The benchmark at Barefeats — Power Mac G4 and G5 versus Pentium 4, Dual Xeon, and Dual Athlon — has being updated to include new information and tests.
Benchmark: PowerMac G4/G5 Vs P4, Dual Xeon & Dual Athlon
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Opteron? Duh!
We’re comparing a bunch of 32-bit CPUs against a 64-bit CPU. What does that tell us? Absolutely nothing!
What does it matter if something is 32bit or 64bit? As long these devices are made to do the same thing (computing) the fact that they are using water instead of silicon as their base, does not matter. What matters is what performs better, this is what people are after when reading a benchmark.
I’d love to see a comparison between Apple’s G4 and AMD’s Athlon 64 (when it is released).
More importantly though, and this is something that most benchmark “duels” seem to forget about, I’d like to see a comparison of the speed and feature set of an AMD, Intel, and Apple computer than all cost about the same amount of money.
In other words show 3 computers that all have about the same video card, ram, HD, etc., but adjust the cpu and motherboard based on cost.
This wouldn’t compare an Athlon 3200 to a P4 3.2 GHz, but would instead compare a $200.00 Athlon to a $200.00 P4.
I think that is the farest comparison.
Well, some of what I just posted doesn’t add up, but I think the main point about comparing computers (and CPUs) witht he same cost, holds up.
Sometimes, when you buy a computer, the cost of it isn’t the whole story. I purchased matching 17″ LCD iMacs for my wife and myself last November (2002), mostly because of the OS (OS X is really nice, and this is coming from a Mac hater!) and the hardware. These little iMacs probably won’t play the latest and greatest 3-D games, but hey, that’s not what we’re into. We use them for everyday, work-horse computers, and they just work. We never shut them down (hardly ever have to reboot them, except for the odd Apple update that forces a reboot) and they just keep working. While I could get similar uptimes and reliability from Linux (in fact, I have several Linux computers, running Debian, and like it very much), the desktop still isn’t quite there yet. The OS X desktop experience is the best I’ve ever used.
I’ve owned many computers over the last 21 years, and I can say that they are, without a doubt, the best built and supported computers I’ve ever owned.
While I have found OS X to be technically a great OS, I have noticed it is a bit slow from time to time. From what I’ve read about the upcoming Panther (10.3) release, it should fix most of the remaining problems I have with OS X.
Oh, and they have the best display I’ve ever used also! You should really try playing a DVD on a wide-screen display! Nice 😉
Yeah, that would be nice, but it would also be problematic. In the PC world, a good performing system is a balance of components. The best priced hardware depends heavily on the application you are using.
For scientific computing (or a CPU benchmark), I would use a crappy graphics card and the best CPU I could afford. It would produce good results and would lower the system cost. For a game, I would put in a $400 graphics card (probably at the cost of getting the best CPU).
I’d be happy just to see benchmarks given for machines with a complete description of the hardware/software used. I’d also like to see it from multiple independent sources so that deliberately skewed results are easier to see. I can make up my own mind about the “value” of the components (based on what I want to do with the hardware).
P.S. The Photoshop 7.0 MP and non-MP test results seem kinda funky. Are these actions different in nature in some way other than MP-awareness? It just seems strange that the MP G5 and Xeon perform so closely on the MP test, but the G5 destroys the Xeon on the non-MP test. This seems even more strange since I’d expect the G5 to MP scale better than the Xeon given the independent FSBs of the G5. Can anyone explain this?
Let’s not get into one of those “Apple is better but more expensive” vs “x86 is better and cheaper.” Because those always just end up being flame wars. I completely agree with Captain N. If you were comparing cars, putting a lexus vs a honda civic vs a bug wouldn’t really work, because although they all are made to do the same thing (get people from point a to point b) they are made differently and to do it in different ways, plus it wouldn’t be a very useful “benchmark” for buyers, because some could only afford the lowest car anyway. It’s the same with computers. It’s the same with everything actually. You compare like things, and since cost is the main thing all three have in common, it makes sense to use that as your “control” category and everything else can fluctuate
Jesus Christ! Of course it matters if a computer is 32-bit or 64-bit. Except I’m a lunatic I wouldn’t purchase a 64-bit PC for my family. A benchmark that compares a 32-bit system and 64-bit system only serves to distort the perception of computers the general public.
If I didn’t know any better, based on the benchmarks, I’d be looking at purchasing a 64-bit PC thinking it is better and faster than its 32-bit counterpart. We all know that is false. The point is, you don’t compare apples with mangos.
Comparisons should be based on even and fair grounds and should be conducted scientifically. It’s like comparing a 200 horsepowered with an 800 horsepowered vehicle and telling people what does it matter.
Regards,
Mystilleef
The stated Mac OS X was the 10.2.7 release. The latest bata 10.3 is about 30% faster. If we assume only a 15% application improvement, then the 2GHz G5 is:
Cinebench = 56, same as the Athlon 2.1GHz
Photoshop(MP) = 24, best result
Photoshop(non-MP) = 48, best result (both ways)
Bryce = 18, best results (both ways)
Cinebench(software render) = 449, closer to the Athlon.
I think they should retest using the 10.3 bata. The bigest problem is that they are using a non-optimized OS for the G5 tests. Since 10.3 is a good 30% faster for things like Memory allocation, task switching, video graphics, and disk IO; I don’t know why they aren’t using it.
It all depends on what you’re comparing and for what reason.
-if you plan to spend $X on a CPU, comparing two processors that cost $X is a valid comparison, no matter what the clock speeds are, or if one is 64-bit, etc, etc
-if you’re comparing based on user-friendliness, again the actual processor inside doesn’t much matter
It’s like comparing a 200 horsepowered with an 800 horsepowered vehicle and telling people what does it matter.
What does it matter, if the 800hp vehicle weighs so much it can barely move? Or if the 800hp one costs $20M? Or if the 800hp one breaks in half after the first use?
Sometimes one number alone isn’t everything….
How can this be a fair comparison when the system have different amounts of RAM (some with 512 … some with 2GB)???
CB isn’t optimized for the G5 yet. I recall coming across a forum posting somewhere from Maxon saying that in their internal builds which *are* optimized, they’re getting scores of “over 500” on the G5, although he didn’t specify which G5, IIRC. (And I presume he means the “CB Rating” chart that they have at the bottom of the Bare Feats page where higher numbers are *better,* rather than rendering time!)
>How can this be a fair comparison when the system have different
>amounts of RAM (some with 512 … some with 2GB)???
And to everyone else complaining about the scientificness (or lack thereof) of the tests… the website clearly states that they are a low budget outfit without the means to go buying or borrowing the latest, greatest hardware out there. Honestly, only a preponderence of multiple benchmarks on different OS’s and different hardware will give you any indication of how fast a particular system really is compared to others. This is just one tiny (and early which is why it made the news) bit of information. By the time we have enough information, noone will care because something faster will be out and we’ll by complaining that some shmo in Sheboygan who was lucky enough to get an early delivery of the AMD 128 bit SuperDooperon, or a 20 GHz G7 and posts his benchmarks to the neurodomain didnt do his test scientifically enough.
These are useful data points. Obviously, one is dealing with moving targets. I, for one, am pleasantly surprised that the much maligned G4 holds up so well. I also expect the G5 to pick up speed compared to the Intel and AMD chips (at least until the 64 bit AMD chip goes into production).
I, for one, am pleasantly surprised that the much maligned G4 holds up so well.
Why would you be happy – the G5 is suppose to be significantly better than the G4, not marginally.
Comparing an opteron to a g5 or an xeon or to just about any cpu is fair as long as the same test is run. plus, the opteron will replace all of amd’s old processors. its not like they’re running doom3 on all processors and running quake 1 on the g5 and saying SEE! g5 gets a higher frame rate!
anyway, I want a g5. cant wait till i have the dough.
You are wrong. First of all have you seen this 30%. Second of all if I would use your extrapolation method then for opteron 246 results would be as follows (in similar tests opteron is 30% faster than xeon 3GHz):
Cincbench 35.7
Photoshop(MP) = 18.2
Cinebench(software render) = 675
But I think that this type of predictions have no value. So lets wait and see.
he he .. very funny!
I like what Apple has done with the product line. They make some of the best looking systems and without a doubt the best looking OS there is.
I do however have a serious problem with their marketing team.
They always spout this “The World’s Fastest Personal Computer…” stuff, when in fact, and the benchmarks prove it, they are *not* the world’s fastest.
Sure, they are great performers, yes, they look good and hell yes, I would not mind owning one at all.
But please, Apple, enough with the marketing lies. It’s insulting and I don’t think anyone in this day and age will take that statement seriously any more after years of the “MHz Myth” ad campaign.
Anyone notice this line… or am I reading it wrong? (from barefeats)
“The Apple Power Mac G5 models were borrowed for testing from anonymous sources. They were running Mac OS X (10.2.7). Photoshop had the new G5 plug-in in the case of the 1.6 and 1.8GHz machines.”
Does this mean the dual 2 G5 DID NOT HAVE THE PLUG-IN? Doesnt photoshop show huge speedups using this plug-in??? 100%?
I agree with Grusic. You can obviously benchmark for different things. But when Apple is touting their computer as the world’s fastest, then you compare the speeds! This is what annoys me: when mac-fanatics tell me that using the same compiler is fair because it’s one less variable. Well, we don’t limit them to the same transistor count or architectures either now do we.
Micmoo: There’s no use making speculations like that. 100% faster how? And who’s to say that it didn’t have the plug-in? Etc…
Good point, about the Photoshop plugin. That would make a huge difference. And only a slight increase in speed would push the dual G5 past all the others.
” Micmoo: There’s no use making speculations like that. 100% faster how? And who’s to say that it didn’t have the plug-in? Etc…”
They specifically said they only used it on the two lower models.
is probably gonna take a lot of wins in many benchmarks like this.
but it looks pretty balanced now, with strong performance by all 3 platforms (opteron included if you interpolate other benchmarks)
What is the point of these tests? Talk to anyone who uses a Mac and sure, performance is important BUT it is only a small part in a bigger equation.
Up until the G5 was released, yes, there were people leaving as many of them weighed up the downside of the PC with its performance advantage, however today, there is a justification to by a Mac over a PC. I am not going to get into specifics, however, it is the old story of using the right tool for the right job.
Every vendor has its niche, and instead of PC and Mac fanboys trying to convert people, how about a little acceptance that other people may value certain aspects of their platform more highly than other parts.
well, what people miss is that this is a real world test. Actual machines with the software currently available. And what does it show? that the G5 by far doesn’t have the general superiority which is claimed by Apple. It has caught up with x86, but didn’t manage to surpass them yet.
Opteron chips and M/Bs are readily available yet G5s are always compared with (not even the fastest) Intel and AMD 32 bit CPUs.
A cynic would think Apple is too scared to be tested against Opteron machines!
The barefeats article is not sponsored by Apple.
Opteron systems are not readily available where G5s are sold or will be sold like CompUSA and Best Buy.
Apple totes the G5 as a personal computer. You can see it in the advertising all over the place.
I don’t think Apple considers the G5 in the same class as the Opteron if they are even considering the Opteron at all.
Regardless the real speed demon is the dual G5. The lower end G5s should do well on bandwidth intensive operations. You really need a 500MB+ file in Photoshop to really stress these systems.
The Opteron isn’t a personal computer chip because Windows does not yet support 64 bit addressing. Linux is the only OS that can use the Opterons power, and until Linux doesn’t require the terminal for anything it won’t be a personal OS. This isn’t meant to start a debate on the goods and bads of command line interfaces or the terminal in general, I am just saying that most people either won’t or can’t use one, and Linux still requires it’s use, although less now than ever before.
Just heard from Rob at BareFeats about his testing. he told me via email that the machine he had access to (the dual G5) did not at the time have the Adobe PS plug in installed!!! It was not yet available. It is very possible that the machine he was benchmarking was a preproduction unit. Rob was not sure what exactly the adobe plug-in improved, and was contacting Adobe. I had heard it has major effects on performance, but have seen no test to proof. A friend in the Video app dept at apple has been using a pre-production dual G5 for months, and reported that due (oddly enough) to bugs in the cooling system and fans, his machine often ran at greatly reduced speeds… so who knows exactly what has been tested in these benchmarks. Rob from barefeats is a super nice person and promptly answered my questions. I’d love to hear about After Effects performance as well.
I just installed 10.3 (7B53) on my G4 533 1GB RAM last night, and for some reason the CPU stays pegged at 100%. Seeing quite a bit of latency, and QuickTime stutters. Not sure what is doing it as it is in the Kernel. Anyone have thoughts?
I can’t say I’m entirely happy with the changes Apple has made to the design either. Aqua is getting more like metal, but that doesn’t really matter as everything in the Dock by default is Metal, except Mail and System Preferences.
Micmoo, the Photoshop plugin for the G5 gives Photoshop 64 bit addressing. This would mean a HUGE speed improvement. Currently the dual 2 GHz G5’s in the test were running 32 bit code, not 64.
Nnooiissee, sounds like an Energy Saver problem. It’s probably not actually doing anything at all, empty clock cycles or something. I’m sure it will be fixed in the final release.
As for the appearance, you can always use themes. Max Rudberg has some good ones for 10.2, and as I remeber when 10.2 came out, he was very quick to port everything from 10.1.
Thanks for the info. I just tweaked with Energy Saver to no effect, but that doesn’t mean much. I can still test the functionality of everything, but I can’t test speed and that is frustrating.
I’m going to repost my comment to the newly posted panther story.
I wasn’t talking about the Energy Saver prefpane speccifically, but about what it does. I can control the CPU usage. If you have a laptop, you can set energy saver to cut the clock speed down to improve battery time. I just thought that since Panther was keeping to the CPU usage up, it may have had something to do with whatever it was Energy Saver used to control things like that.
The G5 isn’t going to see any real performance benefit from 64-bit addressing along unless the test actually uses > 32-bit address space and the machine has > 2GB memory. However, the plug-in may also optimize the generated code for the G5, so a healthy performance benefit is definitely possible (though 100% seems optimistic).
Also, the ONLY thing keeping the Opteron from being a personal computer chip is pricing. Sure, the 64-bit version of Windows isn’t out for it yet, but it runs the 32-bit version quite well. The Athlon64 is the same core with single channel memory (though the AthlonFX, also a PC chip will have dual-channel).
BTW, does anyone know how much cache the Athlon64 and AthlonFX will have?
Nothing against the Opteron, I’m a big fan of AMD. But for a personal computer, the Opteron is a waste of money until Windows can really use it’s power. I didn’t mean to say it could not in any way work in a personal computer, just that it’s not considered a personal computer chip yet.
Yeah, the Opteron is a waste of money right now (though I’m drooling over two Opteron 246s on a Tyan K8W). I’ll be curious to see how much only having single channel DDR hurts the Athlon64. I hope they make a non-Opteron (read non-ECC memory) dual-processor version.
I don’t think a 64-bit OS is absolutely necessary for the Opteron though. In the PC world (not workstations) hardly anyone has >2GB of memory (though this is probably going to change in another year or two). Sure, the ISA enhancements will be nice, but we are only talking about a 5-10% performance improvement from those.
It won’t have to be a full 64 bit OS, there is usually nothing in the OS that requires that anyway, but just 64 bit addressing for processor intensive apps. I doubt OS X will really be full 64 bit for years to come, and thats saying a lot since a new version is out once a year.
Most apps don’t even use what they have to work with now, but processor intensice things like video, CG, games, that stuff is going to be brought up a huge notch. It won’t be a big deal for most PC users, seeing as about 60% of them are still on Win98, but it does open a ton of doors for everyone else.
I think the biggest thing Apple should be touting about the advantages of the G5 over the G4 is the bus speed. It was something like 167 MHz and now its at 1 GHz. Thats big. Real big. They said a 30% speed increase with Panther alone. Plus some 64 bit apps, not just plugins, is going to change the whole game around. The new Photoshop, Illustrator, and InDesign, coming out this year, are all going to be full 64 bit on the Mac versions. I’m sure Apple will do a big update for their own Pro apps (read – Final Cut Pro), as well as Logic.
The Apple G5s are advertised but not in stock.
Opteron 240 and 242 CPUs, Tyan (dual CPU) and Asus NForce3 (single CPU) ATX motherboards are *in stock* at my local computer shop in Brisbane Australia.
The Opteron already runs XP Pro (in 32 bit mode).
The Opteron is just as much a desktop as the G5. However the G5 isn’t a true workstation class machine – no SATA RAID, No ECC RAM support, no workstation class graphics cards supported.
Ummm, the G5s are in stock. People have them. They have been shipping for quite a while now. And we already said the Opteron can be used as a desktop, but using it in 32 bit mode is just a waste of money, as is any 64 bit chip running Windows, until Windows supports 64 bit addressing.
RE: Anonymous (IP: —.tpgi.com.au) – Posted on 2003-09-07 03:24:57
The Apple G5s are advertised but not in stock.
Opteron 240 and 242 CPUs, Tyan (dual CPU) and Asus NForce3 (single CPU) ATX motherboards are *in stock* at my local computer shop in Brisbane Australia.
If you had 10,000 friends, you could buy a MB and processor directly from IBM if you so wished. Instead of comparing parts, how about compare AVAILABLE PRE-BUILT MACHINES FROM BIG NAME VENDORS, IBM has announced support, however, AMD64 machines probably won’t appear until Windows XP AMD64 is made available, meaning, your argument is pointless.
The Opteron already runs XP Pro (in 32 bit mode).
Aren’t you the same Australian is who rabbidly anti-Apple and claimed that the G5 will suck because MacOS won’t be a 100% 64bit operating system? here is a hint sunshine, Windows XP AMD64 won’t be 100% 64bit, either now or anytime the future.
The Opteron is just as much a desktop as the G5. However the G5 isn’t a true workstation class machine – no SATA RAID, No ECC RAM support, no workstation class graphics cards supported
There is nothing stopping you from installing an workstation card later on, the only thing that may cause problems is the lack of drivers, but that is a hard vendors problem not Apple. As for ECC memory, explain to me why one would want it on a workstation? sure, on a server, I could see the value, however, on a workstation? please, I don’t think you need to have 5 9’s on a workstation.
Actually I got Opteron 246 a week before before G5 were available 240-244 are available since July.
PantherPPC, 64-bit adressing under winxp is done through PAE.
CooCooCaChoo, Windows xp for AMD64 is fully 64-bit (currently beta). No idea from where you got the info that it is not.
“how about compare AVAILABLE PRE-BUILT MACHINES FROM BIG NAME VENDORS”
I would say that IBM is a little bigger than Apple. Take look at latest IBM eServer 325. Here you have some tests: http://www.tpc.org/tpch/results/tpch_perf_results.asp
e series with opteron 246 is currently the fastest cluster.
IBM already got an contract for thir most powerfull cluster in the world built on eserver/opteron.
Actually I got Opteron 246 a week before before G5 were available 240-244 are available since July.
They are parts, not a complete system.
PantherPPC, 64-bit adressing under winxp is done through PAE.
PAE gives 36bit addressing, IIRC, Opteron allows 43bit addressing.
CooCooCaChoo, Windows xp for AMD64 is fully 64-bit (currently beta). No idea from where you got the info that it is not.
“how about compare AVAILABLE PRE-BUILT MACHINES FROM BIG NAME VENDORS”
I would say that IBM is a little bigger than Apple. Take look at latest IBM eServer 325. Here you have some tests: http://www.tpc.org/tpch/results/tpch_perf_results.asp
e series with opteron 246 is currently the fastest cluster.
IBM already got an contract for thir most powerfull cluster in the world built on eserver/opteron.
You are talking about a server not a workstation. The original debate was about the fact that there are no large computer vendors offering opteron or AMD64 equiped machines, and no, don’t point to boxxtech.com, because I or any other schmuck are not going to try and order a system from overseas, get hit with a $500 shipment fee THEN have to pay GST no only on the system but the shipping as well. That DOESN”T include the fact that if something goes wrong, the system will have to be shipped BACK to the US at a cost of $500.
I have NOT said that Opteron is a good or bad product, infact, if they were around and priced reasonably at the time I bought this (eMac), I would have gone with it, however, since it wasn’t, I went with a Mac. Sure, it doesn’t set the world on fire in terms of benchmark scores but it does get the work done with minimum fuss and bother.
CooCooCaChoo,
can you provide me with link explaining that opteron is 43-bit not 64-bit processor?
I am sorry tnat you dont have any vendors selling opteron workstations but it does not mean that they dont exist. In Bangladesh they dont have G5 and for long time probably they will not have latest apples because this is very poor country and they have bigger problems. So following your thinking G5 does not exist as a desktop/workstation.
CooCooCaChoo,
can you provide me with link explaining that opteron is 43-bit not 64-bit processor?
I am sorry tnat you dont have any vendors selling opteron workstations but it does not mean that they dont exist. In Bangladesh they dont have G5 and for long time probably they will not have latest apples because this is very poor country and they have bigger problems. So following your thinking G5 does not exist as a desktop/workstation.
Here is the difference:
“AMD 64-bit technology builds upon the x86 instruction set, one of the industry’s most proven and widely supported technologies. The AMD Opteron processor seamlessly integrates into current 32-bit computing environments, enabling enterprises to deploy high performance 64-bit capable systems that build upon the billions of dollars already invested in 32-bit software. “
“Any application should run faster on the AMD Opteron processor due to its doubled data paths. Current 32-bit processors have a 4 GB memory addressing cap. The AMD Opteron processor’s 40-bit physical and 48-bit virtual addressing remove that limitation, permitting up to 1 Terabyte of physical memory space and 256 Terabytes of virtual memory addressing space.
Regarding 43bit, I was wrong with that, I was getting confused with the Itanium which has a 53bit virtual addressing IIRC.
64bit refers to the instruction set. The virtual addressing, however, can be something completely different. This is where people get confused about 64bit processors.
CooCooCaChoo,
I found it, 64/42 G5 and 48/40 opteron. If that would be real advantage eseries would built on ppc not opteron
“can you provide me with link explaining that opteron is 43-bit not 64-bit processor? ”
64 bit or not, only Linux can address it all. PAE allows Windows to address more than 4 gigs of ram, but not 64 bit addressing.
http://www.microsoft.com/whdc/hwdev/platform/server/pae/default.msp…
That said, Windows can’t take advantage of the Opteron. It can run on it, but only in a sort of crippled state.
BTW, an eServer and a personal computer are very different things. Telling us that IBM has a contrat to build a supercluster doesn’t pertain to this disscussion at all. No one is argueing that the Opterons aren’t being used, just that they aren’t personal computer chips yet.
“64-bit capable systems”
Thus requiring a 64 bit capable OS, and that whole post sounded more like a high end server thing than a personal computer thing.
Where os X is 64-bit?
opteron runs 32-bit as well as 64-bit OS/apps
I dont know how it sounds to you.
Now Windows XP 64-bit Edition for AMD64 can utilize up to 16 terabytes of physical memory. And again this is fully 64-bit OS in contrast to OS X that can only address more than 4GB barrier but it is 32-bit OS
The bottom line is that it may be some time before OS X become goes truly 64-bit.
It seems that you are lost in the disscussion about 64-bit cpu and OS (not benchmarks):
“CooCooCaChoo
Aren’t you the same Australian is who rabbidly anti-Apple and claimed that the G5 will suck because MacOS won’t be a 100% 64bit operating system? here is a hint sunshine, Windows XP AMD64 won’t be 100% 64bit, either now or anytime the future.”
My response was that Windows XP 64-bit Edition for AMD64 can utilize up to 16 terabytes of physical memory. And again this is fully 64-bit OS in contrast to OS X that can only address more than 4GB barrier. The bottom line is that it may be some time before OS X become goes truly 64-bit
Next he complained about CPU 64-bitness and that was clarified.
Also opteron workstations were available before G5, that also was clarified.
Finally I was wrong about PAE and CooCooCaChoo corrected that.
AMD is advertising opteron as workstation/server class cpu because they want to sell Athlon64 and AthlonFX. Otherwise who would bother with those?
If you want to you can buy opteron worksations.
CooCooCaChoo,
by the way BOXX workstations are available in Australia
H-Digital Pty Ltd. in Sydney, Melbourne, whole Europe and several countries in Asia
OS X doesn’t need to be a full 64 bit OS. That would be almost useless. 64 bit addressing does more than access large amounts of RAM, which is all WIndows XP can do on a 64 bit chip. It won’t run 64 bit apps. OS X will. How’s that for a bottom line. I think you are the one lost in discussion here.
I will be getting a dual G5 on 9/26. It will be a long wait.
To the person who says it does not have Serial ATA drives; well you are wrong. I’ll have a 160gig SATA –just look at the store front, it says right on it; “SATA Drive”. Not that that is such a jaw dropper. It improves 133 by going to 150 and tweaks a few things to become a little more reliable — progress at least.
Looking at the performance tests at Bare Feats, I think that the G5 will get faster over the next few months just do to software updates. Cinebench on the Mac holding its own against Athlon 64 and Xenon?!? I’m hereing Panther is 30%-40% faster than Jaguar. And the 970’s architecture appears to be very affected by code organization and optimization (like Altivec which can be up to 800% faster with optimized code). Granted the Athlon 64 may have some of the same hurdles, but is not such a new design as the Power 4 architecture.
So overall, I will be running a sweet box.
And for what its worth, 64 will sometimes be slower than 32 bits for the same task. But if we are talking about largeInt or doubleprecision floats — there can be a huge gain. The point is, we don’t need Word or PowerPoint faster — the 64 bit machine will help where it counts, in highend database, video, 3d and computer games. And 3D is so dependent on software optimization it is crazy to fight over this now — I’ve seen renderers improve 400% from one version to the next just based on better code. What matters is how easy it is for developers to achieve good code. And rumours from Maxon are incouraging–a score of 500! Um, whatever 500 stands for 😉
To MP who says;
Now Windows XP 64-bit Edition for AMD64 can utilize up to 16 terabytes of physical memory. And again this is fully 64-bit OS in contrast to OS X that can only address more than 4GB barrier but it is 32-bit OS..
One; OS X 10.2.7 is 32 bit OS with a 64 bit bridge. That means the speed and support of 32 bit apps with whatever you want at 64 bits. From here on out, there won’t be much improvement by going fully 64 bits for most of the apps we currently use. By the time the big apps arrive, we will be ready.
Now 64 bit XP –wow 16 terrabytes. That just means it is 64 bit –what do you think 64 bits is for? Now, try putting that much ram on your 3 slotted board –oops. Oh, my 32 bit apps and games won’t run–oops. You are talking about a good server/workstation architecture with like 10 applications in beta. And why can’t I have more than 4 gigs on my mac box? I only have 4 slots and the ram comes as 1 gig. My bank account couldn’t afford that much. But it would be nice to have 8 slots. The 64-bit aware apps can use more. 32 bit apps cannot. So the 16 terrabytes will help you with nothing at this point in time. Especially since none of this is even available.
What exactly is your point? I will be producing movies at HD before you run Office in that WinXP64-AMD.
And I am a real fan of AMD– I only use their processors when I build a Win box. But if these guys don’t make money soon, you may never see processors from them in 2 years. It will just be IBM and Intel. And IBM has a better roadmap to improve processor performance (at least until Intel can invent a cooling laser or something).
And when has Microsoft NOT announced something cool when somebody else does? They won’t be pushing out this baby until there is demand. And probably only to high-end just to compete with Linux. I don’t see demand at the low end.