G5 Benchmark Roundup

"The G5 is the fastest PC in the world." "Yes, it is." "No, it's not!" Whatever. Read on for more on the subject, if you really want to. Matt Johnson writes "Well it looks like we finally have our first comparison of G5 vs. AMD Opteron, completed by none other than Charlie White, the individual which gained much oh his fame by publishing misleading benchmarks to make Apple's Final Cut Pro Software look like a bad performer. Mr. White's latest comparison shows the Opteron operating roughly 50% faster but what he doesn't say is which compiler was used to generate those SPEC scores. When Apple declared its benchmarks I feared that whoever made the first comparison would likely make this mistake. It seems only appropriate that Charlie White would be first."

An anonymous reader writes "In an ironic twist to the recent benchmark wars, Intel referred the Mac site MacFixIt to an analyst at Gartner Group who actually backed the PowerPC G5 platform with this assertion: 'These models certainly equal Intel's advanced 875 platform and should allow Apple to go until 2005 without a major platform refresh.'"

Another anonymous user writes, "While browsing the Xbench benchmark comparison site, I discovered some G5 benchmarks! The 'G5 Lab Machine at WWDC' got an overall score of 164.78, but much higher scores in certain areas. All of the tests are calibrated to give 100 on an 800MHz DP Quicksilver G4."

vitaboy writes "Sound Technology, one of the "leading UK distributors specialising in musical instruments, music software and pro-audio equipment," seems to have some data regarding the real-world performance of the G5 compared to the high-end PC. They state, 'The dual 2GHz Power Mac G5 with Logic Platinum 6.1 can play 115 tracks, compared with a maximum of 35 tracks on the Dell Dimension 8300 and 81 tracks on the Dell Precision 650 each with Cubase SX 1.051 ... More impressively, the 1.6GHz single-processor Power Mac G5 played 50 percent more tracks than the 3GHz Pentium 4-based system.'"

Comments

It's important to know...

[NickV]

That Charlie White gets off on doing nothing more than trashing the Mac and he often makes tons of things up...

A real good point, and one to points to the fact that Charlie White stats are COMPLETELY cooked up and fake, is that apparently AMD benchmarked against a SINGLE G5 2Ghz Powermac...

Hmm... Where did the Single 2ghz G5 Powermac come from? We know Apple doesn't make them...

If you're gonna lie, at least do it right. Sigh.

(And another thing, AMD has more credibility than Apple regarding self-reported benchmark scores? There is no reason for that other than bias.)

Re:It's important to know...

[GreenHell]

Actually, if you look at Apple's G5 Performance page you'll see that those are the numbers for the dual 2GHz G5.

Given that, I'm still inclined to take the comparison with an entire shaker of salt. I mean, if he's suspiscious of Apple's numbers but not suspiscious of numbers obtained from another processor manufacturer than I don't know what to say other than 'Mr. White, your bias is showing.'

I'm waiting until they hit the market so that the comparisons are done by people who actually got to test the machines themselves, not some guy who knew what he wanted the data to say before he even began writing. Until then, all I'll say is that it looks like nice hardware. But faster or better? Who knows.

Re:It's important to know...

[saden1]

Never trust a company that is trying to sell you something. The test should have been done by him rather than being provided by the companies themselves. This guy has zero credibility and I for one don't put much stock on what he says.

Oh, and that last line about AMD having more credibility is just one of the most stupidest things I have ever heard. I don't buy Macs because of the $$$ and I pack a AMD chip but to say something like AMD has more credibility is very silly.

It is good to see that the old adage of "like assholes, everyone has an opinion" hold true.

Re:It's important to know...

[CompVisGuy]

I think that the $$$ argument is flawed -- it's a myth.

If you compare the price of a G5 (or pretty much any Apple system), with an equivalently-specced PC from a reputable supplier (such as Dell) -- if you can in fact find an equivalent (and frequently you can't, Apple now out-perform PCs, and often you get standard features on an Apple that you can't get on a PC) -- then you will find the Dell system to be more expensive than the Apple.

Granted, you will probably be able to build your own system, or buy from a local PC shop, a PC with a decent spec that is cheaper than an Apple system. However, I have had a couple of very bad experiences with small/mid-size PC builders, and a horrible experience building my own system (I'm a qualified electronics engineer, but I was let down by some dodgy components and very poor aftersales service). Others may have better experiences, but I think it's a matter of luck over judgement. So, ever since, I've vowed to only buy from the big boys.

Next we come to software. On an Apple system, OS X is included in the price of the system -- you often have to pay extra on the PC system for Windows (OK, Linux etc. are often cost-free if you want to go that route). Sometimes Apple software is more expensive or unavailable on the Mac -- but in my line of work (statistical modelling), all the software I need is available. For Word document monkeys, you also have MS Office on the Mac (I'm told it's better than the PC version). Games are slower to appear on the Mac -- that's a potential drawback.

I spend almost every working day in front of a computer. If you had to drive around for a living, you'd want a decent vehicle: any extra cost of an Apple system over a PC system, amortised over the time spent using the system, is almost zero. Oftentimes, the Apple system is cheaper anyway.

But here's the real reason to buy a Mac: The integration between the hardware (some of the best-engineered in the industry) and the OS (OS X is probably the best OS around at the moment). "It just works" is something I hear from people who make the 'switch' from PC to Mac, and it's true.

That's my opinion. Maybe I'm an asshole, though.

Re:It's important to know...

[Paradise Pete]

I build my own systems and I sure as hell can build a screaming system for much less than any Mac out there.

That's like dropping a crate full of automobile parts next to your neighbor's new sports car and saying "see how much cheaper mine is?"

Useless article

[cioxx]

First benchmarks? This is a joke. He didn't even get to test any of the G5's, nor bench'd them.

DMN has obtained SPEC benchmark data from AMD

Right! He obtained them.

It's a biased opinion piece. Now I'm aware that Apple kick-started the G5 with lots of smoke, which is the nature of the business in the computer hardware world, but to discount these numbers just because of some hype during WWDC presentation is silly.

How about we wait for the REAL benchmars from Anandtech and put away some speculation from webmasters who can't even hire anyone older than 14y/olds to design their websites?

Re:Useless article

[MalleusEBHC]

Apple states they are the first 64-bit desktop.

Actually, Apple never has said that it is the first 64-bit desktop, as they know that they would be wrong due to previous 64-bit workstations. They have repeatedly said that it is the first 64-bit personal computer, as shown on their PowerMac page.

Re:six of one half a dozen of the other

[goombah99]

Sir,

if you had read anything in the last week you would know that the only area the G5 is slow in is integeres. It dominates on floats. it's basically a tie on integers: in Single processor mode it loses by perhaps 10% on the SPEC tests and in Dual processor mode it beats the Dual xeons by a margin of maybe 20%.

not even shipping yet...

[boomerny]

I'll wait til the systems are actually shipping and I've seen some independent real-world benchmarks before making any judgements. Xlr8yourmac.com should have some good information once they ship, and maybe barefeats.com

Re:Can't find SPEC results at spec.org for Apple??

[Anonymous Coward]

Probably because we're still talking about prerelease hardware. The Veritest results are reliable, but they're not necessarily going to be perfectly representative of what you'll see in production hardware running the final build of 10.2.7.

Apple will submit figures to SPEC when they've got final hardware and software.

benchmarks; can't live with or without them...

[flaroche76]

People obviously shouldn't form an opinion on a new platform in the first week following its much hyped anouncement. I think the only thing this first week proves is that at least Apple was able to put itself back on the map and be worthy of performance comparison with high-end systems. Or else, why would these PC-centric doofus post early benchmarks and make asses out of themselves if not to try to defuse an apparent threat? What I want are options. I think Apple just gave me another one. But I won't base my judgement on the number of times Steve Jobs says the word 'awesome' in a keynote address or on shady benchmarks done on an apparently non-existing model (single 2ghz cpu)... I think people should let their emotions settle down and wait to get their hands on a real machine and try it out themselves...

Re:benchmarks; can't live with or without them...

[Anonymous Coward]

People obviously shouldn't form an opinion on a new platform in the first week following its much hyped anouncement.

Of course they should. That opinion is perfectly valid. And it is, "Wow. Those are going to be really fast. They look cool. I'm excited."

Or else, why would these PC-centric doofus post early benchmarks and make asses out of themselves if not to try to defuse an apparent threat?

In my experience, PC doofuses have always been big with the benchmarks. It's like a bragging right to them. "I tweaked my dual Smockron 4500 and got it up to 313.3 on SPECdickweed_base!"

Meanwhile, us Mac doofuses (and I use the term with the greatest affection) spend that same time actually working. Because we need the extra cash to feed our $4000-a-year Mac habit.

What I want are options.

Oh, come on now. No you don't. What you really want is a computer that satisfies all of whatever your personal criteria for goodness are. If there were only one computer in the world but it were perfect, you'd be happy.

The whole "what we really want is choice" thing just ain't so.

I need a G5 to keep track of all the claims

[GurgleJerk]

Looking at everything I've seen so far, it looks like the G5 at 2.0 GHz is comparable to a current Xeon or P4 on raw speed. Maybe it lags a little bit in some areas, and in a few areas it can beat the Xeon or P4. But I think we've gotten a little too anal about the processor specs. If I'm not mistaken, Apple didn't claim "World's Fastest Processor." they claimed "World's Fastest Personal Computer."

At 2.0 GHz, the G5 is on par with the current top processors, but what I think people need to look at is that the 1GHz bus is a monster. It allows data transfer rates that smoke other desktop systems. This is where Apple picks up a lot of speed, especially with disk-hungry programs like Photoshop. So the total system is significantly faster than the PC in terms of that kind of real-world performance.

And there are two more things that give the G5 an advantage: price and GHz. If the claim of twelve months to 3.0GHz is true, then at 3.0GHz the G5 will be exponentially faster than a 3.5 or 3.6 GHz P4. I don't know precisely how fast the Intel chips will be in 12 months, but a whole GHz? Unlikely.

Lastly, price is a fantastic advantage for the G5 systems. At $3000 you can buy the fastest Mac and a machine that can run certain apps twice as fast as PC systems. And it's cheaper than these top-of-the-line PCs by more than $1000. The G5 is simply the fastest, cheapest system with the most potential in the future to get even faster. When looked at in total, there really isn't a lot of debate on those points.

Re:I need a G5 to keep track of all the claims

[FueledByRamen]

This is where Apple picks up a lot of speed, especially with disk-hungry programs like Photoshop.

That probably should read "memory-hungry." Disk transfers are still really, really slow - although SATA (which is used in the G5) can go at 150 megs/sec, so can full-duplex Gigabit Ethernet (also included). The real performance ass-kicker is the memory bus - they use 128-bit DDR400, and I'm assuming it can be interleaved (since you're probably going to put multiple sticks in it anyway) for even better performance. They get 6.4GB/sec (gigabytes) out of it (stated at the Stevenote), which is pretty damn good. Not quite enough to saturate the processors' FSBs, but if you need to move a lot of stuff to/from RAM, PCI/X slots (optional), AGP, and the I/O controller (sound, ethernet, etc), like in any game, any high-end 3d app, or any audio app that includes an effects processor (especially when running it on a real-time audio input, recording, while also outputting the results, at 96khz 48000/stereo), the G5 will dominate.

Re:I need a G5 to keep track of all the claims

[Anonymous Coward]

The dimms used are standard 64bit parts, so you need a matched pair for 128-bit access. muchlike an i875 chipset needs dimms to be added as identical pairs.

real world apps

[goombah99]

The keynote address was fairly long so I would guess most slashdot readers actually watched all of it. In it they did on stage examples of tests they did with real world apps.

They showed four top-shelf apps: Photoshop, Mathematica, Emagic, and one other I'm spacing on. In each case the apps were not demoed by mac but rather by someone from the app company. And the examples they gave were clearly practical ones not special cases noone would actually want to do. In the case of Photoshop it was actually a commerical product (movie poster) that was recreated by replaying the artists commands. In the case of the Emagic it was the compositing of the actual musical composition that the musician had done. In the case of mathematical it was the calcualtion of a fractal curve: theodore grey pointed out they had to dumb down the calculations so they xeon would not run out of memory.

in all cases the Apple ran more than 2X faster than the Xeon.

now you could try to say these were tweaked apps, but that wont wash. these are pro-sumer apps that these comanies sell for a living. you better believe that would optimize the heck out of both the wintel and Apple versions. Certianly, if there was any tewaking tobe done they had lots of time and no shortage of manpower and experts to do it on the intel instruction set. Another test they did not demo live was the 40% higher frame rate in Quake

If all they had shown was some single case like photshop or Quake I might have been less convinced. but here are five different genres of applications, in the most demanding fields of Imagery, music, (real world) numerical math, Gaming and others. Okay so your application--say MS word or web browsing--isn't so demanding. That's not the pointis it: you aren't doing things where the machine is the speed limit.

I think its pretty reasonable to assume that over time compilers for the new G5 will imporve more that those for the i86 instruction set since there's new things to exploit. Likewise relatively few compilers do a good job of taking fulladvantage of the Altivec extensions yet. And with the fat, independent pipes to disk, and memory apps will need to be re-written since many of the old bottlenecks they were designed to avoid aren't there anymore

So argue all you want about SPEC tests, but were taking shaving ten or 20 minutes per hour of real world usages. Its phenomenal. In my opinion the diveristy of tests clearly shows the mac is not only the fasest currently on-sale platform, but that there is not even any wiggle room to doubt that.

Re:real world apps

[Andy_R]

They showed four top-shelf apps: Photoshop, Mathematica, Emagic, and one other I'm spacing on. In each case the apps were not demoed by mac but rather by someone from the app company.

Emagic is the software company, not the program, and the fact that their Logic program one was demoed by Gerhard from Emagic rather than someone from 'mac' ( I think you meant Apple!) is a rather dubious disinction when you consider that Emagic is actually a subsidiary of Apple.

Having said that, my contacts in the pro-audio community are hugely impressed by the specs that were being thrown around. Apple's decision to but Emagic and discontinue development on the PC version of Logic was widely criticised, but I think the pay-off of having Logic optimised for G5 will win Apple a lot of sales.

Still stuck on benchmarks?

[xyrw]

I'm surprised that slashdot is still stuck on benchmarks as an indication of processor speed. Hasn't it already been pointed out over and again that it is incredibly difficult to compare across platforms?

I think it is best leave the pointless statistics to hardware fanatics, and use whatever platform makes one most productive. As such, if any benchmark is even minimally admissible, it is `real world' benchmarks. Yet they do not complete the picture, since productivity is a function of other things, such as user experience, planning required (for the type of job), ease of use-- the list goes on, but you get the idea.

After a point, increasing the number of FPS you get in Quake 3 is not going to make it any more fun for you; likewise, beyond a certain threshold, it becomes pointless trying to get those pro tools to run faster.

Re:Still stuck on benchmarks?

[FueledByRamen]

Actually, a higher FPS rating in Quake3-based games (I think the magic number is around 125) lets you jump higher and run a little faster. The key is that the engine physics are computed per frame, and something about the way they're written (maybe a rounding problem somewhere in there, don't ask me) allows for higher jumps and faster movement when you hit around 115 - 125 FPS.

Re:Still stuck on benchmarks?

[dbrutus]

But past a certain framerate I can compile in background and still run Quake at an acceptable speed. Sure, it makes the compile slow down some but if you're going to take a 10 minute break, isn't it nice to be able to get some work done in the background at the same time?

real world apps?!

[andrewleung]

now, how come everyone is just focusing on SPEC benchmarks?! which compiler, what options were set, etc.?!

i saw the keynote, they had photoshop/mathmatica/etc. going on there... photoshop has been out on PC for a while... REALLY enhanced with MMX/SSE/SSE2... and it probably was using the intel compiler... but the G5 version was only a few months old, barely optimized, and using whatever tools apple gave them (probably GCC 3.3)... and the G5s still kicked a lot of ass.

benchmarks are important but it's not my job. if i can get shit done faster in photoshop with BSD guts, i'm all for it.

fuck the benches. welcome to the REAL world...

maybe

[jbolden]

I think its pretty reasonable to assume that over time compilers for the new G5 will imporve more that those for the i86 instruction set since there's new things to exploit.

Actually from an optomization standpoint x86 is pretty new too. What you need to do for Pentium IV (pre HyperThreading) is very different than what is needed for Pentium III and different from what is needed for PIV w/ HT. Further the complexity is so great that compiler science of today is really not up to the task.

Conversely the G5 is much simplir problem due to better design. OTOH it also much newer. It may be that in practice (especially when people are willing to lose 32 bit and/or G3 compatability) you might get some truly wonderful improvement.

So I'm really not sure where there is more room for improvement over time. I just don't think its nearly as easy to say as you had it in the above. In my opinion its going to come down to a political choice regarding the G3s vs. advances in compiler technology.

Re:maybe

[FueledByRamen]

Losing 32-bit compatibility shouldn't be a problem at all. That's the great thing about the Mach-O Executable format (used by OS X) - you can stick binaries for as many different architectures as you want in there. Hell, if windows supported the format, you could stick an X86 and a PPC binary in there and run exactly the same file on both platforms. Ditto for Solaris, Linux, IBM's zOS - you get the point.

My guess is that Apple will make the 64-bit versions of the Mach-O binary loader look in a different place (I don't know how the Mach-O format is organized - the next slot? a different directory tree?) for a 64-bit native version, and fall back to the 32-bit version if one can't be found. The existing loaders will just keep looking in the same place they always have, and see the 32-bit version.

Re:Can't find SPEC results at spec.org for Apple??

[Anonymous Coward]

Because the first thing my Mom does when she buys a new computer is to run over to the SPEC website and check their numbers.

All kind of pointless...

[Arkham]

At this point, does it really matter if Intel, AMD, or Apple is the slightly faster computer?

They're all extremely fast and all run one or more UNIX-like Operating Systems (Linux or BSD or OSX). For the Slashdot crowd, Windows is an afterthought, but I'll mention it as well.

What a person decides to buy is not going to be based on speed anymore. All of the fastest current machines will blaze playing Quake 3 or UT2003.

People who buy Macs may enjoy the speed, but that's not why they buy them. They buy them because they're cool, the have a really nice, easy-to-use, elegant OS that allows them to be productive. Also, they can use the commercial applications (Photoshop, Office, Filemaker, etc) they need on a stable, reliable UNIX platform.

Linux/BSD users have a very different set of criteria. They're looking for cheap, super-secure, stable, configurable or some other particular criteria, but are not particularly concerned with the UI experience or with running commercial desktop applications.

Windows users are a different group too. They want to run their commercial and vertical applications. They are not looking at Linux or Mac because their apps are not there.

That's why there's not a lot of crossover right now between Mac and Intel/AMD. The audience is just different. Thanks to things like Lindows, there may be some Windows->Linux crossover, but this too is pretty small.

Re:This is worse than political campaigning...

[Maelikai]

ahem... the Dell wasn't running Logic.

Re:Not Only Deadlier...

[presearch]

if you can't build it yourself why bother?

I feel the same way about cars, major home appliances, and especially consumer electronics.
Does anyone else smell wire burning?

I gotta go....

Perspective - my 25MHz NeXTstation

[garyebickford]

It's not always the cycles, it's how they're spread around and how you use them.

I still have an original 25MHz NextStation. CPU is a Moto 68040, plus (Intel?) Digital Signal Processing (DSP) chip that does (most of?) the rendering for both the display and the laser printer.

Back in 1999 I compared this box in actual usability with a Mac Powerbook 5300, admittedly the slowest and lamest PPC Mac ever built.

I found that in general usage, opening windows, updating display, doing word processing, etc., the NeXT outran the PB 5300.

Compiling speed sucked big time. Stuff that took a few minutes on the PB5300 ran overnight on the NeXTstation. This demonstrated to me the advantage of having a display coprocessor.

The user interface was also better by far than the Mac that stage. I used several 3rd party enhancements, such as one that provided an infinite-size virtual window, so it's not a completely fair comparison. The NeXT also scame with a bunch of cool apps, like Mathematica, Webster's, Lotus Improv (completely unique approach to spreadsheets, so far unduplicated.)

Most impressive thing about the NextStation was the industrial design. It is still the most elegant design I have ever seen in a desktop computer. For example, the ribbon cables from the mainboard to the floppy and the disk are about 1.5 inches each - just a 90 degree curve, essentially. Those are the only wires inside the box!

I've still got the NeXT, though it's back in the original boxes. I'll probably sell it eventually. I've also got three Perq workstations from 1982-3, but I haven't benchmarked them.

It's worth noting that NextStep's complete object integration across all apps was cited as a major inspiration for Tim Berners-Lee's original proposal for the World Wide Web. In fact, I even have a running copy of that first version of TBL's code, called (surprisingly) "WWW".

I think you missed the clue train.

[phillymjs]

The machine that a dual 2GHz G5 trounced in all the real-world app tests was a Dell with dual 3.06GHz Xeons. Notice, I said "real-world app tests," not the questionable benchmarks. You can dispute the benchmarks, but it's hard to argue the performance differences I saw with Photoshop, Mathematica, etc. The Dell was flat-out dusted.

If a dual 3.06GHz Xeon system was shown to be slower than the dual 2.0GHz G5, please explain how a Dell with only dual 2.4GHz Xeons (which is what I presume you meant) is faster.

The Dell dual 3.06GHz Xeon system has been repeatedly spec'd out in recent /. discussions at ~$4000 in configurations comparable to the G5's. I just did it myself. I configured my Dell PWS 450 by selecting two 3.06GHz Xeons, downgrading to 512MB of RAM, upgrading to a 120GB hard drive (still smaller than the G5's 160MB), upgrading to the cheapest drive that could write DVDs, adding a modem, adding a FireWire card, and subtracting a monitor. Components not specifically listed here were left at their default settings. Final price: $3772.

Since the bone-stock G5 is $3000, please explain how the dual Xeon costing $3772 is cheaper.

BTW, the exact Dell system above configured with 2.4GHz dual Xeons is $2522, not "under $2000" as you seem to have claimed.

~Philly

Re:No Appropriate Fortran compiler

[Jeremy Erwin]

Why don't you read the benchmarks before speculating? Some of the benchmarks are written in Fortran 90. Neither f2c nor g77 (available from fink) support this code, so Apple used NAG Fortran f95 v4.2

Yes, there is a GNU Fortran95 compiler, but it's "in a pupal state."

benchmarks, stenchmarks

[nuckin futs]

Intel (and others) could dispute every benchmark out there, but no matter how fast a P4 or Xeon is, it has one major problem which prevents me from buying one...
It still can't run OS X.
And no...rumors about an Intel based Mac running OS X deep inside Apple HQ doesn't count.

What a joke

[coolmacdude]

I stopped reading the article when I got to the subtitle where it refers to Apple as a "Cupertino Fruit Company." Look, Mr. White, if you aren't even going to show any respect at all and even mock one of the companies in your so called comparison, how do you expect anyone to take you're evaluation seriously?

Real World Benchmark

[inertia187]

Here's my idea of a real world benchmark. Take 75 people with varying levels of technical no-how. Divide them into three groups of 25, and assign various real world tasks.

Obviously one group of 25 is using only the latest and greatest that the wintel people has to offer, while another group is using only the latest and greatest that Apple has to offer.

What is the third group doing? Each person in the third group gets to choose which platform they can use.

All three groups would be given real world objectives. Some would be as simple as writing a report. Some would be as technical as application development. Others would be as pointless as a Quake III tournament. All would be measured for how much time it took to complete, and/or other pertinent measurements to see which platform stood out. This is less of a performance test and more of a productivity test.

What is the third group for? It's the preference control group. Do people really prefer one platform over the other AND are they more productive when they can choose? That's what I'd really like to know. Most companies are dead set on one side or the other (usually wintel). If anyone goes off the beaten path, they are the black sheep.

Personally, I like to work on multiple platforms - some at the same exact time. If the current BitTorrent implementation is better on OS X, I'm using it. If the best IRC implementation is in the X Window system, I'm there. If it's quicker for me to pull up the Windows calculator when I'm trying to convert a decimal value to hex, that's what I'll do. But am I really being more productive (and why am I using BitTorrent and IRC to measure this)?

Re:Can't find SPEC results at spec.org for Apple??

[Quikah]

No you don't.

"Submission to SPEC's review process is not required. Testers may publish rule-compliant results independently. "