PowerMac G5: World's Fastest Desktop Computer

[tychay]

Re: Ars

Originally posted by Brother Mugga
Guv said:

"sorry 2 break ur hearts"

Although some people have gone a bit loopy about it (check out his amusing replies to some fanatic-nutters who have flame-mailed him), I think he's raised some fairly valid points.

I don't think so:

1. He deliberately misrepresents postings he culled from MacNN (not exactly the cream of the Mac world) as flame-email.
2. He misreads and misinterprets the report (saying SSE is disabled; implying that hyperthreading would increase the benchmark) to the point where I suspect that he is deliberately doing so.
3. He never mentions that it is rare for benchmarks to be documented in such a way that any scrutiny can be done.
4. He engages in an ad hominem "I am a Mac user" defense of his opinion. (By the way, I am a PC user--I actually use and own more PCs than Macs even though I've been using Macs since 1984--so this means that you should believe that I speak the truth when I bag on him.)
5. He managed to get this crap posted on slashdot. Their editors should have known well enough to spot a troll instead of giving it instant street cred by pointing it out.
6. I won't go over the arguments presented in the paper. Some of them are so absurd as to be self-incriminating. Read them as they are great for a laugh.
7. You could have guessed all the fudging going around from the WWDC Keynote if you read it instead of getting caught in the reality distortion field. By the time this person reported it, it was common knowledge to anyone who has even a basic understanding of benchmarking (remember the days when Apple used to use ByteMarks?)

Frankly the only valid point he raised is the one he didn't. That benchmarks can be manipulated to the point of uselessness. As Hannibal put it in one of the articles you linked, ``...anyone who makes a purchasing decision based on these 'benchmarks' deserves to get taken for a few grand.''

 

[Sun Baked]

Re: Re: Ars

Originally posted by tychay
Frankly the only valid point he raised is the one he didn't. That benchmarks can be manipulated to the point of uselessness. As Hannibal put it in one of the articles you linked, ``...anyone who makes a purchasing decision based on these 'benchmarks' deserves to get taken for a few grand.''

He didn't hold it against Steve for trotting the numbers out for the "show"...

If you leave it at that key sentence people would forget what follows.

The Adobe, Emagic and Mathematica benchmark bakeoffs were a bit more believable and relevant. I wouldn't be surprised at all if the dual-G5 really did out-do the P4* on these apps, because they're very bandwidth intensive and they give the G5's excellent frontside bus and memory subsystem a chance to shine. What makes a media workstation is bandwidth, and not just raw CPU power. In fact, high-end Unix workstations from SGI and Sun have always been beefier on the bandwidth side than on the CPU horsepower side--this is one reason why they're so expensive. The G5 tower looks like a real contender when it comes to moving data around, and that's what counts for pro applications.

 

[tychay]

Originally posted by gezzas525
AMD has Madrake 64, SUSE 64 and Redhat 64 all of which have been released.

Since this guy has been banned, I won't argue. I just want to highlight some observations.

Mandrake Linux, SUSE Linux, and RedHat Linux are not 64-bit operating systems. What he is talking about would be, in Mac terms, like Mac OS X 10.2.7 (Smeagol). Compilations that allow the OS to have 64-bit support. In particular, the kernel and glibc are recompiled. Note that this is where most of the performance gain will be anyway. The rest of it still uses the x86 binaries with the system running in in compatibility mode. The problem is a lot of x86 Linux libraries are not 64-bit clean, especially in its file handling. I really think this is more a technicality though. What matters is what counts, not some terminology.

A friend of mine at AMD used to always mention 64-bit Windows--he's one of the lucky few that actually gets to work with such a beast on a daily basis. I don't know when it will see the light of day (certainly, it will have to be released with Windows XP given the delays in Longhorn to be of any use). Because of politics, economics, or bugs, Microsoft has not announced a 64-bit Windows for Opteron (they have had a 64-bit Windows though, for the Alpha).

Speaking of the Alpha (or the UltraSparc that used to grace my desk in grad school). There were a lot of 64-bit machines long before the G5 and the Opteron. They were "desktop workstations" and not "personal computers". I guess when you are the inventor of the personal computer, you can to define what is, and isn't a "personal computer". So by some definition, Apple can claim that they are the "world's fastest personal computer"

Take care,

 

[Cubeboy]

Re: Lies, Damn Lies, and Benchmarks

Tychay, your joking right? Because thats the funniest piece of garbage that I've ever heard. Honestly, look at the test configurations yourself, Veritest was using GCC 3.3 Build 1379 which (after a little digging) turned out to have specifically optimised it's compiling code (by Apple) for the G5 (usually more significant than your basic x86/PPC optimizations). Obviously, their is an difference between the GCC compilers, and by all accounts a significant difference. The part of your post about the library especially interests me considering Veritest's notes regarding the malloc library indicated that Apple included an optimized library probably like the LIBMoto library you mentioned.

Regarding GCC, it's at best is an mediocre compiler, especially for complex architectures like the Pentium 4 which takes an significant performance hit on it due to the lack of streaming code (and not just in SPEC, look up the SciMark scores for ICC and GCC since you mentioned scientific and engineering apps). Opteron, PPC970, SPARC and pretty much any other chip will run slower on it than with their premier compilers.

Might I also suggest that you wait for some real world benchmarks to come out from an third party before you shoot your mouth off or do you just believe whatever Apple tells you to believe. (Not that Apple is unique in this aspect, Sun, AMD, Intel are just as prone if not more to make extravagent claims and favorable comparisons, IBM's probably the only "mellow" company out their)

 

[ddtlm]

Cubeboy:

Regarding GCC, it's at best is an mediocre compiler, especially for complex architectures like the Pentium 4

GCC has been shown to be competent for Athlons and Pentium 3's when compared to Intel's latest ICC. GCC lost, but not too badly. It's only when the target computer is a Pentium 4 that GCC starts to loose regularly by large amounts.

 

[Rower_CPU]

Originally posted by wizard
Hi Rover;

I have no intention of going shopping for somebody else. It is really up to the purchaser to determine the make up of his or hers machine. Assuming one knows how to use the web, Opteron based machines can be found using AMD or NVIDIA chipsets. I will continue to ignore your fishing excercise, you should know as well as I what is on the market, if you don't you really have no business responding to this thread.

What is clear is that people should not be going out and buying a G5 based on assumed performance. Specific applications yes, OS/X yes, a cute enclosure well maybe, but not because they believe that it is an outstanding machine performance wise. Performance will have to be judged against its (G5) contemporaries. Lets face it Panther is a much more compelling reason to buy a G5 than its much touted performance.

Thanks
Dave

Well, wozard, I guess you have no desire to actually debate the topic and would rather troll through the various threads here. You have continually said the the G5 was a poor performer for the price, but have consistently failed to prove your point by comparing it with an equivalent PC. The burden of proof lies with the person making the claims.

You are the one who has no business posting here.

 

[Sun Baked]

This sums it up nicely...

Originally posted by DeltaOP:
All I know is that at Approx. 11:00am PST on Monday June 23, 2003, the world came crashing down on PC fanatics and they are not liking it one bit.

Of course the cheese grater image does hold up well to all the benchmark whine, plus there are plenty of crackers to go around.

---

Until the machines hit the market and people start playing with them, it's all guessing games.

 

[adamfilip]

i wish that when apple demo'd the comparison at wwdc

they compared the new dual 2gig g5 to the old TOTL Dual 1.42 g4
and not against a pc, only cause then mac users might appreciate how much of an improvement was obtained.. and since we have an appropriate reference with a dual 1.42.

but with the comparison to the Xeon's
later on once the g5 is publically available then we will see all the hardware sites jumping at the chance to see if it is actually faster then a pc. if they didnt claim that it was the fastest PC then.. there wouldnt be so much fanfair.. and speculation. think about how much more publicity the g5 is getting even from hard core Wintel sites.. its excellent marketing for apple even if the g5 isnt as fast as they claim

 

[Cubeboy]

Originally posted by ddtlm
Cubeboy:


GCC has been shown to be competent for Athlons and Pentium 3's when compared to Intel's latest ICC. GCC lost, but not too badly. It's only when the target computer is a Pentium 4 that GCC starts to loose regularly by large amounts.

Yes, GCC is an competent compiler, it just might not be the best choice for anything other than say an Athlon (which is known to run poorly optimized/poorly coded programs better than most other cpus) or Pentium 3. Sun's compiler compiles code which runs around 25 to 30% faster than the same code compiled with GCC on SPARC. Similar scenarios are true for Visual Age compilers for Power4 and PPC970s and ICC compilers for Pentium 4 and Itaniums.

 

[soggywulf]

Originally posted by tychay
--
terry chay
MyCasaNetwork <http://www.mycasanetwork.com/>
Spinodal Systems <http://www.spinodal.com/>

Is this THE Terry Chay? Did you used to be a techer?

 

[BigJayhawk]

What about the Software Tests?

I know that this has been briefly mentioned before but not specifically this way:

What benefit would there be to Adobe and Wolfram Research (Mathematica) to show that their products are MUCH FASTER on the new G5 Macs than they are on the 3 Ghz P4 and the Dual 3 GHZ Xeons??? ESPECIALLY IF THIS IS NOT TRUE?!?

I KNOW that Adobe sells A LOT MORE software to the PeeCee world than they do to the Mac world (although they sell much more per installed user to the Mac World). There is also a version of Mathematica for Windows if I am not mistaken. These developers would not want to alienate ANY of THEIR OWN customers with LIES just to make Steve happy for the day.

These should be about the most independent real world tests that we shall see. Who cares if a program like Photoshop is "better compiled" for OS X than it is for Windows XP??? If you want to use Photoshop on any platform, then the Mac is faster (and ALMOST undebatably has a better, more stable OS).

As the G5's start to appear, the more Mac Application to PeeCee Application tests that we win the more clout the results will have. I'm REALLY looking forward to the OS Speed in Panther.

 

[wizard]

I would hardly call myself a troll, I find the G5 to be a seriously interesting machine, especially in combination with Panther. As I have said; my remarks on the performance of the G5 are based on Apple's and other information posted on the web. If Apples information is incorrect I will change my mind, but I do not believe that this is the case. The comparisons have already been made against old Intel technology, since this technolgy is being eclipsed by newer processors, that perform far better; that is where the comparisons should be made. The proof has been supplied by Apple, what more do you want.

What the G5 does offer as a counter point to the old P4 technology is floating point performance, that however is not the whole ball game. To many people seemed to believe that the G5 is an awsomely powerful computer, that is not alway going to be the case relative to other contemporary hardware.

Thanks
Dave

Originally posted by Rower_CPU
Well, wozard, I guess you have no desire to actually debate the topic and would rather troll through the various threads here. You have continually said the the G5 was a poor performer for the price, but have consistently failed to prove your point by comparing it with an equivalent PC. The burden of proof lies with the person making the claims.

You are the one who has no business posting here.

 

[ddtlm]

Cubeboy:

Sun's compiler compiles code which runs around 25 to 30% faster than the same code compiled with GCC on SPARC.

Heh... seeing as how the Sparcs that Sun sells are still in-order processors this would not surprise me. I bet they totally depend on a smart compiler.

 

[Sun Baked]

Originally posted by wizard
As I have said; my remarks on the performance of the G5 are based on Apple's and other information posted on the web. If Apples information is incorrect I will change my mind, but I do not believe that this is the case. The comparisons have already been made against old Intel technology, since this technolgy is being eclipsed by newer processors, that perform far better; that is where the comparisons should be made. The proof has been supplied by Apple, what more do you want.

Thanks
Dave

You're right, Apple should have compared the machine to the other 64-bit desktop that is coming in September, the Athalon-64.

 

[wizard]

Originally posted by tychay
What's so "old" about the P4? The P4 they benchmarked against is a great (new) design using the same process (130nm) as the G5. Just because it has a lot of silicon wasted on legacy doesn't make a chip "old technology".

The p4 is old technology, as die shrink and a few new features does not make a new processor.

If you knew anything about the way the P4 is structured, you'd be surprised, as I was, that the single CPU G5 achieved anything close to SPEC integer parity with the P4 (actually, not that surprised because you would have been aware of IBM's benchmarks in Microprocessor Forum last year).

There is absolutely nothing surprising about the p4. If you want a surprise look into the performance that AMD got out of the Opteron running I86 code. That is impressive!

Finally, I advise you not to talk in a "megahertz adjusted basis" ever again, because your PC-loving friends will laugh you out of the room. Such commentary is Apple-marketting smoke-and-mirrors designed to distract us from the fact the G4 had lost parity with the x86 (Athlon and P4). Now that Apple finally has a real workstation-class processor, we can finally shed ourselves of that embarrassment.

I will talk about what I think fits the subject. The comparison was between a G4 and a G5. Sure there is no G4 running at 1.6GHz or even 2GHz for that matter, but the published information does seem to indicate that it would have very good integer performance relative to the G5. This is actually a good sign for Apple; if a faster g4, with a coresponding faster I/O bus, can be stuffed into a laptop the powerbooks could remain competitve for awhile longer.

(The only time such a thing is close to relevant is in notebooks. In those cases, it is better to look at the amount of work done / watt instead.)

Yes, it is true that the G4 (and G5) do more work per clock cycle than the P4. But the P4 is an engine that is designed to rev higher--that's like saying a PT Cruiser is faster when at the same RPMs as a Honda S2000, so therefore it makes it a faster car...

The P4 was designed to "rev" higher to keep intel ahead of the competition for marketing purposes. The chip itself is not an outstanding performer at the frequency it runs.

Apple hasn't released any G4 benchmark done under the same conditions. Most likely because it performs so horribly in SPEC as to be attrocious. Also, you cannot get a G4 at 2 GHz yet so the whole argument is moot. When you can, then we can compare it to the G5s going at 3 Ghz because by then, it surely must.

Take care,

terry

The ability for IBM and Apple to deliver faster machines is going to be critical to the success of the G5. It will be a challenge as the I/O bus frequency increases. So yeah we agree on one thing that the team needs to get to 3GHz fast. Like far less than 12 months.

Dave

 

[wizard]

Originally posted by The Shadow
Common you PC guys. This thread has disintegrated into BS on top of BS. Bogus price comparisons and everyone's a professor in CPU architecture, with a book due out soon.

Actually I think that some valid point have been made on both sides.

If you look at the G5's competition, and consider the pro graphics/video market it is designed for, it has no competition in the marketplace on a performance for value equation.

The DP 2Ghz G5 is cheaper than that Opteron Boxx thing, I looked at the site myself and other have made detailed posts to this effect. And for floating point stuff like time consuming Pshop, video rendering etc, it pees all over Zeon and Pentium. This is when performance really counts. It's also much cheaper than dual Zeon or Pentium as well as the Boxx.

Box isn't the only one selling Opteron systems. But as I've said before I won't do somebody elses shopping. You have a valid comparison with the Zeon with respect ot float performance, that however is not important to everyone.

On top of that you have an unequalled OS that is about to get better, and make the system faster and we can expect considerable Speed bumps within 6 months - so that should shut people up if a faster competition comes out sooner - you always have to consider that R&D of various companies is always out of sink.

With all due respect, I think you PC guys have got sucked in by Apple's marketing. This kind of interest/furore is exactly what they probably wanted, so people would start to take notice of the new PMac. Let's face it, if Apple did just release the world's fastest computer, who in the PC world really cares anyway. Speed doesn't make most people switch one way or the other. It's more a question of what do you want to do and what computer does it best, AND personal preferences - don't forget they ARE valid too. Many people in this thread seem to have forgotten that. So they put out a new slogan and all of a sudden PC guys are talkin about it. Do Apple really care if it's true? Everything moves so fast it could only be true for 2 or 3 months anyway. Brilliant marketing - and most of it's free.

Its a msitake to believe that speed is not important to some people. That is one of the causes of the slowdown in PowerMac sales. Your comments about preferrences and operating systems are completely correct, this issue I've been arguing is speed. As I've said Apple has reached parity with the PC market place performance wise, but htat is about all.

I think instead of worrying about Apple's marketing, we should be focussing on the fact that, in spite of all the arguments, Apple finally again has a competitive price to performance equation (for the top machine anyway - not sure about the 1.6) and this frees customers to focus on the important issues which have been derailed over the last few years of megahertz wars (Brilliant Intel marketing strategy by the way). Ie What do you want to do and what machine does it best, and what are your personal preferences.

Personally, I think PCs are S**t.

Sorry couldn't resist a joke after all that.

Where the Joke;

Frankly the only reason I'm running PC hardware at the moment is Linux. Now you may not believe this but Panther on a G5 is very appealing, but I'm not about to go out and buy such a ssytem with the belief that its absolutely the fastest PC going. That is not the case.

Thanks
Dave

 

[soggywulf]

Originally posted by wizard
I would hardly call myself a troll, I find the G5 to be a seriously interesting machine, especially in combination with Panther.

Unfortunately, if you criticize Apple at all, some folks here will automatically assume you are a Mac-hating wintroll.

Would anyone care to recall the fate of the British motorcycle industry? There are some parallels to be drawn to the situation at hand.

 

[Frohickey]

Originally posted by tychay
Apple hasn't released any G4 benchmark done under the same conditions. Most likely because it performs so horribly in SPEC as to be attrocious. Also, you cannot get a G4 at 2 GHz yet so the whole argument is moot. When you can, then we can compare it to the G5s going at 3 Ghz because by then, it surely must.

Most likely Apple hasn't done any G4 benchmark done under the same conditions because there is no reason to do so. The G5 machines are pending. Why spend the marketing budget on doing some G4 benchmarks? What is that going to accomplish?

 

[wizard]

Apparently this is the case. I'm sitting here on a very old PC trying to look at the G5 with an open mind. Between the G5 and the other intros at WWDC I saw things that really tweaked my interest. At the same time though I'm not going to take every thing as its feed to me, I do have a mind of my own. That mind tells me parity not an overwhelming lead for the G5. For Apple that is a good thing.

Now there is alot of wondering going on as to why AMd was not int he comparison. I think the reaosn is fairly simple, AMD is apparently a manufacturing partner. May be wrong on this as solid information is hard to come by, but isn't apple using some of AMD's I/O chips?

I do not want Apple to fail, but if they follow the same behaviour as in the past they will get the same results as they got in the past. And soon will go the same way as the Britsih motorcycle industry. That would be a huge shame.

Thanks
Dave

Originally posted by soggywulf
Unfortunately, if you criticize Apple at all, some folks here will automatically assume you are a Mac-hating wintroll.

Would anyone care to recall the fate of the British motorcycle industry? There are some parallels to be drawn to the situation at hand.

 

[Sonofhaig]

Air Flow.........

I don't know if anyone else mentioned this. The front and back are set up for optimum air flow. This thing is going to have to be dusted all the time. I just know how some people are. They let electronics build up with dust. I think even with constant dusting, the G5 will accumulate a lot of internal dust. Those nine fans are going to suck it in from the outside. Mark my words.......

 

[Rower_CPU]

Originally posted by soggywulf
Unfortunately, if you criticize Apple at all, some folks here will automatically assume you are a Mac-hating wintroll.

Would anyone care to recall the fate of the British motorcycle industry? There are some parallels to be drawn to the situation at hand.

No, only if you criticize Apple without providing any sort of evidence to substantiate your claims. (i.e. troll).

This community, on the whole, is quite "zealot-free".

 

[ddtlm]

tychay:

Apple hasn't released any G4 benchmark done under the same conditions. Most likely because it performs so horribly in SPEC as to be attrocious.

Below I provided a link to a site did some GCC 2.95.2 G4 SPECing a while back, with a dual 1ghz Quicksilver. In SPECint it got 306 base vs 840 for the new dual G5, but of course the compiler for the G5 was better and Apple used that special malloc library. Based on this I'd say that the G5 integer performance does scale somewhat more than linearly.

http://www.heise.de/ct/english/02/05/182/

 

[jbomber]

Re: Air Flow.........

Originally posted by Sonofhaig
I don't know if anyone else mentioned this. The front and back are set up for optimum air flow. This thing is going to have to be dusted all the time. I just know how some people are. They let electronics build up with dust. I think even with constant dusting, the G5 will accumulate a lot of internal dust. Those nine fans are going to suck it in from the outside. Mark my words.......

definitely.

 

[The Shadow]

Originally posted by wizard
Box isn't the only one selling Opteron systems. But as I've said before I won't do somebody elses shopping.

Well Dave, you were the one claiming that the G5 was too expensive compared to PCs, yet noone in this forum has come up with a PC cheaper than than the DP G5 Mac! Since this price/performance was your initial main concern, a retraction would be nice.

Originally posted by wizard
You have a valid comparison with the Zeon with respect ot float performance, that however is not important to everyone.

Its a msitake to believe that speed is not important to some people. That is one of the causes of the slowdown in PowerMac sales.

You can't design a machine that suits everyone. The G5 is aimed mainly at graphics, video and music pros and scientists. Nearly everything they do that is time consuming involves floating point and requires a lot of bandwidth. These are the G5's strengths, and where it will massively out perform the "old technology" you speak of. By the way, Intel doesn't call it that. They've just released a 3.2 chip. Funny that!

Originally posted by wizard
As I've said Apple has reached parity with the PC market place performance wise, but htat is about all.

For the market and real world apps the G5 is aimed at they probably have more than parity for the short term. But you're missing the point - Apple only needs parity. There are other reasons to use a Mac. And at the moment, one of those is price, which seems to have taken many by surprise.

Generally speaking, if 2 systems had parity, including the apps you want them to run, who in their right mind would choose a PC? IMHO

Thanks anyway for your contribution, we can choose to disagree. It has been entertaining debate.

Best Regards,

TS

 

[soggywulf]

Originally posted by wizard
I do not want Apple to fail, but if they follow the same behaviour as in the past they will get the same results as they got in the past. And soon will go the same way as the Britsih motorcycle industry. That would be a huge shame.

Indeed it would.

Personally I think the G5s are pretty good. I'm a bit curious as to why they didn't adopt the 256mb version of the 9800, but I suppose that's a minor issue (until you start antialiasing ). The prices are a little disappointing, but that is just one part of price/perf. As far as performance, I don't think we have the data to come to any conclusions yet. All we have is a few app tests done on-stage, hardly what I would call an objective comparison. I think we will really have to wait until the machines become available, and get tested independently against new PCs available in August (or whenever the G5s actually arrive), on a wider range of apps. My prediction is that the G5s will be slightly behind price/perf wise, but well within the value difference represented by OSX. Which would be great news.

And because Man and Woman do not live on bread alone...I think the G5 actually looks pretty good. It has a sophisticated, understated air. And the tremendously free-flowing ventilation system gives me a warm fuzzy feeling. The lack of expandability is disappointing, but then again messing up that great airflow with a bunch of clunky HDs would be a shame. No matter, I can get a spare ATX case cheap and stick the extra HDs there.