PowerMac G5 Photos, Benchmarks and Smeagol

[MacRumors]

- After some controversy regarding Apple's G5 Benchmark -- which has been discussed extensively in this thread. Apple's Hardware VP Greg Joswiak provides comments/responses to the criticism.

- Confirming previous reports, it appears the PowerMac will ship with Mac OS X 10.2.7 per this PowerMac G5 Tech Overview PDF (as noted Spymac). Apple also confirms the codename to this Jaguar update by referencing the update in this G5 Developer Note:

This API will be named OSSystemInfo and is scheduled to appear in Smeagol and Panther.

- Numerous photos of the new PowerMac at WWDC have also cropped up:

MacCentral Photos
MacCityNet.it Photos
MacMinute Photos

- An XBench comparison site claims to have XBench benchmarks of a 2GHz PowerMac G5. However, these numbers are different from these XBench numbers posted to the forums. It's unclear what accounts for the discrepancy.

 

[Mallardx]

brench Mark

What value does xbench really have at this point? It is not optimized for 64 bit, OS X is not optimized for 64 bit, it seems all us MAC users do is scream for faster and better and when we get it we tear it apart because it is not fast enough or better than what the hype made us believe.

Personall, i am waiting for the dual 3Ghz machines before I purchase a new PowerMac.

 

[rice_web]

Anyone care to offer Xbench scores for other systems?

 

[noverflow]

Originally posted by rice_web
Anyone care to offer Xbench scores for other systems?

they are on the site... thousands of them

 

[Mallardx]

What I really want to see is xbench for windows. That way we have a comparsion between the two platforms.

The fact of the matter is this: I am never ever ever going back to a windows base computer. Even if it means going without.

 

[copperpipe]

some people are going through denial...

It seems to me that we have some people who just can't accept the fact that Apple has the fastest system around right now, so they argue these miniscule points about the benchmarks. The funny part is that when the G5 goes head to head in "real world" tests, the results are nothing less than astounding. But maybe they just missed the part when the G5 SPANKED the competition in Photoshop, and the other apps which are exactly the kind of stuff you'll be using a processor like this for...

 

[macnews]

A few thoughts:

1. Some people are never satisfied and will find faults with everything and anything you do.

2. If Apple did "cheat" by enhancing the G5, they aren't the first ones on the block to do so. Intel, AMD, even Microsoft have all been acussed of fudging/tweaking performance tests to hardware (software in MS case). If you look around there really is NO good apples-to-apples (pun not intended) comparision/benchmark.

3. The ONLY good benchmark is to go BUY two, closely (you won't get exact) OFF the shelf computers. Take them home, install the same programs and hook it up to the internet. Use both the same amount of hours for a week (min 20 hours), browsing the exact same websites, saving the exact same files, etc. Now run your benchmarks and tell me which one is faster. THAT would be the kind of benchmark I would care about. Also include how many times the OS crashed/froze/kernal panicked.

4. Let's face it, Apple is playing catch up. Do I really care if the new G5 BEATS a top of the line Intel (or AMD) chip? Not really - SO LONG AS IT COMES CLOSE! The OS makes up for the extra speed any PC might have. The few seconds I lose in CPU time, I gain back in a well designed OS and apps interface. Not to mention all the stuff that JUST WORKS when I plug it in compared to driver hell with Windoze (yes, even in XP I have had problems with drivers).

5. Look at the big picture, if Apple and IBM can be at 3Ghz in 12 months, where will the Intel chips be? They are pretty much at the top limit. AMD might have an option with Opteron, granted (although from what I understand this will be a lower Ghz and should help to further dispell the Ghz myth). Again, if Apple can keep in the race I will be happy. There is so much more to owning a Mac than just the chip. I hope they get G5's in all their machines soon so I can get my first powerbook!

 

[Arcady]

Optical drive bay?

I can't tell from looking at any of these pictures how the optical drive bay works. I already own a 4X Pioneer "SuperDrive" so I would like to save $$$ when I order my G5, and get it with a combo drive instead. That shaves $190 off the educational price.

Has anyone seen this in person, and can answer these questions:

How does the drive tray work on these new Macs? Does the metal bezel connect to the tray, or is it a flip-down door like on the previous Macs? Can I just stick my DVR-105 in there without any modification?

Thanks

 

[Megaquad]

Cam anyone say why are those Xbench results so bad?
GCD recursion and altivec CPU tests are SLOWER than 1.42 G4? WTF? Who can explain this?

 

[matthew24]

Comparison

The slashdot article sets the record straight. Besides Steve's demo's were quite impressive. Also all those guys of the different Software companies would be lying as well!

So we have:
1: The fastest Hardware
2: The best Hardware
3: The Best OS
4: The best software (besides games, let them have it!)

Apple keep the good job going!

 

[Thundercorp]

10.2.7?

If this is true, that means that the first crop of G5s will not have a full 64-bit OS running on it. And since 10.2.7 is a compatibility stopgap OS + G5s ship on or before September 2, this means that Panther will not ship until after late September (probably Dec 31).. darn.

 

[Bengt77]

Anyone care to explain?!...

See those two benchmarks:

1. Power Mac G4 (DP1.42)
2. Power Mac G5 (DP2.0)

Is this why Apple decided to 'keep' the Power Mac G4 DP1.25 instead of the DP1.42?

I'm very curious. Could it be true the G5 offers soooooo little improvement over the G4?



Nooooooooooooooooooooo!




EDIT: Direct URL for G5 added.

 

[herocero]

1) People in general are hard to satisfy (especially us but questioning a company's testing methods isn't a mark of disdain, i consider it good shopping. You gonna believe every test and every word that comes out of a company's PR department? of course not, which is why you read PC Magazine and eWeek (we could start a whole 'nother thread on ziff-davis objectivity, but i'll hold off) I will read every decent review I can get my hands on before I plop down 3k for a computer, as I'm sure many of you would.

2) How hard will it be for Apple to straddle the 32-bit/64-bit world? 64-bit workstation, 32-bit iMac, g5 xserve and a g4 pb . . . two different OS paths? two teams? photoshop coming in both 32 and 64 bit varieties? i don't know the extreme technical difficulties, but seems to me a more streamlined path (all 32 or all 64 bit) is less confusing and more cost effective. i'm sure a simple recompile is not all that stands in the way of 32 and 64 bit apps and OSes. how long will apple have both 32 and 64 bit systems? something to ponder until the g6 and mobile g5 come out next week ...

hard to argue monday wasn't a good day though

 

[sparkplug]

matthew24 where you reffering to this posting on /. ?

All this talk of gcc removing a variable is naive at best, misinformation if the speaker is knowledgable on the subject. Gcc is not a constant, the quality of it's code optimization varies from platform to platform. To be more specific, gcc is used by Apple to build MacOS X and Apple has been improving gcc PPC code generation. Apple provides gcc to Mac developers. Apple is also IBM's partner in the development of the PPC970. Gcc is the developer optimized compiler for the chip in many ways and is more comparable to Intel's compiler in this respect.

Or maybe this one?

He gives the illusion of parity by using the same compiler on both platforms. But the back ends to these compilers are different pieces of code written by different people. There is no parity.

Or perhaps this one?

Several other posters have noted that GCC/970 is really not the same compiler as GCC/Xeon. Sure there may be a bit of code in common between the versions, but the job of a compiler is to produce object code... and by definition, the object code for 970 is different from that for Xeon.

What matters to a purchaser is "How much performance can *I* get out of this machine". If I am performing CPU-intensive scientific calculation that require the fastest CPU I can find (at least for a given number of kilodollars), I'll almost certainly spring a few hundred extra for the compiler that produces the fasted object code on that platform (if needed, there's nothing ruling out GCC automatically because it's free).

It happens that for a Xeon or P4 (or Opteron, for that matter), the compiler that produces the fastest object code is ICC. Intel has done an amazingly good job with their compiler.

Now, sure, I *could* get a similarly optimized 970 compiler for comparison.... if one existed, that is. It looks like right now, GCC is the best you can get on a 970. It doesn't do a buyer any good to know that IN PRINCIPLE a more optimized compiler could be written.

All that said, the 970 looks like a very respectable chip. And Apple is selling their new machines at a very competitive price; and Macs have extremely friendly and stable OSs. All that means that it is probably well worth buying a PowerMac even if it will crunch big computations a few percent slower than a more expensive Xeon. But still... the "GCC is the common element stuff is pretty darn bogus."

Im with this last guy, The G5 dual is a very nice machine (the singles are average to poor, where can I buy that 2ghz single the single benches are based on? it doesnt even exist!). I'll be buying a bundle of these duals next year (depending on how well it compares to the dual opterons when I can bench them myself) . Apple have caught up.

However this was all they needed to do, the apps and os complete the picture. Thats enough, they dont need to be faster, just not horribly slower as the G4 was. The compelling arguments for me are software and performance that is comparable to what I get for the money elsewhere.

What is galling though, what dampens my enthusiasm is the purile dick wagging desire/need to be in #1 spot. Its not real, as a result theres this distatefull and sleazy, even dishonest ad campaign. Go read the white papers on apples site, they are gagworthy.

This rubbish doesnt help the cause, instead of my being plain impressed by the merit of the machine I am left feeling deeply suspicous. It has allways been a terrible failing of the company and of the users, It's somekind of technological small man syndrome.

 

[iGav]

Re: some people are going through denial...

Originally posted by copperpipe
It seems to me that we have some people who just can't accept the fact that Apple has the fastest system around right now, so they argue these miniscule points about the benchmarks. The funny part is that when the G5 goes head to head in "real world" tests, the results are nothing less than astounding. But maybe they just missed the part when the G5 SPANKED the competition in Photoshop, and the other apps which are exactly the kind of stuff you'll be using a processor like this for...

I'm with you here...

I know the bake-off's are not a good comparison etc etc... but the performance of the G5 was truly amazing when compared to the Xeon... it didn't just beat the Xeon, it totally nailed it, (and this wasn't the result of a little tweaking by Apple, the performance gap was way too big) and this is on an OS that isn't optimised, and as such if people wanted to be pi$$y it could be argued that the Mac suffered because of it... but yet it still flattened the competition.

I expect to see similar perfomance characteristics through all of Adobes line when they've optimised it for Panther, and the inherent memory characteristics that will go along with this new architecture.

 

[Stella]

Benchmarks

Speed of a machine may be important to some people. But, personally, I really don't give a damn.

I'd rather use a good quality, but perhaps slightly slower machine, whilst using a very good and rock solid OS, than a faster computer that is crippled by a second rate, user unfriendly OS such as Windows!

Only thing windows is good for is playing games. Nothing else.

 

[Cubeboy]

"Several other posters have noted that GCC/970 is really not the same compiler as GCC/Xeon. Sure there may be a bit of code in common between the versions, but the job of a compiler is to produce object code... and by definition, the object code for 970 is different from that for Xeon.

What matters to a purchaser is "How much performance can *I* get out of this machine". If I am performing CPU-intensive scientific calculation that require the fastest CPU I can find (at least for a given number of kilodollars), I'll almost certainly spring a few hundred extra for the compiler that produces the fasted object code on that platform (if needed, there's nothing ruling out GCC automatically because it's free).

It happens that for a Xeon or P4 (or Opteron, for that matter), the compiler that produces the fastest object code is ICC. Intel has done an amazingly good job with their compiler.

Now, sure, I *could* get a similarly optimized 970 compiler for comparison.... if one existed, that is. It looks like right now, GCC is the best you can get on a 970. It doesn't do a buyer any good to know that IN PRINCIPLE a more optimized compiler could be written.

All that said, the 970 looks like a very respectable chip. And Apple is selling their new machines at a very competitive price; and Macs have extremely friendly and stable OSs. All that means that it is probably well worth buying a PowerMac even if it will crunch big computations a few percent slower than a more expensive Xeon. But still... the "GCC is the common element stuff is pretty darn bogus."

IBM's own Visual Age is a much better compiler as was seen by their own preliminary PPC970 SPECmarks. For a 970/Power4, it's capable of translating source code into object code just as fast as a Pentium4 with ICC 7.0. Makes me wonder why Apple used GCC as it seems to only be a mediocre compiler at best.

For equity, Apple should've used GCC on the G5 and BSD on the Pentium4/other x86 processor, that way, they won't have to worry about a custom Intel compiler, while still providing a fair comparison since BSD and GCC are widely regarded as equals (the PPC version of GCC and x86 version of BSD anyways).

I really wouldn't trust the real world benchmarks, they don't look like they were done in a honest way and so far at least one of them (Quake III) is definitely false.

I suggest waiting for benchmarks from Barefeats or Mac Speed Zone, or some reputable PC review site (Ace's Hardware) before concluding on real world performance.

As a general rule, it's better to let a third party review site benchmark the systems rather than let the company who is trying to sell it.

 

[Fender2112]

Hyperthreading turned off?

In the Greg Joswiak interview, he states that hyperthreading was turned off because it made the Intel chips run faster. I find this interesting because I thought hyperthreading was a major breakthrough for CPU design.

He also makes a few other good points folks should hear before getting too excited about false benchmarks

 

[Cubeboy]

Greg Joswiak is an tad dissapointing, not only did he prove himself to have no idea what he was talking about, he also made a complete fool of himself. An analysis:

"He said Veritest used gcc for both platforms, instead of Intel's compiler, simply because the benchmarks measure two things at the same time: compiler, and hardware. To test the hardware alone, you must normalize the compiler out of the equation -- using the same version and similar settings -- and, if anything, Joswiak said, gcc has been available on the Intel platform for a lot longer and is more optimized for Intel than for PowerPC."

He forgot to mention that GCC runs natively on OSX, and that Apple sent literally tons of modifications to be implemented into GCC. Honestly, PPC GCC and x86 GCC are two entirely different compilers, if you wanted a fair comparison why don't you compare a Pentium 4 with a BSD compiler against a G5 with GCC, or better yet, a Pentium 4 with ICC against a G5 with Visual Age, like IBM who's in the least honest with their SPEC scores, oh wait, the G5 wouldn't actually beat the Pentium 4 in those instances.

"Joswiak added that in the Intel modifications for the tests, they chose the option that provided higher scores for the Intel machine, not lower. The scores were higher under Linux than under Windows, and in the rate test, the scores were higher with hyperthreading disabled than enabled. He also said they would be happy to do the tests on Windows and with hyperthreading enabled, if people wanted it, as it would only make the G5 look better."

Could've fooled me, sure makes me wonder WHY Intel, AMD and every other PC company out their uses Windows when they attempt to achieve the highest SPEC scores possible for their CPUs. Scores were higher with hyperthreading disabled you say? Than why were they only disabled for the rate tests where having multiple CPUs and implementations like hyperthreading really counted? I mean it's well known that Hyperthreading lowers scores on single cpu SPEC, I suppose thats why Apple decided to leave Hyperthreading ON in those cases. I mean come on, this guy is just full of it.

 

[Mr.Hey]

The controversy is over
http://www.theregister.co.uk/content/39/31416.html

According to Joswiak, HT was disabled in the SPECint and SPECfp base tests because it yielded higher scores than when HT was enabled. VeriTest did keep HT switched on when it performed its SPECint and SPECfp rate tests.

Indeed, a number of Register readers have pointed out a report on Dell's web site that supports Joswiak's claim. Essentially, it says HT is good for server applications, but less well suited to compute-intensive apps. It uses SPEC CPU 2000 as an example of such an application, and found a "system performance decreased 6-9 per cent on the CPU 2000 speed tests and decreased 27-37 per cent on the CPU 2000 throughput tests" with HT enabled.

 

[evolu]

Re: Anyone care to explain?!...

Is this why Apple decided to 'keep' the Power Mac G4 DP1.25 instead of the DP1.42?

I'm very curious. Could it be true the G5 offers soooooo little improvement over the G4?

This could be a hint of future G4 speeds for other products. If Apple were to press forward with an iMac or PB G4, we could be seeing a preview of their speed here. Chips cost less with larger quantities.

Am I reaching?

 

[Johnny7896]

Let's keep it basic here

It's hard to compare hardware that isn't running on a 64-bit OS
and applications that aren't 64-bit. All those p4's and xeon's run in their native 32-bit OS and run "optimized" 32-bit applications.
By the nature of design, the p4 will run faster than the G5 or Xeon. Because it has a different architecture and it's strength put on sheer Mhz. The G5 and Xeon will blow away the p4 in different test that measures their strengths. The G5 will blow them all away on tests that show their strengths. And so on and so on........ this is like comparing apples to oranges. With no over all winner. The real big difference will be shown in the next year or so. The G5 has a lot of growth potential. The aritecture is wonderful and the Mhz will grow. Better than what we had with the G4. Apple's new optimized 64-bit OS will be released. Applications will be compiled for 64-bit. It's all good. Speed speed speed. Please don't put all the attention on Mega-hz! I have a 2.0 Ghz cordless phone. Doesn’t mean it runs faster than a P4 or G5. It's a number to measure cycles. Not bandwidth to move data. Remember when the P4 and P3 where out at the same time.the P3 was faster than the P4. Because the P3 did more work per cycle. Bottom line is, Mhz is a miss leading measurement.
Intel puts it's money on the Mhz. An Amd will run faster at much lower Mhz than a P4. Welcome to the wonderful world of marketing. Intel designs their chips to run at high speeds-cycles.
With less work done per cycle. Why would the P4M be replaced by the Centrino mobile chip? The P4M ran at much higher Mhz almost double the centrino. The P4m is a dog. It runs at a very hot temp and isn't a better chip than the centrino chip. The Centrino chip does more work at half the Mhz than the P4m.
Apple has a true advantage over the wintel world. It can make a 64-bit OS when it wants to and put in a 64-bit processor. On the dark side, you have to wait for mirco$$$ and Intel to roll them out together. Which is a long process. Apple will be the #1 64-bit PC seller for a long while. It has little or no competition in this market and has no comparison PC. Apple is in a unique market. It's really pointless to compare an Intel 32-bit to a 64-bit G5. We got a 64-bit processor they don't. OS X will be 64-bit and Xp isn't. It's a little like comparing a PS2 64-bit to a Sage saturn 32-bit.

 

[copperpipe]

This is the problem with benchmarks

In my opinion, benchmarks should be taken completely out of the equation. As we all can see, all they do is serve to get everybodies panties in a big ol' bunch. The only conclusion you can draw from benchmarks is that they can be altered in many ways to paint many pictures.

So, how do you measure the speed of a computer then?
1) Pick the App you most use, and compare the overall speeds of that machine

2) Pick an OS that is stable and intuitive. How much time is wasted in computer crashes, and quirky install prblems, etc.? Lots. This is a hell of a lot more important than a few hundred megahertz. Add in the pleasure you get from a hassle free machine, and well, I rest my case.

So, by this I think it's fair to say that in 2d, 3d, movie, and sound editing and creation Apple has FAR AND AWAY the best machines money can buy. And they are very competitively priced.

And if you're doing word processing and internet surfing, save yourself the cash and get a G3...

 

[shadowfax]

Originally posted by Megaquad
Cam anyone say why are those Xbench results so bad?
GCD recursion and altivec CPU tests are SLOWER than 1.42 G4? WTF? Who can explain this?

how about that the test is bogus? there's no way. the G5 has 2 altivec units on it and is clocked higher, and has massive amounts more CPU throughput. conclusion: Xbench doesn't test the G5 correctly.

 

[jayscheuerle]

Let me get this straight...

Y'all are frothing at the mouth over benchmarks and a system that won't even be shipping for 2 months?

If the system isn't shipping, there's a reason for that. These benchmarks and bragging rights don't mean a thing until a finalized, testable machine is available, at which point the PC world may have faster systems as well.

Lighten up Francis!!