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.
PCMag's PowerMac G5 2.0GHz Review
- Thread starter MacBytes
- Start date
- Sort by reaction score
Good stuff I say... it'd be interesting to see if, as other people have asked here as to whether they changed the energy save setting to manual and highest performance rather than automatic.
I think it's a good show from the G5 on it's first outing though, considering that technically it is crippled by an un-optimized OS and software, and there appears to be discrepancies on the PC results.
Once Apple ships Panther as de-facto and software companies get their products optimized then the performance will increase by significant amounts over the current specs.
What I'd be interested to see if multi-task benchmarks, carry out several tasks at once and get performance figures from that.
I and all of the creatives I know spend a huge amount of time multi-tasking on projects... and after suffering having to do some AE work on a PC last week, I can honestly say that from my experience, Windows can't heavily multi-task for sh*t!!
As i_b_joshua said previously though, it's quite a positive review from a PC Mag, and that can only be a good thing for the Mac community as a whole.
Wicked!!
I think it's a good show from the G5 on it's first outing though, considering that technically it is crippled by an un-optimized OS and software, and there appears to be discrepancies on the PC results.
Once Apple ships Panther as de-facto and software companies get their products optimized then the performance will increase by significant amounts over the current specs.
What I'd be interested to see if multi-task benchmarks, carry out several tasks at once and get performance figures from that.
I and all of the creatives I know spend a huge amount of time multi-tasking on projects... and after suffering having to do some AE work on a PC last week, I can honestly say that from my experience, Windows can't heavily multi-task for sh*t!!
As i_b_joshua said previously though, it's quite a positive review from a PC Mag, and that can only be a good thing for the Mac community as a whole.
Wicked!!
Next time, do a little comparison yourself first....
I'm patently a Machead but I pride myself on staying pretty unbiased.
You can get a Dell PWS 450 with dual 3.06 Xeons, 2 HDs (one at 120GB and the other at 40GB -- there's no 160GB ATA option and it wouldn't be fair to pay for a SCSI for the Dell), 512MB of RAM, and a 64MB video card (the nVidia Quadro NVS 280, dual monitor VGA) for $3314 from Dell. I just configured it on their web site.
So I don't think that one is "smoking" the other -- much more so that Macs are, for the first time since the introduction of the G3s about 5 years ago, price competitive.
The Mac couldn't use 8 GiB of RAM
Completely fair - since Mac OS X is a 32-bit operating system that wouldn't let Photoshop use the extra RAM !
By the same logic, however, on the Intel side they should have tested a quad processor Xeon MP with the 2 MiB L3 cache and 16 GiB of RAM. Why cripple the PC just because the Mac doesn't have a quad?
Completely fair - since Mac OS X is a 32-bit operating system that wouldn't let Photoshop use the extra RAM !
By the same logic, however, on the Intel side they should have tested a quad processor Xeon MP with the 2 MiB L3 cache and 16 GiB of RAM. Why cripple the PC just because the Mac doesn't have a quad?
Re: Re: Re: How are you configuring this Dell???
Well, keep in mind that on tasks that make heavy use of Altivec, the G5 is not much faster clock for clock than the G4. Its only advantage for Altivec tasks is the high-bandwidth bus; the longer pipelines hurt it, and the G5 Altivec unit itself is actually slightly inferior to the MPC 745x Altivec unit (it uses the MPC 7400 Altivec unit, the one on the original G4). But this isn't a very bad thing, because remember that the G4's vector performance was already outstanding (this was in the only area in which the G4 could consistently match or beat the P4).
Well, keep in mind that on tasks that make heavy use of Altivec, the G5 is not much faster clock for clock than the G4. Its only advantage for Altivec tasks is the high-bandwidth bus; the longer pipelines hurt it, and the G5 Altivec unit itself is actually slightly inferior to the MPC 745x Altivec unit (it uses the MPC 7400 Altivec unit, the one on the original G4). But this isn't a very bad thing, because remember that the G4's vector performance was already outstanding (this was in the only area in which the G4 could consistently match or beat the P4).
I think its ridiculous not to include loading the dialog box as time in performing the filter! That said, the G5 performance is excellent.
Get ready for even better test results, as Steve has promise 3GZ speed by mid 2004. Thats a full 1.5 increase in raw speed! Anyone think the zeon be at 4.8 GZ by next summer?
Get ready for even better test results, as Steve has promise 3GZ speed by mid 2004. Thats a full 1.5 increase in raw speed! Anyone think the zeon be at 4.8 GZ by next summer?
The last intel cpu road map I saw showed a 4ghz next generation chip (pentium 5? - haha) with 1mb cache and 1066 mhz bus towards the end of 2004. Intel just annouced a desktop p4 with 2mb of cache so I would assume they will probably keep this available.
I currently think the (single)2ghz g5 is about equal to a 3 ghz p4 so if they have a 3ghz G5 out by the end of summer 2004 a p4 would have to be about 4.5 ghz to be comparable.
There are also strong rumors that intels next chip will support x86 64 (amd's 64 bit extensions).
yeah, their photoshop test is highly suspicious. Instead of running a script to run a suite of tests en mass, what they did is to select a filter, wait for the dialogue box to fully open (which they admit took up to a minute on certain tests), and then run the test. A more accurate photoshop test is the PS7Bench which showed the G5 clearly as the top dog. And there are other issues like Energy Saver settings that could've affected the results.
But it's nice to know that in a set-up that's stacked clearly against it, the G5 comes out a big-time winner.
Re: The Mac couldn't use 8 GiB of RAM
Hey all I'm saying is that I'd like to see real tests
I didn't invent the G5 nor do I own stock in Apple, no need to get snotty
I wouldn't mind seeing the quad either, what I'm saying is load the machines up with max ram
and run similar, simulataneous apps on both machines so we can see what they can really do.
peace
daniel
Hey all I'm saying is that I'd like to see real tests
I didn't invent the G5 nor do I own stock in Apple, no need to get snotty
I wouldn't mind seeing the quad either, what I'm saying is load the machines up with max ram
and run similar, simulataneous apps on both machines so we can see what they can really do.
peace
daniel
Sheesh. Step away from the RDF...
First, that PC Preferred page was so misinterpreted by the Mac fanatics. It was meant as "Do you prefer a PC? That's ok, because Photoshop works well on that too." NOT as "We think you should use a PC for Photoshop because we say the Mac sucks." Huge difference there.
Yes, Adobe was probably pissed about FCE destroying their market for Premiere, but I personally see no evidence to suggest that this affected their stance with regard to Photoshop or other products on Mac. Remember, Adobe is a large company, and different product groups may have very different goals or attitudes with regards to PC vs Mac. But I could be wrong...
Of course there's plenty of optimization that could help on both platforms, but I doubt the outcome is as certain as you suggest. I imagine both could be sped up to some degree, and it's hard to know what that amount is when it hasn't been completed!
And yes, Adobe has more incentive to optimize for PC since that represent a larger percentage of their market these days. It just makes financial sense. But that doesn't necessarily mean they spend that much more time on the PC version (I have no idea), and the conclusion that they must "hate" Apple and Mac users is a bit silly.
Sorry for that little rant...
That said, I like the overall message of the article. It seems very positive towards the Mac, especially when writing for a PC audience. I smell more professional switchers over the next year and more. People like my dad, who floored me recently when he said he was interested in getting a G5. He's always used Windows because his customers use that and it's what he knows. I always tried to sneak in comments about how good Macs are, but it never made sense for his business, so he never paid much attention. But nowadays he's using Photoshop a lot and he's interested in making his software cross platform, and when he heard about the G5, he took a serious interest. After taking him to the Apple Store and being very impressed with a single-proc G5 and the cinema display, I'm pretty sure he's going to get a dual-2.0 after Panther comes out. Articles like this can only mean more good stories of professional switchers...
I think it's great how IBM has a long-term business strategy for the Apple line, all I see are a bunch of huge improvements over the next few years. I havnt heard anything about AMD or Intel, but I guess i dont listen that hard. I will be getting my next computer in 2 years (have to wait 3 years on a computer since i'm a college student, lol) and i'd really like to see it wipe the floor w/ x86. It's a really exciting time to be a part of the apple community right now. And for ppl like me, ppl who were here before the G3, it's so.... it's a huge relief....
Originally posted by bankshot
Sheesh. Step away from the RDF...
First, that PC Preferred page was so misinterpreted by the Mac fanatics. It was meant as "Do you prefer a PC? That's ok, because Photoshop works well on that too." NOT as "We think you should use a PC for Photoshop because we say the Mac sucks." Huge difference there.
Yes, Adobe was probably pissed about FCE destroying their market for Premiere, but I personally see no evidence to suggest that this affected their stance with regard to Photoshop or other products on Mac. Remember, Adobe is a large company, and different product groups may have very different goals or attitudes with regards to PC vs Mac. But I could be wrong...
Of course there's plenty of optimization that could help on both platforms, but I doubt the outcome is as certain as you suggest. I imagine both could be sped up to some degree, and it's hard to know what that amount is when it hasn't been completed!
And yes, Adobe has more incentive to optimize for PC since that represent a larger percentage of their market these days. It just makes financial sense. But that doesn't necessarily mean they spend that much more time on the PC version (I have no idea), and the conclusion that they must "hate" Apple and Mac users is a bit silly.
Sorry for that little rant...
That said, I like the overall message of the article. It seems very positive towards the Mac, especially when writing for a PC audience. I smell more professional switchers over the next year and more. People like my dad, who floored me recently when he said he was interested in getting a G5. He's always used Windows because his customers use that and it's what he knows. I always tried to sneak in comments about how good Macs are, but it never made sense for his business, so he never paid much attention. But nowadays he's using Photoshop a lot and he's interested in making his software cross platform, and when he heard about the G5, he took a serious interest. After taking him to the Apple Store and being very impressed with a single-proc G5 and the cinema display, I'm pretty sure he's going to get a dual-2.0 after Panther comes out. Articles like this can only mean more good stories of professional switchers...
I'll think you'll find the situation is a little worse than one would expect...
Re: Re: wha wha?
Second, Panther's PDF performance is absolutely astounding compared to Jaguar or Acrobat.
Lastly, the G5 is 33% slower (3 GHz vs 2GHz), not 50% slower. Looking at it the other way around, the Xeon is 50% faster than the G5. We're just talking clock frequency here.
First, Panther improves memory performance by a fair margin, and Photoshop is very memory intensive, so Panther should have a positive influence on these test scores.
Second, Panther's PDF performance is absolutely astounding compared to Jaguar or Acrobat.
Lastly, the G5 is 33% slower (3 GHz vs 2GHz), not 50% slower. Looking at it the other way around, the Xeon is 50% faster than the G5. We're just talking clock frequency here.
In response to the comments along the lines of this being an unfair test, I'd tend to disagree (although I do agree that not using a script that counts the filter start time is stupid). If you ignore the testing irregularities (which at least they pointed out), the test involved several pieces of real-world cross-platform software, run on two similarly configured computers.
No, they weren't both maxed out to give the edge to the Mac because it supports more RAM, but that wasn't necessarily what they were testing--they were testing a realistic config (keep in mind that loading the G5 up with 8GB will cost a huge amount of money, and most users just won't) on files in the size range that most people are probably working with.
If you're one of the small number of people who need that 8GB of RAM right now, then you'll be loving the G5, but otherwise your speed will probably be bound by the processor and other subsystems, so that's what they tested.
So what the tests showed, at least as far as I can see, was that a well configured G5 was at least as fast as, if not faster than, a top-of-the-line Dell workstation right now, and it costs a little less. That's cool enough for me.
Of course, the other cool thing about the G5 is that it will continue to get faster over the next few months as more optimized software is released. That doesn't mean that a comparitive benchmark right now is invalid, since that optimized software isn't avalable--I and everybody else who's not a professional developer will be using 10.2.7 and mostly-unoptimized software for at least another few months. But it does mean that you're getting even better speed value for your G5 dollar than it looks like if you look at the long term.
Similar to the situation with the early G4s, before much was altivec-enhanced.
Hey, how often do you buy a computer that's going to seem faster as time goes by, instead of slower?
No, they weren't both maxed out to give the edge to the Mac because it supports more RAM, but that wasn't necessarily what they were testing--they were testing a realistic config (keep in mind that loading the G5 up with 8GB will cost a huge amount of money, and most users just won't) on files in the size range that most people are probably working with.
If you're one of the small number of people who need that 8GB of RAM right now, then you'll be loving the G5, but otherwise your speed will probably be bound by the processor and other subsystems, so that's what they tested.
So what the tests showed, at least as far as I can see, was that a well configured G5 was at least as fast as, if not faster than, a top-of-the-line Dell workstation right now, and it costs a little less. That's cool enough for me.
Of course, the other cool thing about the G5 is that it will continue to get faster over the next few months as more optimized software is released. That doesn't mean that a comparitive benchmark right now is invalid, since that optimized software isn't avalable--I and everybody else who's not a professional developer will be using 10.2.7 and mostly-unoptimized software for at least another few months. But it does mean that you're getting even better speed value for your G5 dollar than it looks like if you look at the long term.
Similar to the situation with the early G4s, before much was altivec-enhanced.
Hey, how often do you buy a computer that's going to seem faster as time goes by, instead of slower?
Re: Re: Re: Re: How are you configuring this Dell???
This isn't exactly true, and this thinking is being perpetuated, unfortunately.
The comparatively lack luster performance of the G5 Altivec was started by comments from Arstechnica's Hannibal. He gathered as much pre-release information and made the observation that the Altivec units were 'tacked on' late in the design process.
When more data was made availalbe, and especially after talking to the 970 project managers, he said he was mistaken and that, although somewhat different, the altivec implementation on the G5 didn't look nearly as bad as he once thought.
In truth, the G5 will be an Altivec monster once code is optimized for the G5. If I remember correctly, the G5 prefetches data for the SIMD units differently than the G4. If you run G4 [simd] optimized code on a G5 you are starving the registers.
Once Altivec heavy apps are fully optimized for the G5, they will run WAY faster than on the G4. I think we'll have to wait until Photoshop 8 to see what the G5 can really do.
This isn't exactly true, and this thinking is being perpetuated, unfortunately.
The comparatively lack luster performance of the G5 Altivec was started by comments from Arstechnica's Hannibal. He gathered as much pre-release information and made the observation that the Altivec units were 'tacked on' late in the design process.
When more data was made availalbe, and especially after talking to the 970 project managers, he said he was mistaken and that, although somewhat different, the altivec implementation on the G5 didn't look nearly as bad as he once thought.
In truth, the G5 will be an Altivec monster once code is optimized for the G5. If I remember correctly, the G5 prefetches data for the SIMD units differently than the G4. If you run G4 [simd] optimized code on a G5 you are starving the registers.
Once Altivec heavy apps are fully optimized for the G5, they will run WAY faster than on the G4. I think we'll have to wait until Photoshop 8 to see what the G5 can really do.
Re: Re: Re: Re: wha wha?
I'm looking forward to more tests from magazines that usually don't test stuff from Apple, the future is indeed looking bright.
Also about workstation graphics, the difference isn't as big as it used to be, both in performance and price. What you pay for is the drivers specially tweaked for programs like solidworks, where they make a huge difference.
Eh, no, not on those ones. They stopped being P3s around 1.5 GHz. P3 and G4 scale about the same in clock, G5 goes higher and P4 even more so.
I'm looking forward to more tests from magazines that usually don't test stuff from Apple, the future is indeed looking bright.
Also about workstation graphics, the difference isn't as big as it used to be, both in performance and price. What you pay for is the drivers specially tweaked for programs like solidworks, where they make a huge difference.
I´ll have 1100 of those...
Correct me if I´m wrong but doesn´t the performance table clerly state that the Dell is faster overall in the PS7 tests? I know loading the filter wasn´t concidered part of it, but still, the number crunching (filter execution) is faster on the Dell on most tests.
Time in seconds (12 tests):
G5 131.01
Dell 102.50
G4 187.47
Now, that´s half a minute faster!
Correct me if I´m wrong but doesn´t the performance table clerly state that the Dell is faster overall in the PS7 tests? I know loading the filter wasn´t concidered part of it, but still, the number crunching (filter execution) is faster on the Dell on most tests.
Time in seconds (12 tests):
G5 131.01
Dell 102.50
G4 187.47
Now, that´s half a minute faster!
Article
Not to be negative but.....my new G5 arrived, finally, last night. The machine is beautiful, quiet and fast (1.8gz) but I agree strongly with this statement from the Article from PC mag...
"Unfortunately, the G5 also ships with the standard unremarkable keyboard and one-button mouse, which look and feel more out of date with each main system update."
the Keyboard is especially toylike, though perfectly functional and the white doesn't match anymore either. As for the anachronistic 1 butt mouse I won't even take it out of the box. A shame. I do not understand why APPLE seems to blow off the peripherals, they work OK but they are not as special as the rest of the package. this was even more true with the older PowerMacs and the small original iMac-type keyboards.
FWIW
Steve
Not to be negative but.....my new G5 arrived, finally, last night. The machine is beautiful, quiet and fast (1.8gz) but I agree strongly with this statement from the Article from PC mag...
"Unfortunately, the G5 also ships with the standard unremarkable keyboard and one-button mouse, which look and feel more out of date with each main system update."
the Keyboard is especially toylike, though perfectly functional and the white doesn't match anymore either. As for the anachronistic 1 butt mouse I won't even take it out of the box. A shame. I do not understand why APPLE seems to blow off the peripherals, they work OK but they are not as special as the rest of the package. this was even more true with the older PowerMacs and the small original iMac-type keyboards.
FWIW
Steve
I guess there aren't any Lightwave users here, because PCMag's Lightwave results are completely flawed. Their numbers are half of what you see PCs and Macs are getting at www.blanos.com (Chris' Lightwave Benchmarks).
This is what I think PCMag did wrong - they didn't set the number of threads higher than 1. Lightwave requires you to manually set the number of threads, and people with dual processors set Lightwave to 4 or 8 threads. The default is 1, so those dual processors systems PCMag tried to benchmark gave very slow scores.
Very sloppy.
Their Photoshop benchmark is completely bogus. They didn't use PSBench7, they used their own list of filters, and they didn't use a few filters that are normally included that do very well on the Mac. They didn't normalize the score either. Right now, on Ars Technica, they have normalized PSBench7 results from the dual 2 Ghz G5 and it's much faster than the dual 3.06 Ghz Xeon. Just doing a bunch of filters and then adding up the score is completely idiotic. You have to normalize it.
Very sloppy.
What's the price of the Dell, PCMag? They are happy to show us how expensive Macs are in the past, but not now? Could it be because the Mac is a lot less expensive?
1 meg cache Xeons or 512K cache Xeons in that Dell?
Controls refers to the dialog that comes up when you choose the filter. So you make a selection, then you select the filter from the menu, then the "controls" come up. 1 minute to wait for that dialog to come up is an incredibly horrible thing, and PCMag would have made a bigger deal of it if that happened on the Mac.
After Effects is much faster on the G5 than the PC.
Squeeze is much faster on the G5 than the PC.
Acrobat is much faster on the G5 than the PC.
Photoshop is much faster on the G5 than the PC.
And wait until Panther comes out, which is showing huge improvements in speeds.
And this time next year we will be at 3 Ghz - 50% faster!
This is what I think PCMag did wrong - they didn't set the number of threads higher than 1. Lightwave requires you to manually set the number of threads, and people with dual processors set Lightwave to 4 or 8 threads. The default is 1, so those dual processors systems PCMag tried to benchmark gave very slow scores.
Very sloppy.
Their Photoshop benchmark is completely bogus. They didn't use PSBench7, they used their own list of filters, and they didn't use a few filters that are normally included that do very well on the Mac. They didn't normalize the score either. Right now, on Ars Technica, they have normalized PSBench7 results from the dual 2 Ghz G5 and it's much faster than the dual 3.06 Ghz Xeon. Just doing a bunch of filters and then adding up the score is completely idiotic. You have to normalize it.
Very sloppy.
What's the price of the Dell, PCMag? They are happy to show us how expensive Macs are in the past, but not now? Could it be because the Mac is a lot less expensive?
1 meg cache Xeons or 512K cache Xeons in that Dell?
Controls refers to the dialog that comes up when you choose the filter. So you make a selection, then you select the filter from the menu, then the "controls" come up. 1 minute to wait for that dialog to come up is an incredibly horrible thing, and PCMag would have made a bigger deal of it if that happened on the Mac.
After Effects is much faster on the G5 than the PC.
Squeeze is much faster on the G5 than the PC.
Acrobat is much faster on the G5 than the PC.
Photoshop is much faster on the G5 than the PC.
And wait until Panther comes out, which is showing huge improvements in speeds.
And this time next year we will be at 3 Ghz - 50% faster!