I wanted to clear something up. I may have misstated my position to indicate that I believe a 64-bit execution unit could process two 32-bit instructions simultaneously... that's not the case. The performance benefit I'm referring to comes from being able to issue two 32-bit instructions and have them dispatched to their respective execution units at the same time... if they're the same type of operation (say an integer add) they'll still be executed in order... the speed up comes when you have 2 instructions of differing types dispatched/executed/processed simultaneously because the in/out flow from the processor accomodates 64 bits.

[3.1416]

Using 64 bit code, coupled with the excellent (and in the case of the 2Ghz monster ASTOUNDING) bus on the G5 will show huge gains in performance for tasks that require performing calculations over and over on large chunks of data repeatedly[/B]

These types of operations should use Altivec instead, which processes data in 128 bit chunks on both the G4 and G5.

[juanpacolopez]

Originally posted by 3.1416:
These types of operations should use Altivec instead, which processes data in 128 bit chunks on both the G4 and G5.

While that's true in many instances, read the EDIT I made above for an explanation...

Altivec (being an SIMD unit) is designed for just such a purpose, you're correct.

But there would still be a speed up, because not ALL code can be vectorized, and even when an app is heavily leveraging Altivec for a particular operation there will still be plenty of plain 'ol int/fp functions to perform (either for other programs or for OS processes). This is where you'll really start to see the G5's power...

While the single 32-bit data stream of the G4 would require 4 cycles just to send 128 bits of data to Altivec, the G5 would require only 2. Not only that, but if you're sending data in 32-bit chunks ANYWAYS you could simultaneously have another 32-bit instruction issued to an entirely different execution unit. Like I said, this won't "double" the speed, as the individual execution units can still only process one thing at a time, but in cases where you want to perform miscellaneous 32 bit int/fp operations while simultaneously feeding Altivec w/ 32-bit chunks of data/instructions (it takes 4 32-bit vector instructions to fill each issue queue of a G5 if I'm not mistaken, and there are 2).

[kndonlee]

Our school just bought like 15 DUAL (2400+) 1.4ghz opterons... should be up in a couple days..they cost 4k each.. from a manufacturer called BoXX or something like that. comes with a nvidia geforce fx i think... with 2 gigs of ram...

btw, these are for maya 5... but yea...

anyone wanna send some mail to macnn (at) ghak.net on psbench directions and stuff? i'll get those numbers posted in a week or so...

[Hydra]

Originally posted by Brazuca:
Did you see this Ars thread?

People are seeing significant improvements over gcc.

I started the thread and now have trouble following it It got very technical very quickly. But It does look like XLC is gonna make things very interesting on the Mac front in the next few months.

-Jerry C.

[moki]

Originally posted by Brazuca:
Did you see this Ars thread?

People are seeing significant improvements over gcc.

Yes, I've read that thread (and recently participated in it). I'm not surprised xlc is doing better than gcc -- gcc has always been fairly poor at generating PPC code, though it has been improving with the resources IBM and Apple have contributed.

As I mentioned, I saw a nice speed boost just recompiling with gcc 3.3 compared to 3.1. Imagine if that speed boost were applied to the entire OS.

Compilers have quite a bit to do with the performance of any given code written in a higher level language. It isn't just the CPU/computer architecture.

Use a car analogy, perhaps. Pump poor quality, low octane gas into your car, and it isn't going to run as well as it could. So it is with the code a compiler generates.

[jmfr123]

You guys are not getting enough Info, Lets see now the G5's new OS will still be 32-Bit, what makes you think it can't compare to an Opteron system ? it's currently running in 32-bit Windows OS also. Second they are both work stations, Apple claims their's to be a Personal Computer and they are right, you know what a WorkStation is ? it's a more powerfull version of the regular Home personal computer, and it's a Personal computer just with more power, so a Dual Opteron vs Dual G5 is a good match. i'm going to tell you a little history, IBM is helping AMD to manufacturer the Opteron chips and research some technologies, that is why the G5 also has some equal specs to an Opteron bus system, PCI-x and PCI-Express. 8GBite barrier ? wrong, since the release of the Pentium Pro, they address 36Bits making them support up to 64GBites of memory and same goes with the G3-4 Processors, it's making the chipsets address that much memory not the processor. The Opteron and G5 have a 48Bit address making them i think address up to 256GBites, but i might be wrong on that capacity. Anyways Intel had a Prescott with 64-bits but they stoped the project, who knows why because i don't, but they are going to re release a new core with 64-bit extensions and dual 32-bit cores that make that happen, which combine to process in 64-bit chunks and will get released in february 2004, OMG! two cores working as one ? well i don't know how it will perform, but it will be like a Power4 in some way. The 1.6GHz is a weak processor, the 1.8GHz might do the job and the Dual 2.0GHz can not be compared to a single Pentium 4 , instead you use an Opteron WorkStation for that (yes it's a PowerFull Personal Computer).

[kupan787]

Ok, here is my G5 1.8 with 512MB of RAM. Note therer were some times the 1.6 G5 won. this is probably because it had 1GB of ram. If I can get more ram, I will try this again. But also interesting is that the 1.8 here scored better than the projected single 2.0 on the first page. So I think that the 2.0 numbers might have been a little too conservative. I can't wait to see those in real!

Code:
                     Ath   Ath   P4     P4   Ath   2xAth  G5   2xOpt	G5
Test                 3000  2700  3.06   2.8  2200  2200   1.6   244	1.8

1 Rotate 90           0.2   0.2   0.2   0.2   0.3   0.3   0.4   0.2	0.5
2 Rotate 9            2.8   2.9   2.7   3.3   3.5   2.4   2.7   2.4	2.6
3 Rotate .9           2.6   2.6   2.5   3.2   3.3   2.2   2.8   2.2	2.4
4 Gaussian Blur 1     0.8   0.8   0.8   0.9   1.1   1.0   0.6   0.7	0.6
5 Gaussian Blur 3.7   3.0   2.9   2.0   2.1   3.8   2.4   2.3   2.0	2.4
6 Gaussian Blur 85    3.4   3.5   2.3   2.6   4.6   2.9   3.5   2.3	5.1
7 Unsharp 50/1/0      1.0   1.1   0.9   1.2   1.3   1.0   1.1   0.8	1.1
8 Unsharp 50/3/7/0    3.1   3.1   2.1   2.4   4.0   2.5   2.8   2.2	2.6
9 Unsharp 50/10/5     3.1   3.1   2.2   2.4   4.0   2.5   3.4   2.2	3.3
10 Despeckle          2.6   2.7   2.2   4.0   3.3   1.9   0.8   2.9	0.7
11 RGB-CMYK           8.1   8.2   7.3   9.4  10.0   5.4   4.2   7.4	4.0
12 Reduce Size 60%    1.2   1.2   0.9   1.0   1.5   1.1   0.8   1.0	0.8
13 Lens Flare         3.9   4.0   2.5   3.8   4.3   3.1   6.0   4.2	5.5
14 Color Halftone     2.8   3.3   2.2   2.4   4.3   4.5   4.3   2.6	3.8
15 NTSC Colors        2.3   2.4   2.4   2.6   3.1   3.1   4.3   2.4	3.9
16 Accented Edges    10.4  10.9  10.9  12.0  13.7  13.9  16.2  12.1	14.4
17 Pointillize       17.5  17.7  12.1  18.8  21.3  12.0  25.0  20.1	21.9
18 Water Color       22.6  23.6  26.4  29.2  29.4  29.9  35.8  25.7	31.1
19 Polar Coordinates  8.0   8.3   7.0   8.5  10.2   6.1   4.9   8.0	4.4
20 Radial Blur       46.6  46.9  33.1  43.6  62.7  34.4  54.3  38.2	46.9
21 Lighting Effects   2.0   2.0   1.9   2.1   2.5   2.0   3.4   2.0	3.3

Total               148.0 151.4 124.6 155.7 192.2 134.6 179.6 141.6	161.3

[Hydra]

Originally posted by jmfr123:
You guys are not getting enough Info

Yeah, but you seem to be overstating things a bit. IBM is working with AMD but that has ZERO to do with the new G5 towers incorporating many new technologies that AMD is behind. Apple sits on the Hyper-Transport board too (IBM I believe recently signed on as well). Apple is responsible for what goes in their boxes not IBM (well, expect for the CPU and support chips being manufactured for Apple). If Apple wanted to hook up a hamster wheel drive and have it power their new G5's, IBM would have little to do with the decision.

The Opteron has been assailed as of late as not being a PC but more of a workstation (which I do agree is a matter of semantics). The problem is that until recently all dual Opteron MB's lacked an AGP slot. That IMHO is a serious blow to the argument that the Opteron is PC. How many PC's do you know of today lack this most basic feature? The Opteron is a very nice chip and in dual CPU configs is very attractive but to be honest I have played with twice as many G5's at this point as I have even seen Opterons in use (exactly 2).

As to Panther not being fully 64bit, I think you may have a bit of a wait on your hands if you think MS is gonna have a fully 64 bit OS in your lifetime. There is no reason to rewrite everything down to solitaire to take advantage of the 64 bit capability. I wonder if 64 bit heavy iron stuff that has been running for years has all 64 bit apps. The question is why would you go through all the trouble to do it when a certain app will gain nothing from it.

As far a the 1.6 chip being weak, I think the jury is still out on that one. Why would the 1.6 be weak and the 1.8 be so much better? IBM's SPEC scores show almost parity with Opterons and the PPC970. As far as multimedia stuff goes the Opteron can't really compete with Altivec anyway (one big weakness early on with the Opteron has been it's poor showing in multimedia things compared to a P4). After 10.3 and the eventual widespread use of XLC the PPC970 will really fly. I've seen reports that ALL apps under 10.3 will get a significant boost and the G5 will get a bigger boost.

-Jerry C.

[Simon]

Originally posted by Hydra:
As far a the 1.6 chip being weak, I think the jury is still out on that one. Why would the 1.6 be weak and the 1.8 be so much better?

Possibly because the difference between 1.6 and 1.8 is not just 200MHz, but also a faster, higher-bandwidth and refined motherboard...

After all, the 1.6GHz model is the low-budget config.

[MindFad]

Hmm, these machines seem to really get a good jump in performance with lots of added RAM. Personally, I bet benchmarks and apps will get a lot better -- OS X included -- when things get optimized in any ways they can for the G5. I can't wait to get ahold of Panther on one of these.

[klinux]

Originally posted by Hydra:

The Opteron has been assailed as of late as not being a PC but more of a workstation (which I do agree is a matter of semantics). The problem is that until recently all dual Opteron MB's lacked an AGP slot. That IMHO is a serious blow to the argument that the Opteron is PC. How many PC's do you know of today lack this most basic feature?

I know and see plenty of PCs without an AGP slot. Ask yourself, who really must have an AGP slot or else they cannot do they do on their computer? Graphics professionals? No, they have plenty of PCI based solution out there and most of their bottle-neck is CPU based. Gamers? Yes...

Plus, if they are hardcore enough, they would still play their games even if the vid card is "just" a Radeon 9000 or GF FX PCI card.

[Shaddim]

Here's the PS7Bench on my "new" (well it's new to me) dual 1.42GHz with 2GB of RAM.

Code:
rotate 90 - 0.5
rotate 9 - 2.2
rotate .9 - 2.0
1 pixel gausian blur - 0.6
3.7 gausian blur - 2.1
85 gausian blur - 3.1
50%/1/0 unsharp mask - 0.9
50%/3.7/0 unsharp mask - 2.5
50%/10/5 unsharp mask - 2.9
Despeckle - 0.7
rgb to cmyk mode change - 2.5
60% Reduction - 0.8
Lens Flare - 4.5
Color Halftone - 4.8
NTSC Colors - 5.1
Accent Edges - 15.2
Pointilize - 13.6
Watercolor - 34.2
Polar Coordinates - 3.8
Radial Blur - 27.5
Lighting Effects - 2.7

This is on an average of 3 runs.

[idyll]

Unfortunately yours has 2 GBs of RAM, whereas the G5 has 1 GB.

Could you take 1 GB out and re-run the tests for reference?

[Shaddim]

Would the RAM make that much difference on this test? I can see 256MB vs. 1GB, but 1GB vs 2GB? Surely PS7 (with files that small) wouldn't be using any scratch disk space with 1GB installed.

Oh well, I can try it tomorrow, I've got it doing some rendering, encoding, playing Tiger Woods 2003, and browsing the web. Gawd, I love this machine. My PB would be coughing up lung-biscuits if I tried doing all this on it at the same time.

Oh yeah, I've got SETI running too.

[Anand]

With all this testing and the results of the old DP G4 machines, could it be that, well is it possible that, do you believe that, the old G4 DP machines were pretty nice computers?

[Hydra]

Originally posted by klinux:
I know and see plenty of PCs without an AGP slot. Ask yourself, who really must have an AGP slot or else they cannot do they do on their computer? Graphics professionals? No, they have plenty of PCI based solution out there and most of their bottle-neck is CPU based. Gamers? Yes...

Thank you, you furthered my point. Gamers make up a huge portion of the Personal Computer market and without what most gamers would consider essential I think it hurts the entire case that the dual Opteron is a personal computer. AMD doesn't even consider it a chip for personal computers, that is what the soon to be introduced Athlon64 is for.

I like the Opteron and feel it is a damn fine piece of engineering ( I think it has the much bigger Intel worried a bit) a dual processor box that is gonna cost $3000 plus (conservatively) is not really a Personal Computer w/o a basic PC feature. A workstation yes, a server yes but not really a Personal Computer. Could you buy one, put it on your desk andsurf the web? Sure thing. I could do the same with a Xserve (which has an AGP slot) but it really is not exactly what it was meant to do.

The line between PC and workstation is most definitely blurred. More powerful computers for less money than most people ever dreamed of. Right now average ordinary human beings can walk into an Apple Store and walk out with a 64 bit personal computer (supplies are limited as of now of course) and soon COMPUSA, Best Buy and others as well will have G5's too. Will you be able to buy an Athlon64 in a few months at COMPUSA? Yes probably. An Opteron at a regular retail outlet, doubtful now or in the future. I just did a search on COMPUSA.com for G5, Opteron and threw in Xeon too for kicks. Guess which one of the three was the only to find ANY hits?

-Jerry C.

[Hydra]

Originally posted by Simon:
Possibly because the difference between 1.6 and 1.8 is not just 200MHz, but also a faster, higher-bandwidth and refined motherboard...

After all, the 1.6GHz model is the low-budget config.

Yup, the 1.6 G5 has lower specs on it's MB but it appears to be the same MB with a few different things on it. The RAM is not as fast but I really think the 1.6 G5 will perform pretty well. The 1.6 uses slower RAM but the CPU is slower to begin so the faster PC3200 RAM might be wasted on it. I'm not questioning that the 1.8 and 2.0 dual with be superior. I only think too many people discount the 1.6 without giving it a fair chance. Compare the bus speed/bandwidth and just about everything else on the 1.6's MB with the previous G4's and it looks mighty good. The 1.6 is far closer in specs to the dual 2.0 than any prior G4. (all this defense of the G5 1.6 from a guy with a dual 2.0 on order)

-Jerry C.

[Simon]

Originally posted by Hydra:
Compare the bus speed/bandwidth and just about everything else on the 1.6's MB with the previous G4's and it looks mighty good. The 1.6 is far closer in specs to the dual 2.0 than any prior G4.

I absolutely agree.

The problem IMHO is that people won't compare the 1.6GHz G5 with the previous expensive dual 1.42GHz G4, but with the rather inexpensive current dual 1.25GHz G4. And this dual G4 is really a swell deal.

Before the G5 arrived this board was full of people complaining how bad Apple's G4 compared to P4s with HT on 800MHz boards. Now, people are amazed to see how well the G4s compete with a 1.6GHz G5.

It's amusing to observe how fast perception changes.

But, I'm pretty certain that the 1.6GHz model will be the first to leave the line and from then on DDR400, 8 DIMM slots and PCI-X will be across the line. I suppose this reminds many people of the Yikes model when the G4 was introduced. And I'm sure that further hurts the 1.6GHz G5 model.

Before the public opinion kills the 1.6GHz completely, I believe we should wait for Panther and more updated apps. By then I'm sure it will show quite nice numbers. Nevertheless, we are comparing a single processor machine to duals. That's already quite impressive.

[aimeesteinberge]

Hey guys, really stupid question, but does anyone know how much faster the 1.6 G5 would be compared to a 500 mhz single processor G4? (for graphics type things such as Premeire and Photoshop)

just a rough estimate perhaps?

[moki]

Originally posted by aimeesteinberge:
Hey guys, really stupid question, but does anyone know how much faster the 1.6 G5 would be compared to a 500 mhz single processor G4? (for graphics type things such as Premeire and Photoshop)

just a rough estimate perhaps?

My guess, based on the stats we've seen, would be 4-5x faster.

[aimeesteinberge]

heh, I guess I really have been waiting long enough to get a new computer then, haven't I? I was hoping to start making films again, and I use photoshop and final cut all the time... if it really is like 4 or 5 times faster, I wouldnt have to let the poor machine render files overnight while I drool sleeping under my desk...

[StiZeven]

Interesting how Apple didn't use the 3.2GHz 800MHz FSB P4 chip in their comparisons. Anyone know why? It's been out for months and when you're benching high-ends, you're supposed to use the best from both platforms. Right now, they are comparing the G5 with a 3GHz P4 with a 533MHz FSB which is over 9 months old.

[blackwind]

Originally posted by StiZeven:


Interesting how Apple didn't use the 3.2GHz 800MHz FSB P4 chip in their comparisons. Anyone know why? It's been out for months and when you're benching high-ends, you're supposed to use the best from both platforms. Right now, they are comparing the G5 with a 3GHz P4 with a 533MHz FSB which is over 9 months old.

Actually, a 3-GHz Pentium 4 does run on a 800-MHz FSB. A 3.06-GHz Pentium 4 runs on a 533-MHz FSB.

As for why Apple only compares a 3.0-GHz Pentium 4C rather than a 3.2-GHz Pentium 4C, it is because Apple hired VeriTest to do the test. The 3.2-GHz Pentium 4C was released on June 23, which was the exact same date that Apple announced the G5. Since the tests were already done, Apple did not want to pay VeriTest to do SPEC tests again.

[3pDesigns]

I have a dual 2.0ghz G5 on the way with 2gb of ram and the radeon 9800 video card. could anyone tell me where i could get bench mark testing software for it. ?

[ourisman]

I am noticing a HUGE difference when *printing* a file from Photoshop. I used to use a 500 mHz G3 iBook. The iBook would take perhaps a minute to process an 8 MB file and send it to the printer. My dual 2.0 gHz G5 processes the file virtually immediately.