Home > Benchmarks > Athlon 64 vs. Apple G5 Systems: Not Even Close Athlon 64 vs. Apple G5 Systems: Not Even Close Submitted by Prognathous 2003-10-14 Benchmarks 75 Comments “Apple Power Macs did well on Photoshop, but the 64-bit AMD-based systems won handily on most tests”, says PCWorld. About The Author Eugenia Loli Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker. Follow me on Twitter @EugeniaLoli 75 Comments 2003-10-14 6:08 pm We should see alot of these test come out before Panther. 2003-10-14 6:16 pm I did not care before the G5 came out, I did not care after the G5 came out, and I certainly don’t care now. at least they used an alienware PC so the hardware quality is comparable. 2003-10-14 6:24 pm Of course, when you are dealing with computers at these speeds.. how much does the benchmark difference matter? I can see it now.. “GEEZE! I’m only getting 200fps in Quake.. I wanted 350!” Er.. I understand the differences for some needs, but for most people these two are close enough to not make a difference. (As opposed to comparing a G4 to a recent P4 or AMD chip..) 2003-10-14 6:25 pm Comparing a duel 2ghz cpu system to a 1 2ghz cpu system in my eyes isn’t that great of a test. The way I understand duel systems is that the job is still assigned to one cpu, so doing rendering in the adobe program would still be using one cpu. Since the arch of the two cpus are very different I expect differences in the specs using the same test. I wouldn’t mind see how they both handle rendering from two or more applications, correct me if I am wrong, but I would think the G5 would do better with a test like that. But I am no hardware engineer so I don’t know for sure 😀 2003-10-14 6:26 pm First, note that the Mac side of the tests were performed by MacWorld, so I’m sure they knew how to set things up properly. Second, I don’t think these benchmarks are indicative of actual performance. There are a couple of factors at play: 1) The code is probably optimized for the G4. I don’t think any of the companies involved have released G5-optimzed code yet. The P4 had the same problem when it came out, and that was remedied eventually. I’d rather see Linux benchmarks, with software compiled in G5 mode by the latest version of GCC. 2) OS X might be holding the machine back. The kinda 64-bit but not really setup might incur a performance loss. Of course, note that all the software is running in 32-bit compatibility mode on the Athlon too, and the Athlon has a higher theoretical performance hit for running in 32-bit mode because the software can’t access certain architectural improvements. 3) The Mac versions of some software (especially Word) are probably less optimized than the Win versions. However, the results aren’t entirely surprising. The Athlon FX is one hell of a CPU, and it has certain architectural features (built-in memory bus) that make it potentially faster than the G5. However, the kind of delta you saw in the article is probably unrealistic. I’d guess a 2.2GHz Athlon FX should be a bit faster than the G5 in integer performance, and a bit slower in FPU performance. The memory-latency advantage on the Athlon FX side, though might tip the scales towards the Athlon, even in FPU benchmarks. 2003-10-14 6:27 pm So the RAID setups were the fastest. So to make the test fair, did they bother setting up any G5 systems with RAID also? No. Gee whiz they put the standard HD G5 systems against the Athlon64 boxes with RAID .. in this case they should compare the standard HD G5 with the standard HD Athlon64. 2003-10-14 6:33 pm to fair. The only two really comparable machines were the dual 2 ghz mac and the alienware aurora with standard hdd setup. Most of these programs and the things they were doing were single-threaded operations which would not have the advantage of the 2nd CPU. This article was obviously posted as flaimbait. Also, Premier sucks, they should have used Final Cut Pro, but alas, there is no PC version 2003-10-14 6:35 pm However, how realistic a scenario is that? Anyway, the benchmarks should be SMP-aware, as you can see from the big performance difference between the 1x 1.8 and the 2x 2.0. Note that Quake III, which is especially SMP optimized, saw a huge boost and still couldn’t touch the Athlon. As for who needs this power — if you’re dropping > $3500 for a machine (welll over $4300 if you factor in a monitor and good speakers) then you want the fastest machine, period. Note also that the Alienware is rather overpriced. Athlon FX’s in general are overpriced, because they’re not sold in large quantities yet. When the Athlon FX hits the Athlon’s volumes, the Alienware machine should go down several hundred in price, bringing the delta between the two to about $500. That delta goes up even more if you don’t buy from some foo-foo company with fancy cases, and buy from Dell instead. It might not be fancy, but Dell will be at your door the next day if something goes wrong, for free. And their default warrenty is longer too. 2003-10-14 6:36 pm RTFA. They did. Thats why there are two rows for the Alienware machine. It turned out to make little difference, because the benchmarks were not I/O bound. 2003-10-14 6:39 pm you know…a month ago you said the same thing…you know what I found when I went looking..you have to pay EXTRA to get warranty where dell comes to your door. so how is it different than applecare? 2003-10-14 6:45 pm Performance does not matter but the price does. The Atlon XP is not tested on the price benchmark so it is useless. Z 2003-10-14 6:45 pm No RAID or not Athlon64 won. Take look again on the table. RAID did not helped that much as systems were tested for cpu/memory. Polywell Polystation Two (Opteron won most – no RAID), Alienware Aurora was set with or without RAID. Besides OS X implementation of RAID are still crappy so it would make it even slower on disk intensive tasks 2003-10-14 6:50 pm The only benchmark even worth considering among those provided is the the Photoshop benchmark. No Mac owner in their right mind would be using Premiere over FCE/FCP/Avid, especially considering that Adobe, being unable to compete with FCE/FCP/Avid, recently dropped support for the Mac. Mac users typically do not care about games (at least, the caliber of professional who will be purchasing a G5 PowerMac as opposed to an iMac/eMac certainly doesn’t), and for those looking for a platform for MS Office a PC is clearly a better choice (and for professionals using Office because they already have a Mac anyway, it certainly isn’t going to be slow on a G5). A better point would be made by comparing the performance of applications which have a predominantly Mac userbase, such as QuarkXPress, InDesign, Illustrator, Painter, ProTools, Cubase, Reason, etc to see if the Athlon FX is capable of trumping the Mac in performance of Mac-optimized applications. 2003-10-14 6:50 pm Panther is said to have significant speed gains over Jaguar. It’s pretty telling that they say “Later this year Apple will launch another OS revision, code-named Panther, with even more 64-bit enhancements.” Try later next week, and that’s about as long as this review will be relevant, if at all. 2003-10-14 6:51 pm even if it is CPU intensive, when you are looking at premier and quake, you will have swapping out to the hard drive. raid makes a difference (depending on the raid set up…if it was striping, then you get a nice swap latency boost, otherwise, it makes no difference in performance. 2003-10-14 6:56 pm I’d already posted this in a different news thread, without much reaction. Noone reading German computer magazines here?!? In c’t issue 20 of 2003 (now 3 weeks old) there is a huge comparison of the G5 to P4, Xeon, Athlon 64, A64FX51, Opteron. They tried to replicate the G5 specs with their other systems, so it’s ATI9600 everywhere, 512 MB RAM, the same Pioneer 106 SATA 160 GB etc. Can’t quote all the results here, would take me days. Just a couple: 1. Photoshop 7 (special c’t macro that I don’t really know anything about, except some plugin is already G5-optimized), seconds: 2xG5 2GHz: 278 1xG5 1.8GHz: 454 1xG4 1GHz: 796 2xOpteron 2GHz: 275 2xXeon 3.06GHz: 287 Athlon64FX51 2GHz: 337 Opteron 2GHz: 366 Athlon64 2GHz: 668 AthlonXP 3200+: 431 P4 3,2GHz: 362 2. CineBench 2003, OpenGL-Shading, score (higher is better): 2xG5 2GHz: 1295 1xG5 1.8GHz: 1014 1xG4 1GHz: 163 2xOpteron 2GHz: 2643 2xXeon 3.06GHz: 2256 Athlon64FX51 2.2GHz: 3360 Opteron 2GHz: 3190 Athlon64 3200+: 3181 AthlonXP 3200+: 2641 P4 3,2GHz: 2643 Obviously for x86 dual processor doesn’t make any sense here; the same is true for the Mathematica 5 benchmark (not quoted here). There are a few more anomalities c’t notes: – the 2xG5 2GHz is 4 (!) times as fast as the single G5 1.8GHz at re-encoding a DVD with DVD2OneX. Prices (Euro, German market) 2xG5 2GHz: 3210 1xG5 1.8GHz: 2560 1xG4 1GHz: ./. 2xOpteron 2GHz: 3700 (sum of parts; made by c’t) 2xXeon 3.06GHz: 3100 (sum of parts; made by c’t) 3970 (Dell Precision 450) Athlon64FX51 2.2GHz: ./. Opteron 2GHz: 2180 (sum of parts; made by c’t) Athlon64 3200+: 1950 (sum of parts; made by c’t) AthlonXP 3200+: 1760 P4 3,2GHz: 2070 (Dell Dimension 8300) Philotech 2003-10-14 7:00 pm I have never owned a Mac in my life and I still say this review was rather pointless. The Word test is utterly worthless on any platform. Premier is pre-G5, and discontinued on that platform. People don’t buy a Mac for games, which only leaves Photoshop which the Mac did just fine on. The real winner was the Opteron, with the Mac just behind it. And none of the platforms, with the exception of the lonely P4 at the bottom are even optimized at all yet. Find some better benchmarks and test again in a year and you might have an good idea of performance of the systems. It’s WAY too preliminary at this point… 2003-10-14 7:07 pm Why is that anytime the little Mac is compared and beated, is wrong for the zealots? RTFA, like Rayiner said. They did compared the Athlons 64 with and without RAID. Also, Athlon 64 does not take a performance hit when is running 32 bit code. Just step out of Steve Jobs Reality Distortion Field. 2003-10-14 7:27 pm So are they going to re test the machines once Panther comes out? Everyone knows that 10.2.7 is only a stop gap until 10.3 comes to market next week. You’d think they’d wait for the OS Apple intended to run on the G5. 2003-10-14 7:27 pm I was disappointed that they didn’t quote the machine’s EPS rating (that’s eggs per second — the number of eggs you can fry (and have reasonably cooked) on the processor in one second). It would be nice to start seeing power consumption listed in these system shootouts. 2003-10-14 7:28 pm if Raid v non-raid makes the comparison unfair then one should point out that the dual-opteron cpu outperforms apple on adobe photoshop dual G5 even in apple’s supposed platform strength, adobe photoshop, opteron dual outperforms apple dual g5 2003-10-14 7:31 pm Photoshop with the same I/O: Dual G5: 16s and 51s 2.2-GHz Athlon 64: 21s and 60s Wow. So, in other words, in an application that is actually designed for the chip at all, with the same I/O, the G5 is faster. “Not even close,” indeed. I find this entire section of the article completely bizarre. Why the hell did they bring the G5 into a review at all, while ignoring, say, Sparc 64 machines? Or Itanium 2s? Or, since you’re not running a database or BLAST (the only real uses for 64-bit code), why not a P4? Moving beyond that…Apple makes setting up a RAID array so ridiculously easy in OS X that they must have *tried* to not test this feature. And why exactly are they testing a program that isn’t even supported by its developer any more? 2003-10-14 7:35 pm However, how realistic a scenario is that? That scenario is realistic to some of the 3d modelers I know, they use different tools depending on what needs to be done for their project, and they normally render 2 or 3 at the same time. This is what I am told from them, they are probally exagerating, I don’t model I code 2003-10-14 7:40 pm @Debman: Hmm, I don’t see where you got that I’m looking right at Dell’s website, and the default option comes with 1-year of at-home service and a 1 year warrenty. The default Apple warrenty is a mere 90 days of telephone support, and a 1 year warrenty. The onsite support is terribly useful — I used it once when I managed to fry my AWE64 by playing with it on a carpet. Wth my laptop, the product quality itself doesn’t seem that good, though, I think that’s a problem limited to the high-end Inspiron’s. Both of my dad’s Latitudes, especially the new C400, have been very well built and trouble-free, and my dorm-mates all have Dimension 8200s (special educational deal) and none have had trouble with it. However, their support is still top notch — they’ve replaced my LCD twice because of scratches on the screen, my keyboard once because a key popped out, and my battery once because it stopped holding a charge. Every time, they sent a shipping box to my door to pick it up, all without me paying a cent. @xnetzero: I highly doubt if 10.3 will help these benchmarks. 10.3 still runs with that 32/64 bit melange. 10.3 is supposed to be faster, but unless they fundementally change the memory model, it shouldn’t affect CPU-bound benchmarks like these. 2003-10-14 7:43 pm RTFA. They *did* test a P4. They *didn’t* test database benchmarks because the G5 is a workstation, not a server. Also, look how much of a jump the Photoshop benchmarks get from having dual CPUs. The Athlon machines were single CPUs. Its quite clear that a single 2.0 GHz G5 would lose to a single Athlon FX51 even in Photoshop. 2003-10-14 7:50 pm I was looking into getting a Dell laptop and noticed that the in home was extra. and here is the Apple warrenty: The AppleCare Protection Plan extends your computer’s 90 days of complimentary support and one-year warranty to up to three years of world-class support. so please stop making up lies the dell site says: Limited Warranty, Services and Support 1 Year Limited Warranty2 plus 1 Year Mail-In Service with the option of upgrading to: 3 Year At-Home Service upgrade now here is the link http://www1.us.dell.com/content/products/features.aspx/featured_ins… the link: 2003-10-14 7:52 pm Using Premiere 6 for closs-platform benchmark is bulls**t. Premiere 6 in Macs doesn’t run on OS X, unless perhaps in Classic mode, which is far from ideal to run a benchmark because it’s an emulation mode. On Windows XP Premiere runs natively. But besides that, Premiere is one of the most buggy programs that ever came out of Adobe, both Mac and PC versions. I used it on both platforms, and it crashes like hell. Photoshop is OK, but testing with Word is also useless because Office for Mac, specially Word, is a chock full-of-bugs. I mean, it’s made for OS X and it crashes as much as the worse OS 9 software. And also it’s pretty different from the Windows version in features and just about everything, so it’s far from ideal for cross platform testing. For that you have to use some Adobe and Macromedia programs, like Photoshop, Illustrator, Dreamweaver, etc. But also it would be better to wait until new versions of software are optimized for the G5, becaue otherwise it’s useless. 2003-10-14 7:53 pm Sure the g5 optimized version of mac os isn’t out yet. But at least the version they tested supposedly helps 64 bit operation. The version of windows they used is “only” 32 bit and it showed the athlon beating the apple in every category. After 10.3 comes out will the macs take the lead? Maybe. But if they do when windows users point out that they aren’t using xp 64 the mac fans will say tough sh*t, ms should develop faster. I guarantee it. 2003-10-14 7:54 pm Reviews are like belly buttons, everyone has one. PC Magazine (the Oct 28th edition) also reviewed the G5 up against a Dell Precision 650 and the G5 won 3 of the 5 tests that were head to head. I am going to look around their website and see if it’s up yet. If it is, I’ll post a link later. Aaron 2003-10-14 8:07 pm As many posters have noted, these tests prove nothing because of the apps picked (besides photoshop). Premiere is not even made for the mac anymore. Why not use at least the new version of After Effects? Word is buggy and slow on the mac and suffers from the conversion. Most games are still written for windows and directx and then ported to the mac, so they suffer from conversion and lack of optimization too. 2003-10-14 8:09 pm The version of windows they used is “only” 32 bit and it showed the athlon beating the apple in every category. Well, not quite. Here’s the categories the Athlon 64 had a definitive lead in: Quake 3 Performance Word Performance Premiere performance These are 3 categories which are completely irrelevant to the pro Macintosh user, the type who would be purchasing a G4 PowerMac. As I said before, here are some applications that would matter to the pro Macintosh user: Photoshop QuarkXPress InDesign Illustrator Painter ProTools Cubase Reason DVD Studio When I was on a newspaper staff one thing I heard from every photographer and graphic artist whenever the issue of Photoshop on Windows was broached was how much they hated the interface simply for the pointless gray MDI background which obscured the desktop icons. On a Macintosh the desktop icons remain visible because of the shared menu interface. That’s just one of the many interface issues on Windows which keep the majority of graphics/preprint/audio professionals on Macs. 2003-10-14 8:10 pm the type who would be purchasing a G4 PowerMac s/G4/G5/ 2003-10-14 8:17 pm As others have pointed out, the numbers cited in this article don’t tell us much: – MS Word is a dog on OS X (not too surprising, considering who makes it). Moreover, Word is going to run “fast enough” on any of the tested hardware. – Premiere for Mac is discontinued. Mac users use FC Pro. – As for Quake, PCs are simply better for games. I’m a huge fan of OS X, but if I cared about having the absolute best frame rates in games, I’d get a PC or a console. I hope no one is buying a 3000-4000 dollar machine just for games! Photoshop is a valid real-world benchmark. Unfortunately, the total time taken by 10 filters doesn’t tell us much. First, they don’t tell us what filters they used. If we assume they used the ones from PSBench, there’s still a problem. The PSBench set of filters include a few uncommon ones (like Watercolor) that take a very long time compared to the more common ones (like Gaussian blur), and seem to perform much worse on the Mac (probably because, being uncommon filters, Adobe hasn’t bothered optimizing them for the G5 yet). That’s why people usually use an indexed score for the PSBench test, rather than total time. For example the first 9 filters might take machine A .5 sec each, and machine B 1 second each. But that last uncommon filter might take machine A 100 seconds and machine B 94 seconds. If you go by total time, machine B wins, but would say it’s better to use for Photoshop? I’m going to wait for more numbers for more apps by more thorough sources before passing any judgment. I won’t be at all surprised if the Opteron is faster than the G5 for more than gaming, but articles like this don’t tell us much. And considering that I really like OS X and have no need for ridiculous speed, I’ll stick with my lowly old powerbook either way 2003-10-14 8:33 pm I’m disappointed that Apple would claim that the G5 is the worlds fastest when it has been proven to be false time and time again. Whenever anyone points this out people think they can invalidate the facts by saying things like “that software sucks, nobody uses that anyway, wait until the next version of the OS”. If you don’t like the software fine; does it change the speed? Of course not! I’m surprised Apple can make such claims with every processor they use and people still swallow it. You can find some great history at http://www.netherworld.com/~mgabrys/clock/back.html It’s a long read but boy is it funny! The part about the first iMacs is certainly the best! 2003-10-14 8:38 pm I’d like to see some more hands-on benchmarks, such as how long does it take for a user to change screen resolution in WinXP or MacOSX. How long it takes to find a specific document on the machine, how long it takes to configure a machine to connect to a LAN or the internet, what happens if the power fails. I recently replaced my Celeron 366 PC with a 2,6 GHz one, and I don’t feel particularly much more productive. It doesn’t help one bit, if your slow grandma is using a 128-bit quad-CPU system or your average 500 MHz P3 and can’t saturate the CPU power on the machine. It doesn’t help one bit, if you’re a fast-moving professional with a monster workstation, but you’re spending more time learning how to use a specific, complex program than actually producing something. I’d say we got the hardware now. It’s time to nurture the software. 2003-10-14 8:40 pm Panther/10.3 will probably help the benchmarks more because of overall speed improvements than because of specific G5 optimizations. Even lowly G3 users have reported 25-40% speed improvements for many tasks from Panther. And since none of the benchmarks likely used 64 bit operations I am not sure why that is relevant to the discussion here (other than the fact that the article was about 64 bit processors). Regardless, it won’t be the 64 bit code in Panther that will result in most of the improvement. I think this is a silly pissing contest anyway. 2003-10-14 8:42 pm I’m disappointed that Apple would claim that the G5 is the worlds fastest when it has been proven to be false time and time again. Whenever anyone points this out people think they can invalidate the facts by saying things like “that software sucks, nobody uses that anyway, wait until the next version of the OS”. If you don’t like the software fine; does it change the speed? Of course not! I’m surprised Apple can make such claims with every processor they use and people still swallow it Mabe you have not read all the post on this form. Premere IS NOT MADE FOR OS X. It would have had to have been run in CLASSIC Mode (emulated enviornment) Want a compairison? Open up Virtual PC (yes, its made for windows) start Windows 98 – then benchmark Premere. Read the facts before you post, pal. 2003-10-14 8:48 pm Marketing in this country doesn’t have to be honest. Does Dell really make high quality computers? Not according to most studies!! Do Dell night tech support actually work in the US and speak English as their first language? Not to my knowledge! “Best” is a matter of opinion “Fastest” can only be determined in the context of “doing what?” It’s called marketing people……don’t take it personally. 2003-10-14 8:48 pm If they’re going to compare Premiere, why not compare Logic Audio as its a resource intensive application, and it is not developed for PC anymore… instead only being developed for OS X and linux. 2003-10-14 8:51 pm Windows Logic Audio against OS X logic audio 2003-10-14 8:56 pm Will any mac user change their environment to windows because photoshop or some other app is a second faster at doing some actions? Nope. Imho, benchmark values this closely matched are nearly exclusively useable for machines doing numbercrunching or dedicated gaming boxes. The only thing these numbers tell you right now is that macs at this moment are at least comparable to their windows counterparts, which makes them a reasonable option for your workstation needs if all the apps you need are available for the platform. 2003-10-14 9:04 pm Premiere is a discontinued app and isn’t even optimized for G5 like Photoshop. So unless Adobe ships a G5 optimized version of Premiere I don’t think its fair to use it a benchmark for a G5, it only makes sense. What the article failed to mention was that in Photoshop the G5 beat all of the Athlon64 systems and was very close to beating the Opteron. PC World concluded that the G5 was slower than the Athlon64 by using a discontinued non-optimized app and a Microsoft product made for Mac. They descibed the Photoshop win as “sparkled” Yeah sounds fair to me. Why not use a half gig poster file with an action and FCP versus Premiere or Avid on both systems. Another anomaly was the Opteron beat all the systems but was slower than the G5 in Quake? What gives? 2003-10-14 9:09 pm is that in spite of great story about fastest desktop ever it is simply not true (I had opteron before G5 was launched). When Apple released the results it was what 2x or 3x times faster than PC? Well G5 is not faster newer was. It is on par with AMD/Intel. There are apps that G5 optimised and those that are designed for AMD/Intel. Thing that I dont like is that Apple is making outrageous statements. Not only about benchmarks. I remember complete idiot from Apple who endorsed /dev/null add (end of UNIX OS X is coming). There is more like that. What suprise me is that people actually belive in this. Arguments about premiere are as stupid as BLAST test. Some time ago there was BLAST benchmark: Altivec optimised against not optimized at all BLAST on PC. Now that was o.k. G5 with OS X are good machines but they are far from best in crowd. 2003-10-14 9:18 pm well, the actual Panther and whatnot whining. Deal with it, it was tested with currently available products. If Panther comes out, test again. At the moment, the G5 loses horribly. Admittedly, if the G5 still loses when Panther is out I’d repeatedly smack my head against a wall if I had bought one. 2003-10-14 9:30 pm I don’t care whether or not the G5 is the fastest. I just can’t believe a magazine like PC World would not wait a month so they could use Panther in their tests and article. It doesn’t make any sense. It would be so much more clarifying and intresting if they had waited. 2003-10-14 9:32 pm “If Panther comes out, test again” IF? It comes out next week. Remember, we’re talking about Panther, not Longhorn. “well, the actual Panther and whatnot whining” “I’d repeatedly smack my head against a wall if I had bought one.” Sounds like you already have. 2003-10-14 9:45 pm People relax, its just a test using existing software and OS’s against dual G5 and AMD64. Both processors were handicapped in this test. Wait till next spring when a proper test can be done when both processors are not crippled and enough optimized software is availible. That means a Dual 2ghz G5 CPUs with OSX 10.3 vs AMD64 FX-51 CPU with Windows XP 64-Bit Edition for AMD64. The test can only use software thats optimized for both CPUs. This test should included Mathematica, a 3D software package(Maya is fine), a DTP package(Indesign will work), Photoshop, a Audio software package(must be availible for both platforms), a video software package(must be availible for both platforms), a Illustration software other than one from Adobe(ie CorelDraw or Macromedia Freehand MX), a MySQL test, a older 1st person game(Americas Army), a newer 1st person game(Half Life 2 or Doom 3), and a simulation game(these task the CPU harder rather than the GPU). If the the software is not optimized for both platforms, then the test must used non-optimized software for a fair and balance test. This test has to contain software performances that give an assement of speed and power for all users, not just pro creative users or gamers. 2003-10-14 9:53 pm Wouldn’t it make more sense to pit dual Opterons against dual g5? Those are actually more or less comparable. With FX against 2xG5, you would be unable to explain if the speed difference in some test is caused by some test liking/disliking SMP or the actual speed difference of individual CPUs. If you would like to test FX-51 (which is just rebranded Opteron 148), a single G5 machine would be more appropriate (take 1.8 or remove one of CPUs from dual 2.0) 2003-10-14 9:55 pm > …they hated the interface simply for the pointless > gray MDI background which obscured the desktop icons. > On a Macintosh the desktop icons remain visible because > of the shared menu interface. That’s just one of the > many interface issues on Windows which keep the majority > of graphics/preprint/audio professionals on Macs. I prefer the Windows way, because it keeps the focus on the application. It gets distracting when there are other windows fighting for attention. Actually it’s the main reason why I don’t like Gimp. My guess is that people who dislikes MDI don’t have more than a few apps running at the same time. Back to the main topic: Comparing one processor against another based on performance alone is not interesting for me. I’m more interested in the performance compared to the cost. Processor speeds, or how many they put in a box doesn’t matter, if the competition can deliver better performance at the same (or cheaper) price 2003-10-14 10:11 pm and since we know that the FX is faster then the p4 3.2 in most things, this means the G5 is really nipping at the P4’s heals. all in all, fast computers for every crowd. i’d probably go with a dual opteron first, dual g5 second, and and a distant 3rd, a dual zeon. 2003-10-14 10:13 pm “I prefer the Windows way, because it keeps the focus on the application. It gets distracting when there are other windows fighting for attention. Actually it’s the main reason why I don’t like Gimp. My guess is that people who dislikes MDI don’t have more than a few apps running at the same time.” No, designers are usually working with Illustrator or Freehand, ImageReady or Fireworks, and Dreamweaver at the same time as Photoshop. And they need to see images, updates, etc… in both apps at the same time. Much better to actually see them than to try to size the application windows so you can see both apps side-by-side. 2003-10-14 11:14 pm When Panther and Photoshop CS hit stores, then I’ll call it a real test. 2003-10-14 11:46 pm Arguments about premiere are as stupid as BLAST test. Some time ago there was BLAST benchmark: Altivec optimised against not optimized at all BLAST on PC. Now that was o.k. The argument that because someone else years ago approved of something that is similar in nature to what I am objecting to does not hold water. I am not an amorphous “Mac Users” entity. 2003-10-15 1:50 am At least in the minds of the PC driven press and users at least. Forget about them. 2003-10-15 2:51 am Just compare this: PS7Bench test, here is the total time: Pentium4 3,2 EE (2MB L3): 112,2 secondes Athlon 64 FX-51: 116,2 secondes PowerMac G5 2*2GHZ: 95,1 secondes And for the normalized scores see here: http://www.geocities.com/sw_perf/PSBench.html The G5 is a monster as you can see. A more complete list is available here: http://www.xlr8yourmac.com/G5/Dual_g5_9600_9800_tests.html So how is the faster? And indead not even close. And how many serious test will use Premiere on Mac. This software has never been optimized on mac, and moreover it does not exist any more on mac, so this test is meaningless. And word!!!! What this, every one now that if office on mac is great in his features and interfaces, it still lacks a good optimisation, so again its unfair to use this software to test both platforms. The results foe Quake 3 are wrong. The dual G5 scores at 337 fps for the 1024*768 resolution according to apple or 314 fps according to this site: http://www.xlr8yourmac.com/G5/Dual_g5_9600_9800_tests.html with the Max quality test. The G5 scores at 448 fps for “fastest” setting. But it true that Quake on the G5 does not give good results at 1600*1200 resolution, with a 181,7 fps. But i guess that quake is not optimized yet for the G5, so better results will come. If they want to test fairly both chips, they have to use a more complete set of tests as scientific, mp3 and video encoding, and 3d applications. But because its PCworld, i guess we could not expect fair comparison!!!. 2003-10-15 2:53 am They tested whatever is common for both platforms. See I dont care about Altivec optimised BLAST (althougt I use BLAST daily). If you look ate the tests they were not optimised for G5 or 64-bit AMD with true 64-bit OS. Stop whining. Beside, I prefer long term benchmarks. These are more informative about real capabilities of OS than PC World benchmarks or numbers provided by Apple tests. If you want real test then set up 200 virtual servers on XServe. See what will happen (hint: will die in short time). 2003-10-15 3:02 am Pentium they tested is not extreme edition that is Xeon which is not the same. Also they used dual G5 against single AMD. I don’t see any specs except CPU so not only biased but also not informative. As you can see results favor Apple if made by Apple or PC. So calm down. 2003-10-15 4:01 am PC World should be called Windows World : here is a magazine which title begins with PC yet, instead of using linux or the *BSD for their tests, they preferred to stick with Redmond. Though 64-bit free operating systems exist, they choose to “take a peak at a beta of Windows XP 64-bit”. Intel is unable to produce a 64-bit chip for the desktop. They can’t admit it and resort to stuff like renaming a Xeon “Pentium 4 Extreme Edition” just to rain on AMD parade. It’s quite lame to pretend that nobody needs a 64-bit processor today. Apple has been clamouring for years that their computers are the fastest in the world : they know they’re lying and those who perform benchmarks know it too. With this kind of attitude, they’ll never get my money. 2003-10-15 4:02 am The benchmarks are interesting but they don’t have a ‘bang for your buck’ ratio. In other words much cheaper systems offer about 80% of the performance for 20% of the price. A decent Athlon 2400+ system will set you back about $700. In 12 months you will get the sort of performance of the high end machines machines for a fraction of the price – less than $1000. The ‘excuses’ from the Mac camp are pitiful. Windows XP isn’t optimised for the Athlon either so stop complaining. G5s don’t offer SATA RAID so don’t complain about the lack of that either. 2003-10-15 5:10 am When it comes down to it I wouldn’t mind replacing my slow old putter with any of those in the test; but, going by past experience, whenever I can afford to buy a putter I know the most bang for the buck wont be a Mac. Hardly cast in stone, even if it seems to be, so be it……… 2003-10-15 5:10 am No my friend the scores for the PS7bench come from different source. The Pentium 4 EE and the athlon 64 Fx scores are from a pc site who compored the two processors. If i can find the URL i will give it, but it is really the Pentium4 EE and the athlon64 FX which are tested. The results for the normalized scores come from diffrent source. Why you don’t see any spec. And the PS7bench is chicken or what? There is nothing biased there, or maybe you want to see it biased. For quake results, I gave the score of apple for information, and gave as well results from a different source. Yes they use dual G5, but so far we don’t have results for dual AMD, and it confirms that the dual g5 is very far faster than any single Intel or AMD configuration….and i don’t see a dual AMD really beating the dual G5. And moreover because the dual Xeon 3.06 ghz is not as performant as the dual G5, i think it is also the case for a dual opteron, which does not give better results than a dual xeon. 2003-10-15 5:13 am umm..can I just ask…what exactly are you complaining about? first you complain about the test not using a 64 bit open OS then you complain about Intel chips for what ever reason then you claim apple are a bunch of liars anyway. so, are you saying the test isn’t fair or are you saying the test was a good test that just proves apple are liars? BTW, if a company has a marketing department, they are liars. you just happen to agree with one liar and none of the rest. 2003-10-15 5:20 am If you would ever run BLAST then you would know why I dont care (The advangtage is simply too small, and this is only small portion of the analysis, for the bigger part I use DNAStar). By the way demban what is it with this car fixation? Now Sampras is something new. Still quite naive. Again: these apps were neither optimised for G5 nor for AMDFX/Opteron with 64-bit OS. I also explained that there are better tests in my opinion than simple benchmarks. 2003-10-15 7:31 am Once again, I think you have me confused with someone else. Or maybe you are lumping me into the vauge “All Mac Users” entity. Perhaps a more careful review of my posts would make you understand that my post was theoretical in nature and not based on any specific objection I have made in this thread. 2003-10-15 7:34 am BTW, if a company has a marketing department, they are liars. you just happen to agree with one liar and none of the rest. Perhaps you are unfamiliar with the concept we have used in the last millenia called “science”. One principle is “independant verification.” I don’t have to just believe what the church of Bill or churche of Steve tells me. I can go out and check it for myself. It’s called “thinking for myself”. You ought to try it sometime! 2003-10-15 1:00 pm that is nice, but I was making a comment about how the person is not buying a product because the company is trying to sell something (i.e. lying if that is what you want to call marketing). with that point taken in, my statement does not say “only listen to what the cabal says” it makes a direct point to the original poster that all companies are liars in his/her terms then. as far as independent verification goes, there are good ones (Nasa, PC mag, etc) and bad ones (this one has MAJOR flaws in their methods such as using software the does not perform well on OS X but does perform well in windows). the real SCIENCE in all of this is to have enough critical thinking skills to take yourself out of the fray long enough to think about facts rather than your opinions and decided if the test was really a fair test. in most cases, there are flaws in all of these tests. some however are better than others, and this one here is definitely not a well run test. 2003-10-15 1:28 pm DJ Jedi Jeff (IP: —.rocsth01.mi.comcast.net): my apologies. debman (IP: —.cable.mindspring.com): 1. “as to my analogies…what the hell would you like?” – they are naive? and do not bring anything to the discussion? 2. As I said neither (PC World or Apple’s) benchmarks are informatve. This is third time I am trying to explain how real tests works. Don’t be so excited. Same goes to Hakime (IP: —.sys.hokudai.ac.jp). You both are willing to bend over to find any type of superiority of G5 over other CPU’s. That makes whole discussion pointless. 2003-10-15 2:08 pm Is that Steve jobs opened his big mouth making incorrect claims he can’t back up. The end result is both of these systems will not realize their full potential once they have 64-bit operating systems and 64-bit software.Even with that the IBM PPC processors are released at such a slow rate they will be behind PC processors again in a few months. This is a worthless debate. 2003-10-15 2:49 pm Is this the best PC World can do? They didn’t even use the current version of Premiere for Mac! MS Word as a benchmark program for a Mac? Thats fair too! There was no mention really of how the G5 beat all the Athlon64 systems and was 1 second off the Opteron in the 50MB test and 4 seconds off on the 150MB test. Upgrading to Panther alone may improve those scores. Also they never showed what filters were used? Why not? I bet none of the filters favored MP. Why couldn’t they have used Avid versus crippling the G5 with Premiere or at least use a current version which is 6.5 not 6. How come no mention of how the G5 is a better Quake machine than the Opteron? Why no mention of price on the Polywell? Are these scores valid? Even the Photoshop scores probably don’t take into account MP capable filters which I know were probably not used. Its also interesting that PCWorld didn’t use Panther which will also make a difference but they did use a 64-bit version of XP. Is this the best PCWorld could do? 2003-10-15 3:18 pm First of all, Panther isnt out yet. Second of all, Adobe Photoshop came well out of this test. But it is also the only application that has been optimized for the G5. Everyone will see at better performance under Panther, when optimized apps start to arrive. The problem isnt in the CPU, but in many of the programs.. Like in the game-section. MacGames appear like shitty copy-cats because of the rushed-porting.. If those who had ported the games used their time better, the results would be alot better. 2003-10-15 3:21 pm On the Multiple Document Interface: “I prefer the Windows way” That’s great, because there are few who agree with you. Microsoft does not endorse, and advises against, the MDI anymore. Granted, they’re not pushing developers toward a floating interface like Photoshop for OS X (or the Gimp), rather toward a tabbed interface, but regardless, MDI is clumsy, awkward, and takes up valuable pixel real estate. 2003-10-15 4:34 pm AMD: Nice job. If I were a PC user, this is definately a machine to consider. But, I prefer OS X. Great ease of use, and cross platform development. It’s time Apple started considering putting their GUI and IApps on top of Linux and started selling thier OS on two processors, with a restricted set of hardware so that they can give the same quality experience you get on Apple hardware. Uh, that probably won’t work. So, apple seems to be stuck at producing BMW’s vs. Chevy’s. Price Objections: Every time someone compares Apples to PC’s, they leave out features they don’t think they want or need. If you make a serious effort to add all the features of the G5 into a PC you get a HIGHER PRICE. Apple has a price point for everyone, except people who Only want a fast CPU and nothing else, because Apple bundles it’s fastest cpu’s into the PRO line that offer more FEATURES. In the real world this is better value for most consumers, and developers. 2003-10-15 4:44 pm It doesn’t really matter to me which is faster. A Chevy may be faster than a BMW but it doesn’t have the quality or refinement of a BMW. There is a place for both of them though. Even if both were priced exactly the same. Just like a car, truck, or a Jeep Grand Cherokee Limited like I drive, what you drive should be YOUR CHOICE and not someone else’s choice for you. Again. All that really matters is that there are choices and you get to choose what you get and not somebody else. Nothing else really matters because market pressure makes quality and performance rise and fall. 2003-10-15 8:58 pm I really don’t care if the A64 was 2/10ths of a second faster. I have had enough of Windows nightmarish OS. Just had another disaster upgrading to XP. I am through with Microsoft! 2003-10-15 9:20 pm > That’s great, because there are few who agree with > you. Microsoft does not endorse, and advises against, > the MDI anymore. Granted, they’re not pushing developers > toward a floating interface like Photoshop for OS X (or > the Gimp), rather toward a tabbed interface, but > regardless, MDI is clumsy, awkward, and takes up > valuable pixel real estate. I agree that tabbed interfaces are better than MDI, because it’s easier to handle multiple windows. But given a choice between floating interfaces and MDI, I prefer MDI.