[pliny]

I've never been hot on the 2 bay layout of the G5's either, the towers lost bays from G4 to G5.

[deboerjo]

Originally posted by power142:
Just to clarify, what I was angling at was that if someone is in the market to buy a PC from a big name vendor (such as Apple, Dell, HP...) rather than finding the parts from the cheapest possible online outlets and building it for themselves, then buying a machine with comparable specifications and support as a Power Mac G5 is not always going to save such significant sums of money.

Ah, good point. I and everyone I associate with have been building our own computers for so long, buying a factory-built computer is starting to become a foriegn concept to me Even my Mac is a FrankenMac with an iMac motherboard and a bunch of PC parts.

[Spliffdaddy]

Five years later and there's nothing new to talk about.

Apple still gets spanked by peecees at everything except a few obscure and rarely-used Photoshop filters - and DNET.

Wake me up in five more years.

[polendo]

Good thing that the original post only asks about an opinion of a P4 Vs. a 1.8 G5 in photoshop...

[Eug Wanker]

Originally posted by polendo:
Good thing that the original post only asks about an opinion of a P4 Vs. a 1.8 G5 in photoshop...

A single G5 1.8 would get spanked by a 3 GHz P4 in Photoshop too. (I'd still prefer the Mac for other reasons though.)

[flypenfly]

Why exactly is the P4 a "bad" cpu? Because IPC is lower? Who cares when it can scale so so much higher as it was designed. It can be taken then that the G5 is a very poor cpu because it can't run x86 code natively. Duh, it wasn't designed to but what it was designed to do it does very well.

[deboerjo]

Originally posted by flypenfly:
Why exactly is the P4 a "bad" cpu? Because IPC is lower? Who cares when it can scale so so much higher as it was designed. It can be taken then that the G5 is a very poor cpu because it can't run x86 code natively. Duh, it wasn't designed to but what it was designed to do it does very well.

Because it's still being outperformed by competitors, despite it's scalability, and it has no room left to grow (4.0GHz has been cancelled). Even Intel has admitted it made a misstep pushing clock speed over real performance and features. Thier new mantra is "value beyond MHz" and they're pushing the Pentium M as the platform of the future. Former Intel employees have confirmed that the decision to push clock speed above all else was made by marketing, and that engineering was against the decision from the beginning. This conflict contributed to a number of engineers to leave Intel, including Bob Colwell, the head engineer of the Pentium Pro through Pentium 4. Intel's own engineers thought it's design philosophy was a mistake. Intel's own management now agrees it was a mistake. The Pentium 4 was a bad design. Or more correctly, it was a very good design built around a bad design philosophy.

[i_wolf]

Nothing is as simple as to simply imply 'it would get spanked'. A lot of the time you need to compare an optimized app on one platform with an equally optimized app on another platform.
Presently, I have yet to see any Opteron, Xeon rig in After effects do the kind of things we see being done in real time on G5. NASA have gone with a fleet of G5's for their Jet 3d project.... Dr. Craig Hunter found that when you optimized for it and used a modern compiler such as XLC performance went through the roof and was incomparable. Theo Grey was also present at this years WWDC during some of the sessions. He still believes that in its price bracket there is no other workstation that can touch the G5 in terms of performance on mathematica.
I would not argue that on unoptimized apps the 2.5 G5 (even as a single ... assuming one existed) would get 'spanked' by an Athlon 64 4000+. To argue differently as was pointed out here is 'fanboyish'. However , I believe that on most x86 apps where this would be the case you would be talking about a very VERY heavily optimized x86 app where the Athlon 64 is in its element , never mind the fact that most all x86 apps nowadays have support for SSE/2. Sloppy port syndrome springs to mind.
However again, things are changing on the Mac side. Compilers with support for the G5 thus far have been crap... (exception of XLC/XLF but even that doesn't have autovectorisation for altivec). With Tiger Apple and IBM have done a lot of heavy work with GCC. The GCC that comes with tiger has full autovectorisation for altivec, and has a new scheduler which they claim brings massive performance improvements on G5 class machines. If you ask me, the G5 hasn't run anywhere near its potential yet except in one or two rare cases, whereas the x86 equivalents are in their element right here right now. At the performance tuning session for the G5, Apple gave a few demonstrations of mistakes that people had made when they were porting their x86 Athlon 64/Xeon apps to the G5... its amazing how code that works well on one platform does not bode well for another. They are pushing heavily for developers to read adopt performance tuning strategies for the G5. Simple things that make a large difference to the G5 but you wouldn't even consider possibly with an x86 platform.
Personally I believe that there is a bright future in the G5, at least interms of performance its up there with the best and it is only going to get better and better as software begins to tap the potential.

[Eug Wanker]

Originally posted by i_wolf:
< Long post >

That's interesting, but the G5 1.8 would still get spanked by a 3.0 GHz P4 in Photoshop.

And remember, Photoshop IS heavily optimized for the Mac. It's hard for the G5 to overcome that P4's 67% advantage in clock speed. However, OS X offers some usability benefits, such as Expos� and superior colour management.

[pliny]

Alot of people on here build their on win pc's, no doubt because they like doing so and see how just shitty the pc's are made. IMO they are garbage and I wouldn't spend a dime on one.

Having said that there hasn't been any tests on here showing the 1.8 v the 3.0 p4, the one test that WAS listed had the G5 doing better than the Xeon with the editors saying it spankeded the Xeon if you incldued loading times.

And why exactly are the Apple tests showing marked improvement overthe 3.4 in 45 filters, being discounted?

[Eug Wanker]

Originally posted by pliny:
Alot of people on here build their on win pc's, no doubt because they like doing so and see how just shitty the pc's are made. IMO they are garbage and I wouldn't spend a dime on one.

Having said that there hasn't been any tests on here showing the 1.8 v the 3.0 p4, the one test that WAS listed had the G5 doing better than the Xeon with the editors saying it spankeded the Xeon if you incldued loading times.

And why exactly are the Apple tests showing marked improvement overthe 3.4 in 45 filters, being discounted?

You can't seriously believe Apple is going to post stuff that makes them look bad? This is the same company that created a rating system for the store... and then rated all its products 5 out of 5.

Anyways, I believe one way Apple makes the G5 look good is by putting in a lot of RGB->CMYK. Now, this IS used by some people, but then again, some other people (assuming they use it all) don't use it in the way Apple sets out.

In most 3rd party tests, the top-of-the-line G5 competes favourably with the top-of-the-line P4 in Photoshop. However, a G5 1.8 these days is definitely NOT a top-of-the-line G5. Maybe if we had a single G5 2.5 Power Mac, it would do well against a P4 3.4 (36% clock speed advantage) or something, but that's a completely different kettle of fish from the 1.8 GHz G5 vs. the P4 3.0 (67% clock speed advantage).

[pliny]

Originally posted by Eug Wanker:
You can't seriously believe Apple is going to post stuff that makes them look bad?

I see nothing in these results that calls them into question. 45 filters, 600 mb, 2 gig ram, what's the problem?

[Eug Wanker]

Originally posted by pliny:
I see nothing in these results that calls them into question. 45 filters, 600 mb, 2 gig ram, what's the problem?

It's just as easy to make the G5 look bad, depending on what filters and actions are used most often. If you look at the benchmark subtests, there are certain things that the G5 excels at, and there are certain things it does mediocrely or poorly on.

The bottom line is that if you believe the hype, you're in for a disappointment. While the single 1.8 is most definitely an adequate Photoshop machine for most people, remember that it uses the same architecture and CPU as the consumer-oriented iMac. The single 1.8 Power Mac is the only one that uses a 3:1 CPU multiplier, and of course, there's only one CPU.

I think if you're going to buy a Power Mac at this point, the best bang for the buck is the dual 1.8. It has a faster bus and it has dual processor goodness. Mac OS X loves dual processors of course, and Photoshop is partially dual optimized too. For the price and performance of the single 1.8, you may as well buy an 17" iMac 1.8 GHz. The only thing you gain with the single 1.8 GHz Power Mac is the ability to upgrade the video card (which won't affect Photoshop speed), and the ability to add a second hard drive. It's lame though, since you can't add a 3rd hard drive, nor can you add a second optical drive. And you don't get the free 17" LCD of course.

[pliny]

Originally posted by Eug Wanker:
It's just as easy to make the G5 look bad, depending on what filters and actions are used most often. If you look at the benchmark subtests, there are certain things that the G5 excels at, and there are certain things it does mediocrely or poorly on.

It may be true that there are some tests the G5 does better at; but there's nothing in these tests to show that the results are somehow falsified or inaccurate.

Why should the results be read as showing the opposite of what they show?

These results show at least that the P4@3ghz doesn't "spank" the G5 1.8 at Photoshop in 45 filters, indeed they show that the 1.8 G5 is in range of spankeded the 3.0 P4.

Do any tests you know about show that the P4 spanks the 1.8 G5 in PS, or are you just basing your claims on "there is a 67% difference the P4 MUST SPANK the G5!"

[Eug Wanker]

Originally posted by pliny:
It may be true that there are some tests the G5 does better at; but there's nothing in these tests to show that the results are somehow falsified or inaccurate.

Why should the results be read as showing the opposite of what they show?

These results show at least that the P4@3ghz doesn't "spank" the G5 1.8 at Photoshop in 45 filters, indeed they show that the 1.8 G5 is in range of spankeded the 3.0 P4.

Do any tests you know about show that the P4 spanks the 1.8 G5 in PS, or are you just basing your claims on "there is a 67% difference the P4 MUST SPANK the G5!"

Yes. LOTS of Photoshop filters are faster on the P4. For instance, look at this P4 2.8 vs. G5 1.6 Photoshop benchmark. Fortunately, the G5 does very well at stuff like RGB->CMYK. Do you use a lot of RGB->CMYK?

Anyways, you seem to want us to tell you that THE G5 ROOLZ ALLL!!!!111one but that simply isn't the case. If you want a Power Mac, then buy a Power Mac. It's a fine machine, esp. for Photoshop. But if you were going to make a living off it, I would strongly suggest a dual 1.8 over the single 1.8. That same benchmark I listed above also shows a dual 2.0, Take a look at its performance and you'll see why I say the extra $500 for the dual 1.8 over the single 1.8 is money well spent on a Photoshop box.

[polendo]

Originally posted by Eug Wanker:
A single G5 1.8 would get spanked by a 3 GHz P4 in Photoshop too. (I'd still prefer the Mac for other reasons though.)

I couldn't agree with you more.

regards

[pliny]

Originally posted by Eug Wanker:
Anyways, you seem to want us to tell you that THE G5 ROOLZ ALLL!!!!111one but that simply isn't the case.

You said the 3.0 P4 definitely spanks the 1.8 G5 , then you are shown results showing the opposite across 45 filters, then say the tests are not to be believed, then you post a 1.6 test where a 2.8 P4 outperforms it, and say...what? That unless the P4 is called the spankinator tests are lies?

You offer no contradictory tests for the 1.8 v the P4, just your belief, and the statement that different tests will show different performance, some better some worse; which only suggests that the comparisons (and so perhaps even the performance of the machines themselves) are an art, not a science. Your statement about spanking (the word calls for a ), is not based on any such comparison you have posted here, but simply on your opinion about the merits of a P4, which you have already demonstrated.

If you have any proof to show that the Apple results are flawed or lies as you suggest, or indeed, can link to any other tests about the P4 v the 1.8 that demonstrate something other than what Apple's does, just post them!

p.s. an interesting tidbit from the testing of the 1.6 v the 3.0 p4: "if you notice, on some of the benchmarks, the G5 creamed even the P4 3.0Ghz with HT on."

This even further weakens your claim about the P4's spanking the 1.8.

[Eug Wanker]

if you notice, on some of the benchmarks, the G5 creamed even the P4 3.0Ghz with HT on.

No kidding Einstein. I've said it myself about a bazillion times in this thread already. The G5 is better at certain filters/actions.

Originally posted by pliny:
You said the 3.0 P4 definitely spanks the 1.8 G5 , then you are shown results showing the opposite across 45 filters, then say the tests are not to be believed, then you post a 1.6 test where a 2.8 P4 outperforms it, and say...what? That unless the P4 is called the spankinator tests are lies?

If you wish to take Apple's biased benchmarks complete at face value, that's up to you, but some of us prefer to think for ourselves.

BTW, I have never said the Apple tests are lies, but they are indeed biased, for obvious reasons. Apple makes damn well sure that the tests are going to use a workflow for the test that makes Apple look good.

It's similar to the SPEC test claims from last year. Apple chose to use gcc, because they said it "normalized" the compiler out of the equation. However, everyone criticized Apple, for good reason, because there were a number of MUCH faster compilers for the x86 platform available at the time, which Apple chose to ignore, conveniently. Now that xlc/xlf are out with the G5 2.5, I'm find it curious that Apple refuses to publish comparative tests against the P4 3.6 or Opterons. Furthermore, Apple refuses to submit any official benches to www.SPEC.org, despite having used SPEC last year for its own advertising. No matter, IBM has submitted official numbers:

IBM G5 2.2: int-986, fp-1178 (with xlc/xlf)
G5 2.5 (estimated from perfect scaling from 2.2 GHz): int-1120, fp-1339
Intel P4 3.8: int-1666, fp-1839
AMD Athlon 64 RX-55: int-1750, fp-1741

[pliny]

Originally posted by Eug Wanker:

If you wish to take Apple's biased benchmarks complete at face value, that's up to you, but some of us prefer to think for ourselves.

I see! Biased because...why? Because in one test you like, using smaller files and half the filters, the P4 does better, though even then nowhere near the margin of scaling, so those results are ok, but the Apple tests, because of a heavier load with TWICE the filters, are bad?

Where's the bias again?

The files sizes are the same for both systems and both use the same 45 filters. If anything, it shows that under an even heavier load through a greater range, the G5 really performs, and better than the P4.

You claim that the 3.0 P4 is not a POS and that it spanks the 1.8 G5.

You cannot point to one test showing this to be true.

The tests I point to clearly show your claim to be false.

The old tests you point to show that far from being spanked, the *1.6* does nowhere near 67% worse than the P4 3.06.


If you like/love the P4, great! But don't try to say with zero proof and in the face of tests showing the opposite that the 3.0 P4 clearly spanks the G5 1.8

(And you mention a dispute about the gcc compiler and say "everyone" questioned Apple. Even if this issue is relevant to this test, this is exaggeration. People who wanted the P4 to win or who otherwise have some axe to grind questioned it.

As in these results, in the case of the compiler, both systems are using the same program NOT made by Apple. )

[Eug Wanker]

Originally posted by pliny:
The tests I point to clearly show your claim to be false.

Tell that the UK, which forced Apple to stop claiming in advertisements that the Power Mac was the fastest desktop computer in the world.

BTW, have you even read the details of the Apple Photoshop bench? I have. I can no longer find the PDF of the bench however.

The old tests you point to show that far from being spanked, the *1.6* does nowhere near 67% worse than the P4 3.06.

Stop making things up. I never said the 1.6 does 67% worse than the P4 3.0. I said overall it can't overcome the P4's clock speed advantage.

If you like/love the P4, great! But don't try to say with zero proof and in the face of tests showing the opposite that the 3.0 P4 clearly spanks the G5 1.8

Give me a friggin' break. Stop projecting your fanboi assumptions onto me. I'd buy the Mac myself. But I would prefer to make my purchase on informed decisions instead of Mac zealotry. The reason here to buy the Mac is NOT raw speed, because the PC wins overall in raw speed. However, the Mac adds usability with things like OS X, Expos�, and colour management.

And you mention a dispute about the gcc compiler and say "everyone" questioned Apple. Even if this issue is relevant to this test, this is exaggeration. People who wanted the P4 to win or who otherwise have some axe to grind questioned it.

Uh no. Even many of the Mac programmers questioned it. I guess you chose to ignore that.

As in these results, in the case of the compiler, both systems are using the same program NOT made by Apple. )

Yeah, but like I said, at the time MUCH faster compilers were already available for the x86 platform. On x86, if you want maximum speed, you don't use gcc. Now xlc/xlf are available to Apple, but Apple won't repeat the test.

IBM HAS published official SPEC tests however, and the G5 loses. Oh and IBM invented and built the chip, the OS (AIX), and the compilers.

[pliny]

Originally posted by Eug Wanker:
Tell that the UK, which forced Apple to stop claiming in advertisements that the Power Mac was the fastest desktop computer in the world.

You claim this test to be false or otherwise misleading. Prove it or just stop saiyng that the test doesn't show what it does. Simple! Nothing the UK has done is based on the results of this test.

You keep finding reasons to suggest that what the test shows, it can't possibly.

Originally posted by Eug Wanker:
Stop making things up. I never said the 1.6 does 67% worse than the P4 3.0. I said overall it can't overcome the P4's clock speed advantage.

I'm not making up anything. You said that the P4 3.06 in PS, spanks the 1.8 at same; when not even the 1.6 is spanked DESPITE that there is a 67%ish difference in clocks. So how can the 3.0 P4 spank the 1.8 G5 at PS if even in some of the tests you like, the 1.6 G5 outperforms the 3.0 P4?

Originally posted by Eug Wanker:
The reason here to buy the Mac is NOT raw speed, because the PC wins overall in raw speed. However, the Mac adds usability with things like OS X, Expos�, and colour management.

In raw speed in PS as the Apple tests show, the 1.8 G5 is the clear winner over the P4 in the same 45 filters applied to the same files. Until you can show that this test is incorrect or otherwise flawed, the G5 1.8 at PS is the winner in raw speed.

Originally posted by Eug Wanker:
Uh no. Even many of the Mac programmers questioned it. I guess you chose to ignore that.

You said everyone. I said no, not everyone. Now you say many. Many or some you know, is not everyone. Which was my point.

Originally posted by Eug Wanker:
Yeah, but like I said, at the time MUCH faster compilers were already available for the x86 platform.

We're talking about Photoshop, which is made by Adobe, not Apple.

[macaddict0001]

Man I long for the day's when the mac was hands down twice as fast then we wouldn't have these dumb arguments. The bottom line however is that eug is right the p4 and 64 will be faster at photoshop, the bottom line is that the mac is better even though the pc can do renders that take 30 seconds on the mac in 26 seconds, the mac is better because of coloursync and various other technology's.

[Eug Wanker]

Originally posted by pliny:
I'm not making up anything. You said that the P4 3.06 in PS, spanks the 1.8 at same; when not even the 1.6 is spanked DESPITE that there is a 67%ish difference in clocks. So how can the 3.0 P4 spank the 1.8 G5 at PS if even in some of the tests you like, the 1.6 G5 outperforms the 3.0 P4?

In raw speed in PS as the Apple tests show, the 1.8 G5 is the clear winner over the P4 in the same 45 filters applied to the same files. Until you can show that this test is incorrect or otherwise flawed, the G5 1.8 at PS is the winner in raw speed.

It seems pretty clear now that you haven't actually read the PDF where Apple outlines the details of their Photoshop benchmark.

If you actually had read it, and understood it, then you'd know what I'm talking about. Or maybe you wouldn't...

Originally posted by macaddict0001:
the pc can do renders that take 30 seconds on the mac in 26 seconds, the mac is better because of coloursync and various other technology's.

Yep. I agree completely.

[pliny]

Originally posted by Eug Wanker:
It seems pretty clear now that you haven't actually read the PDF where Apple outlines the details of their Photoshop benchmark.

If you actually had read it, and understood it, then you'd know what I'm talking about. Or maybe you wouldn't...


Yep. I agree completely.

Let's see. First you say that the P4 spanks the 1.8 in PS.

Then you point to tests where the 1.6 spanks the P4 in some filters, but instead of saying that your statement on the 1.8 was hyperbole, state that Apple's tests are biased and even suggest they are lies.

Then, when you are asked to prove your claims, you just ignore this.

Then you say that "everyone" questions the gcc compiler test and when called on that, switch to "many."

Now you say that it is not your burden, but indeed mine, to prove that when Apple says the 1.8 G5 is faster than the P4 3.0 in these filters, that this is indeed NOT what Apple says.

You're the one saying the Apple tests are not correct and that they do NOT show the 1.8 to be faster than the P4 3.0 at PS.

Prove it. Simple.

[Eug Wanker]

Originally posted by pliny:
Let's see. First you say that the P4 spanks the 1.8 in PS.

Then you point to tests where the 1.6 spanks the P4 in some filters, but instead of saying that your statement on the 1.8 was hyperbole, state that Apple's tests are biased and even suggest they are lies.

Then, when you are asked to prove your claims, you just ignore this.

Then you say that "everyone" questions the gcc compiler test and when called on that, switch to "many."

Now you say that it is not your burden, but indeed mine, to prove that when Apple says the 1.8 G5 is faster than the P4 3.0 in these filters, that this is indeed NOT what Apple says.

You're the one saying the Apple tests are not correct and that they do NOT show the 1.8 to be faster than the P4 3.0 at PS.

Prove it. Simple.

Give it up pliny. Your constant misreprentation of my posts is tiresome, and your lack of understanding of the topic is obvious. Perhaps you should read Apple's PDF, think about it, and then come back here with some informed posts.

[pliny]

Originally posted by Eug Wanker:
Give it up pliny. Your constant misreprentation of my posts is tiresome, and your lack of understanding of the topic is obvious. Perhaps you should read Apple's PDF, think about it, and then come back here with some informed posts.

Wrong.

You said the P4 3.0 spanks the 1.8 in PS.

I linked to the Apple tests where it clearly shows the opposite.

You say these tests are not reliable and even suggest they are lies on Apple's part.

I say prove it.

You say, you are wrong.

I say, prove it.

I have been consistent throughout.

You on the other hand, have come up with suggestions that Apple's tests are inaccurate, biased or outright lies, and suggest that "everyone" says this or that about other tests not pertaining to PS, and then backpedal.

It is all in your posts.

You have presented a constantly shifting position and language to suit a claim that you refuse to acknowledge as hyperbole and which you have given no evidence to support.

I can see why you would call a demand for proof, tiresome. You have yet to deliver.

I have said the same thing since the beginning--if this is your position on Apple's tests, prove it.

Otherwise the tests showing the 1.8 G5 spanking the 3.0 P4, stand, and your statements about the P4/G5 are hyperbole.

[Eug Wanker]

So are you going to actually read the benchmark details or not? It seems rather hypocritical of you to declare everyone a liar based on Apple's tests, when you haven't even read how those tests were done.

I'm done responding to you, because I have run out of patience. If you can't be bothered to make the effort to even read Apple's info, then I can't be bothered to explain, yet again, why tests can be biased in either direction.

You are officially now on ignore.

[pliny]

Your response is unfortunate but not surprising.

If you call into question the veracity or integrity of Apple's tests, then the burden is on you to disprove them, not someone else.

If you say that the P4 spanks the 1.8 in PS, it is likewise your burden to show how, especially when there are tests showing the complete OPPOSITE.

Feel free to take me off ignore and respond to my requests for evidence when you have some.

[deboerjo]

All right, since Eug is sick of you, I'll pick up where he left off.

Fact is, Apple's benchmarks used a handpicked set of circumstances to show the G5 in the best possible light. They are perfectly accurate benchmarks, but Apple's marketing misrepresents the situation somewhat by implying that overall performance across the board is that good, when in fact in most situations the G5 isn't nearly so speedy.

And Apple's SPEC benchmark comparison WAS outright decietful. They posted outright false results for a Dell Pentium 4 to show the G5 as being faster, when in fact, the Dell scores much higher. Since they got a few lawsuit threats, Apple pulled the decietful benchmark claims.

Want proof?

http://www.xbitlabs.com/articles/cpu...rpc-g5_18.html
The most concise overview of Apple's SPEC benchmark fraud

http://www.pcworld.com/news/article/...49,pg,8,00.asp
The dual 2.0 G5 looses to a 3.2GHz single P4 in every test except photoshop. The G5 gets absolutly destroyed in Word.

http://www.barefeats.com/p4game.html
From a Mac advocacy site, no less, the Mac is thouroughly shamed in game benchmarks.

-Jon

[macaddict0001]

deboerjo
for the record bare feats say's that the g5 2.5 single processor is about as fast as a p4 3ghz with hyperthreading, and the duel is twice as fast, we can assume that if the pentium is equally efficient at higher clocks that a 3.8 would still be slower.

[deboerjo]

Originally posted by macaddict0001:
deboerjo
for the record bare feats say's that the g5 2.5 single processor is about as fast as a p4 3ghz with hyperthreading, and the duel is twice as fast, we can assume that if the pentium is equally efficient at higher clocks that a 3.8 would still be slower.

I think Bare Feats is giving the Pentium4 too much credit, averaging everything out a 2.5 single should be a match for a 3.6 at least. A 1.8GHz G5 that Eug and pliny have been discussing should match a 2.6 P4 at least. General rule of thumb (and this is a VERY BROAD GENERALIZATION) is that a G5 is a match for the Athlon64 at the same clock speed, and a match for a Pentium4 at 1.5x the clock speed. Xeons/Pentium4EE and Opterons/AthlonFX are another story.

Dual processors never translate to double the performance, and sometimes offer no performance at all. But regardless of that, it makes no sense to compare a dual-proc workstation with a single proc at half (or less) the price. Feature-for-feature and dollar-for-dollar, the Powermac G5 family are generally (very generally) speaking on roughly equal footing as dual-proc Xeon and Opteron workstations. While the 1.8 and 2.0 models are a little pricey or underpowered (depending on how you look at it), the 2.5 G5 is a bargain, and is probably the best price/performance out there.

-Jon

[UnixMac]

I think its honest to say that in certain apps the mac is faster and in certain apps the pc is faster. A LOT of this is because the PC usually has well written code that is optimized for the X86 while the Mac usually gets crappy ports. Photoshop is not this type of software and clearly shows the Megaherz myth to be true.. most filters, favor the Mac clock cycle for clock cycle thus showing the superior archetcture of the Mac... but nevertheless, some PC's run at much higher clock speeds which help bridge this gap.

OS X alone make the Mac faster for me personally, as XP just gets in my way.

[Pierre B.]

Originally posted by Eug Wanker:
That's interesting, but the G5 1.8 would still get spanked by a 3.0 GHz P4 in Photoshop.

I am not sure if this should happen for a properly Altivec optimized application. The example I have in mind is Altivec Fractal Carbon, in which switching to vector code accelerates the processing speed by a factor of 6. So, if you set G5 1.8 GHz = P4 1.8 GHz, you would need a 10 GHz P4 to achieve this level of performance, if performance scales linearly with clock speed. Now, this example is perhaps in the extreme side, but an application using somehow Altivec, should run around 2 times faster than the scalar code. So, again if you assume that G5 1.8 GHz = P4 1.8 GHz, then such an application should run on the 1.8 GHz G5 as fast as it would run on a 3.6 GHz P4.

[Pierre B.]

Originally posted by deboerjo:
http://www.pcworld.com/news/article/...49,pg,8,00.asp
The dual 2.0 G5 looses to a 3.2GHz single P4 in every test except photoshop. The G5 gets absolutly destroyed in Word.

http://www.barefeats.com/p4game.html
From a Mac advocacy site, no less, the Mac is thouroughly shamed in game benchmarks.

Please, leave out of the discussion games ported from the PC world. Here is the counter-example: I tried to run on a Dell Latitude 800 (P-M 1.6 GHz, 1 GB RAM) the Nanosaur port. You know, the old Pangea game that came with the first iMacs. It runs MUCH worse than the Mac version on a Wallstreet 266 MHz . Now, would this mean that PCs are crap for gaming?

Same for any heavily x86_platform-optimized application.

[Eug Wanker]

Originally posted by Pierre B.:
I am not sure if this should happen for a properly Altivec optimized application. The example I have in mind is Altivec Fractal Carbon, in which switching to vector code accelerates the processing speed by a factor of 6. So, if you set G5 1.8 GHz = P4 1.8 GHz, you would need a 10 GHz P4 to achieve this level of performance, if performance scales linearly with clock speed. Now, this example is perhaps in the extreme side, but an application using somehow Altivec, should run around 2 times faster than the scalar code. So, again if you assume that G5 1.8 GHz = P4 1.8 GHz, then such an application should run on the 1.8 GHz G5 as fast as it would run on a 3.6 GHz P4.

Yeah, it would seem Altivec is excellent (note - I'm not a programmer), and presumably one reason some Photoshop filters do reasonably well on low clock speed G5s is partially because of Altivec.

However, it would seem that with most cross platform apps, people are not going to spend a lot of time hand tuning code for SIMD (even if it's the type of app that will benefit somewhat from it). As you said your example is an extreme one - most apps won't benefit that much from it, or might benefit more but only with tons of coding time spent. It's true that Apple spends a LOT of time Altivec-ing stuff up the ying yang, but I don't see 3rd parties doing it so much when it means increased development costs and delayed releases, for say a 10-15% speed upgrade (overall).

Still it's nice to see IBM parts with Altivec. I wonder how much money Apple paid them to develop the G5.

[Lateralus]

Consider this a pre-lock warning. This thread is skirting the edge of civil. Try to keep things polite guys.

[pliny]

Originally posted by deboerjo:

Fact is, Apple's benchmarks used a handpicked set of circumstances to show the G5 in the best possible light. They are perfectly accurate benchmarks, but Apple's marketing misrepresents the situation somewhat by implying that overall performance across the board is that good, when in fact in most situations the G5 isn't nearly so speedy.

They used CS 8, 45 filters and a 600 mb file across both systems and Apple says that the G5 ran these better than the P4's. Where is the misrepresentation?

If we point to tests of x86 optimized apps that run better on PCs than a Mac, we can also point to Mac optimized apps that run better on a Mac; but the original post asked about PS.

[deboerjo]

Originally posted by pliny:
Where is the misrepresentation?

When Mr. Average Joe reads Apple's marketing that says "The G5 is twice as fast!" and thinks that means it's going to surf the web and type his college papers that much faster. Thier marketing is now much more subdued, and they now include the proper "at photoshop" qualifier more often than not, but when the G5 was introduced, the marketing was downright decietful.

[deboerjo]

Originally posted by Pierre B.:
Please, leave out of the discussion games ported from the PC world.

I'm not talking about games "ported from the PC world". The Quake3 engine (which is what BareFeats was using for those benchmarks) was developed for both platforms simultaneously and is equally tweaked. Remember it was first demoed on a Mac? There's not a whole lot you can do with OpenGL to optimize for a particular platform/card. DirectX games like Halo generally are much more tweaked for PC, but that's not what's being benchmarked here.

-Jon

[Lateralus]

Originally posted by deboerjo:
The Quake3 engine (which is what BareFeats was using for those benchmarks) was developed for both platforms simultaneously and is equally tweaked.

For OS 8.6, five years ago.

[pliny]

Originally posted by deboerjo:
When Mr. Average Joe reads Apple's marketing that says "The G5 is twice as fast!" and thinks that means it's going to surf the web and type his college papers that much faster. Thier marketing is now much more subdued, and they now include the proper "at photoshop" qualifier more often than not, but when the G5 was introduced, the marketing was downright decietful.

Let's stick to the subject here.

Not one poster has presented anything to show that there is misrepresentation or any thing else in these 8.0 tests to suggest they're biased, wrong, or lies. So until someone can offer proof that they are, at which time it can be looked at, the G5 spanks the P4 at PS 8.

The other tests linked to above tested PS 7 and I recall Adobe saying (I think it was at WWDC) they were introducing more optimization for the Mac across the entire creative suite; so these tests are consistent with that.

If there is evidence to the contrary, just post it!

p.s. check out some these G5 v P4 numbers over at barefeats.com. The G5 spanks the P4 in these as well (PS 7, others).

[UnixMac]

Originally posted by deboerjo:
I'm not talking about games "ported from the PC world". The Quake3 engine (which is what BareFeats was using for those benchmarks) was developed for both platforms simultaneously and is equally tweaked. Remember it was first demoed on a Mac? There's not a whole lot you can do with OpenGL to optimize for a particular platform/card. DirectX games like Halo generally are much more tweaked for PC, but that's not what's being benchmarked here.

-Jon

DUDE! this is so important in the argument of Mac vs PC.. and I have made it my life's mission to get people to understand it. I am not a programmer but I am an ametuer geek, and I can tell you from talking to people in the software world, the the single biggest disadvantage the Mac has over PC is crap portable software code.

Games is a major place where this is apparent in Q3, the engine was tuned for Mac, and the date bares it out. However, most code is neither MP aware nor Altivec and that is the biggest shame for the "Mac" developers .. but then it's all about $$$ ... the cost of properly developing a Mac app with all that stuff would make software all cost like PhotoShop and FCP.. the $50 games have no chance.

X86 is intrinsically inferior to G5 but the remember the old saying.... Garbage in Garbage out-.. well... that's most of the Mac software ports... garbage!

[macaddict0001]

Originally posted by UnixMac:
DUDE! this is so important in the argument of Mac vs PC.. and I have made it my life's mission to get people to understand it. I am not a programmer but I am an ametuer geek, and I can tell you from talking to people in the software world, the the single biggest disadvantage the Mac has over PC is crap portable software code.

Games is a major place where this is apparent in Q3, the engine was tuned for Mac, and the date bares it out. However, most code is neither MP aware nor Altivec and that is the biggest shame for the "Mac" developers .. but then it's all about $$$ ... the cost of properly developing a Mac app with all that stuff would make software all cost like PhotoShop and FCP.. the $50 games have no chance.

X86 is intrinsically inferior to G5 but the remember the old saying.... Garbage in Garbage out-.. well... that's most of the Mac software ports... garbage!

So true Thats why on barefeats The photoshop mp aware tests are dominated by the duel g5 while the non mp aware tests are more or less equal.
I don't know much about coding but couldn't you create an allgorithm that substitutes for example basic with c or whatever you want all the programmer would have to do then is change file references for different filesystems, or is their some blatantly obvious reason why this wouldn't work. I just think that if it is possible to create a program as complex as mathmatica that a code modifier would be possible.

[Eug Wanker]

So I picked up a Power Mac today... OMGAWD IT ROXORS!!!1 Actually, it's pretty damn slow... It's just a G4 450... but it's a Cube.

I can't wait until Freescale's 90 nm G4 is out in a processor upgrade for the Cube. This is mainly just a surf and iTunes box, so I don't need a huge amount of speed. A 1.5 GHz G4 would suffice.

BTW, I don't think a Power Mac G5 would even fit underneath my desk. The case is HUGE.

[Link]

WTF is with people complaining about "OMG tah g5 is sooo huge how is it gonna fit under my desk!!!!!"

It's about 2" higher than a g4 tower? Do you have a japanese table or something? Gee I have like at least a foot between my g4 and the tabletop

[Simon]

Originally posted by Link:
WTF is with people complaining about "OMG tah g5 is sooo huge how is it gonna fit under my desk!!!!!"
It's about 2" higher than a g4 tower? Do you have a japanese table or something? Gee I have like at least a foot between my g4 and the tabletop

People like to complain and we Mac users sure are demanding.

IMHO if somebody thinks he/she really needs the power of a dual G5 and spends $3000 on the box it can be expected that they buy the suitable $100 table.

If Apple would make something like an LC with a single G5 (aka headless iMac) it would even fit under your japanese table, but then people would be bitching about the missing 3 PCIe slots and the crappy 9800 Mobility soldered to the main board. Not too speek about (chose one of two) a) noise due to the fans keeping it cool b) heat due to missing fans keeping it quiet

[Eug Wanker]

WTF is with people complaining about "OMG tah g5 is sooo huge how is it gonna fit under my desk!!!!!"

It's about 2" higher than a g4 tower? Do you have a japanese table or something? Gee I have like at least a foot between my g4 and the tabletop

Well, it's probably to do with the fact that tah G5 is Hoooge! No, seriously, I have an ikea table and it has a support beam underneath that makes the G5 probably a bit too tall. Not that I'd buy one anyway, since I find the single 1.8 too expensive, and I personally don't need duals for this desktop, since it's mainly just for surfing and basic iLife stuff. If I were a Photoshop jockey though, I would not hesitate to get the dual 1.8 though.

f Apple would make something like an LC with a single G5 (aka headless iMac) it would even fit under your japanese table, but then people would be bitching about the missing 3 PCIe slots and the crappy 9800 Mobility soldered to the main board.

Actually, if you remember, when the Cube came out, although one of the complaints was expandability, the much bigger complaint was its high cost. It died a quick death (although one lives on, on my desk).

[Lateralus]

Congrats on your Cube. I picked up one last night, now listed in my signature.

With the display, I paid $595. Middle of the night sniper bid on eBay.

It occurred to me exactly how far I could take the thing. 1.5GHz 7457, 120GB Hitachi, 1.5GBs of Crucial (already in my Digital Audio), Radeon 9000 Mac Edition 128MB w/ ADC... etc. And then of course, external goodies.

It would involve selling my Digital Audio though, as I am not rolling in cash.

[Link]

Originally posted by Eug Wanker:
Well, it's probably to do with the fact that tah G5 is Hoooge! No, seriously, I have an ikea table and it has a support beam underneath that makes the G5 probably a bit too tall.

That could be easily fixed with a set of bricks or something underneath the legs ... or having the g5 next to the table, hehe

[b]
If I were a Photoshop jockey though, I would not hesitate to get the dual 1.8 though.

*hops on his racehorse* yeeeeeeehar!!!

Actually, if you remember, when the Cube came out, although one of the complaints was expandability, the much bigger complaint was its high cost. It died a quick death (although one lives on, on my desk).

kyoobs are cool. Can shove a 1.7ghz g4, superdrive, and 9800 into one.. not to mention a 250gig HD

[macaddict0001]

How long do you think it will be before a g5 chip can be put into a g4?
On another note this is in a way disapointing because it means the powerbook has room for expansion without going to a g5.