[clarkgoble]

Speed definitely is the issue and no the G4 doesn't have it. The top end Pentiums right now are scoring benchmarks as fast if not faster than what the 970 is predicted to do. While the bandwidth problems due to the bus design is an other big problem, the G4 can't be neglected. There is no way a 1.5 GHz G4 is going to compete, speedwise with even an Athalon 2400.

We just got an Athalon 2200 dual processor system. It will smoke anything Apple has. And it cost about $1000.00.

Right around the time we hope the 970 systems will be coming out AMD will have their new 64 bit systems out as well. And they are faster than the 970 by far. (About 20% faster from what I saw)

The real issue, as I see it, will be how Apple prices their machines and whether they come out with dual processor 970 systems.

Don't get me wrong, I think OSX is the best OS by far. But the hardware is very overpriced and underpowered.

[Simon]

OMG, this last post is going nowhere. Jee wiz.

Originally posted by clarkgoble:
Speed definitely is the issue and no the G4 doesn't have it.

This is baloney. The G4 has incredible amounts of speed. If you wouldn't be blinded by this "P4 at 3GHz does 1 kFrame/s in Q3" masturbation you would see what some people still use nowadays at home, in the office and in scientific or technical environments. They get work done buddy. Word processing, presentations, spreadsheets, e-mail, web, etc. all gets done perfectly by a chip like a G4 or even a PIII. These people don't need the I'm-such-a-geek-wannabee-wanking. They get work done. I can't hear this unqualified crap about the G4 not having speed anymore. How many times have you trolled around like this before? How many times have people tried to explain it to you? Huh? My god, you sure love to hear yourself reiterate the same stuff no matter how unqualified it is.

The top end Pentiums right now are scoring benchmarks as fast if not faster than what the 970 is predicted to do.

Absolutely no serious and sane human being is going to compare a 970 which has not yet been placed on an Apple board and tested in public against a chip from Intel that is out on the market today. Even your always quoted SPEC marks are so close togther when you look at these chips that you should admit we should wait till the chip is on an Apple board till we judge it by comparison with what Intel has to offer at the time. And do we have a bus yet? Has Apple shown to you what their architecture for the 970 is going to look like? Have they told you what it's going to cost? No? Ah, right, because they haven't even said that they are going to use the damn chip!

Right around the time we hope the 970 systems will be coming out AMD will have their new 64 bit systems out as well. And they are faster than the 970 by far. (About 20% faster from what I saw)

In real life? On an existing bus we've all seen? The 970 we have all seen benchmarked on OS X against the 64bit AMD chip we've all seen on XP in public? What the hell are you talking about? You're talking about stuff that hasn't even been shown to us running our OS. Actually, we don't even know for sure if it will ever run our OS.

You have no serious technical interest in this topic at all. If you would, you'd know that
a) the 970 hasn't ever been tested publicly running a real-world OS or any pplications people use.
b) if the 970 comes to the Mac we have no idea what bus it will run on - Apple hasn't EVER showed or said anything about the architecture.
c) we have no idea what memory sub-system will come with the bus architecture.
d) we have never seen one of these intel or amd chips you blabber about run our OS in public benchmarks. we have no idea how they would compete.

So all in all you aren't interested in the future of our plattform, you aren't interested in technical discussions or details. You're just spreading FUD, making a fuss and getting the newcomers nervous. Such posts are a disgrace for a technical forum.

The real issue, as I see it, will be how Apple prices their machines and whether they come out with dual processor 970 systems.

Exactly. If Apple is working with tomorrow's 970 that has a 5% lower SPEC than today's Intel P4, Apple should start selling PowerMacs for 50 bucks a piece and give away OS X for x86 for free. Yada yada. I've heard stuff like this before. The last time it brought Apple right onto the verge of chapter 11 and take-over. I sure hope Steve puts suggestions like these where they belong - into the trash.

Don't get me wrong,

Nobody's getting you wrong. You're just trying to machinate and put fuel into the fire. You may be unhappy because your Mac doesn't carry the numbers your PC friend's machines have and I pity you because it will likely stay that way, but live with it or leave it. There's no reason to start posting senseless comparisons to try to make people believe Apple's engineers are all idiots because they (maybe!) are experimenting with a chip you already know is inferior even though nobody on this forum has ever seen it yet. That is just plain clownish.

I think OSX is the best OS by far. But the hardware is very overpriced and underpowered.

Would it comfort you if they sold the PowerMac for 400$ and just added a 1600$ to the price of the OS? You'd have to say "well the hardware's slow but the price is fair, but damn, why is the OS so expensive?". Would that make you feel any better about it?

I'll give you just one piece of advice: Wait till we actually see what Apple will use as a successor for the G4. Wait till you see how they do it, what bus they use and what memory. Wait until you see some real-life OS and app benchmarks. Wait until they tell you what it will cost. And then go and compare it to XP running on the newest Intel/AMD offering. That's when you can come back and start b!tching about it. OK?

Holy cow.

[galarneau]

Hot Damn Simon!!!!

Way to tell it like it is!!!!!

[eddiecatflap]

..does anyone else find that smiley mildly disturbing ?

[Spliffdaddy]

"I'll give you just one piece of advice: Wait till we actually see what Apple will use as a successor for the G4. Wait till you see how they do it, what bus they use and what memory. Wait until you see some real-life OS and app benchmarks. Wait until they tell you what it will cost. And then go and compare it to XP running on the newest Intel/AMD offering. That's when you can come back and start b!tching about it. OK?"

Speculation and rumor are allowed in this thread.

I'm of the opinion that even when Apple releases a machine with a new processor - it still won't be competitive with AMD and Intel's current offerings.

[Simon]

Originally posted by Spliffdaddy:
Speculation and rumor are allowed in this thread.

Absolutely. This is what all these PPC970 threads are about.

I'm of the opinion that even when Apple releases a machine with a new processor - it still won't be competitive with AMD and Intel's current offerings.

But you yourself deliver the exact reason for my criticism of Clarkgoble's post. We can guess and believe all we want - like you do. But it's technically futile to compare benchmarks of an existing chip with existing motherboards to a chip we only have seen some prelimenary SPECint/fp numbers of. It's even worse if you compare the existing chip running on an existing OS to a developmental chip that has never been shown on the OS you're comparing to.

You just can't try to argue with numbers and pretend to be scientific when your methodology is not scientific at all. To me that sounds more like an attempt to start a flamewar.

[Metzen]

Originally posted by CubeBoy:
Cool, .09 micron PPC970s will probably be running upwards 2.5 ghz. Of course the upcoming Prescott Pentium 4s, which is scheduled to come out Q2 2003 will be the first chip actually to use .09 micron manufacturing process.

I guess we'll see about that... I wouldn't doubt Intel is just going to sit on their asses and reap the mark-up of the current P4's before AMD forces them to release something better.

Originally posted by CubeBoy:
Anyways, I agree, IBM is going to be a lot more commited to Apple than Motorola ever was and they will deliver. Of course, I'd still prefer a Opteron like cpu in my Mac to a PPC970.

You maybe waiting till 2005 for that Operton the way it's been pushed back, delayed, pushed back, delayed, etc.

[freakboy2]

Hi, I like to post off topic and make people upset. My posts will be edited until I can grow up.

[DrBoar]

The main problem with the G4 was not that it started topping out at 500 MHz but the fact that it during the more than 3 years since then scaled to poorly. Had it followed the general CPU trends and doubled its speed every 18 months or so we would be at 2-2.5 GHz now and probably on par with AMD and Intel even using single CPUs.

Sure I want the 970 to come out topping out at 3 GHz so I can get a low end dual 2.2 GHz for way less than 2000 dollar. But the improtant thing is that it will scale well in the future. Jan 2000 the 500 MHz G4 was at half the clock speed of the 1 GHz Athlon by AMD, now it has "progressed" to run only 230 MHz above being at one third of the clcok speed of the P4, very close to be 2 GHz slower

[Eug]

Am I the only one that thinks the new chips will be 7457s for the time being? ie. 1.3 GHz and up. Being a non-engineer I may be out to lunch, but it seems that the current designs are perfect for it. ie. the current PowerMac designs were made in anticipation of the 7457 or 7457-RM, but those CPUs didn't appear quickly enough. Thus, they just overclocked some 7455 chips until they could get their hands on the new chips. And when the 7457-RM appears, they could just change a few things, drop in a CPU, and away you go with a real 167 MHz (or higher) DDR bus.

The PPC970 won't be until later, and it will first be an Xserve.

[Simon]

Originally posted by Eug:
Am I the only one that thinks the new chips will be 7457s for the time being? ie. 1.3 GHz and up.

No, you're not the only one. I think so too. That already makes two of us.

I agree with you on the current PowerMacs. I think the DDR design was basically a cut-down version of what Apple had in mind for the 7457 which would support true DDR. The 7455's MaxBus is all they had since Moto delayed the 7457 and that is what they worked with. Once they get the 7457 they could use "true" DDR running on faster clocked boards.

I don't see the 970 running on Mac boards till at least January (even though I'm dreaming of a September release). And then I suppose it will debut on an Xserve. That gives PowerMacs another year to run on a 7457. Of course that's just my guess, but I really don't see the point in getting your hopes up for a July release of dual 970 PowerMacs.

[Ken_F2]

Make that three...

IBM has said publicly that mass production of the 970 will commence in 2H 2003 (they have said nothing about shipment). From the time that production begins, it takes 10-12 weeks to completely fabricate the processor. And then you have to account for the time for IBM to actually ship the processor to Apple, and for Apple to produce some quantity of PowerMac systems for shipment; this will probably take anywhere from two to four weeks. So if mass production on the IBM 970 begins July 1 (the first day of 2H 2003), we won't see IBM 970 systems from Apple before mid-October. If mass production begins circa October 1 (more likely), we won't see systems until mid-January.

Thus, mid-October is the earliest possible date we can possibly see IBM 970 systems. Imo, January 2004 is much more likely. In the interim, I expect we'll have 1.3GHz to 1.5GHz 7457 (G4) systems.

[Souljah]

See that perspective "Zoom out" button?
Press it.
http://slashdot.org/article.pl?sid=0...thread&tid=142

[CubeBoy]

Originally posted by Metzen:
I guess we'll see about that... I wouldn't doubt Intel is just going to sit on their asses and reap the mark-up of the current P4's before AMD forces them to release something better. You maybe waiting till 2005 for that Operton the way it's been pushed back, delayed, pushed back, delayed, etc.

They've already demoed Opterons and Clawhammers running at up to 2 ghz. At least one of their two factories already have the capacity to manufacture them. I doubt they will be delayed any longer than they already are. If AMD releases it's Hammer and Opteron. Intel will have to release it's Prescott Pentium 4s and maybe even Itanium 3 (which is scheduled to come out rougly the same time) to stay in dominance.

[Ken_F2]

The Inquirer reports today that the Athlon 3000+ (not based on Hammer) will debut on February 7.

Opteron systems have already been previously announced for February 14(?) of next month. However, these first systems (dual and quad) will debut at the relatively low speed of 1.4GHz. At 1.4GHz, the Opteron with its 1Mb L2 cache and integrated dual channel memory controller should offer comparable performance to a 2.6GHz P4. Dual and quad systems will be available, and they will be pricy, as you would expect for a much-anticipated processor's early availability.

The desktop Hammers with top-end performance (2-2.2GHz with 3400+ performance) aren't expected become widely available until July. It will apparently take AMD several months to ramp Hammer up to speed (MHz) for the desktop. Supposedly, a Hammer with 4000+ performance will debut late in the year.

[Un-Inferior]

Originally posted by Simon:
OMG, this last post is going nowhere. Jee wiz.... blah blahh...

Holy cow.

simon you're blind. you've just used the stock 'good enough' excuse, but somehow it took up my whole screen. perhaps you're just pissed that macs really are slower. when you wrote your masturbation argument about gaming, you've completely IGNORED the graphic design market. without visual designers there would hardly be a market for powermacs. many of my peers switched to pc's just because they render faster and are a lot cheaper. rendering time literally is money. there's not a computer fast enough for me when i'm rendering 3d or dealing with live previews. it's safe to say if my computer can wank q3 at blazing fast speeds, it can probably run my 3d apps quickly. hell even 2d apps like photoshop will run FASTER on say a 2 or 3ghz mac. anyway, there are a lot of mac users who would like to run 3d games at their native desktop resolution (ie 1280x1024) without stutter.

you're a prime example of a complacent mac user. good for you.

[UnixMac]

Wintel Guys...

The thing we are all missing here is the move to 64bit OS...That is where this is all going in two or three years (the shelf life of my current hardware)

IBM has a lot of experience with this on the Power4, and they run it with a Unix that is a lot closer to OS X's Mach 3 than Windows 2000 or XP Pro.

So with that said, Intel and AMD who have NO experience in the 64 bit world will have to team up with MS, who only makes 32 bit OS's and make a platform that will beat OS X on a PowerPC970??? Not!

[cowerd]

The thing we are all missing here is the move to 64bit OS...So with that said, Intel and AMD who have NO experience in the 64 bit world will have to team up with MS, who only makes 32 bit OS's and make a platform that will beat OS X on a PowerPC970??? Not!

There are already 64-bit versions of Windows Server and various Linux distros (Redhat, Covalent). The Windows 64-bit runs on the Itanium, which is surprisingly enough, a 64-bit processor. MS has also committed to delivering a 64-bit version to AMD's x86 64-bit based hardware. This OS is already running on AMD test mules. In other words its already product.

IBM has a lot of experience with this on the Power4, and they run it with a Unix that is a lot closer to OS X's Mach 3 than Windows 2000 or XP Pro.

I can't even to begin to follow the logic of this statement.

[UnixMac]

Originally posted by cowerd:
There are already 64-bit versions of Windows Server and various Linux distros (Redhat, Covalent). The Windows 64-bit runs on the Itanium, which is surprisingly enough, a 64-bit processor. MS has also committed to delivering a 64-bit version to AMD's x86 64-bit based hardware. This OS is already running on AMD test mules. In other words its already product.

I can't even to begin to follow the logic of this statement.

Let me help you follow...

The OS that you refer to is a hardly a baby compared to the Unix IBM has been using for the Power4 based servers...Can we say windows 3.1, windows 95, windows 98 etc... how long did it take MS to get it close to right?? Can we say Optimization? Even in HP's own literature it talkes about this. It also says that you can run windows for your "little-endian" apps and HP-UX for your "Big-endian" apps.... i.e. Windows is for the not as powerful as Unix, especially when networking machines.



IBM Power4 based systems

Itanium not up to par with Power4

I'll quote from the above link:

"Internally, even though the instruction set structure may not be as elegant as Alpha's, POWER4 is close enough - in fact, its design philosophy bears the most resemblance to the dying performance leader. As it sheds the remaining baggage that blocks fast speed ramps and moves towards the mainstream (Apple?), this architecture could be well positioned for the final 64-bit showdown."



As for the Itanium, for one its not up to the Power4's performance and secondly most of them are running Linux, and HP-UX... Windows 64 is a baby and not ready for prime time. And Windows 64 for AMD is still an Alpha.

OS X has (a lot) more in common with HP-UX than Windows 2000/XP/64... does that make sense now?

[CubeBoy]

Originally posted by UnixMac:
Let me help you follow...

The OS that you refer to is a hardly a baby compared to the Unix IBM has been using for the Power4 based servers...Can we say windows 3.1, windows 95, windows 98 etc... how long did it take MS to get it close to right?? Can we say Optimization? Even in HP's own literature it talkes about this. It also says that you can run windows for your "little-endian" apps and HP-UX for your "Big-endian" apps.... i.e. Windows is for the not as powerful as Unix, especially when networking machines.

I'll quote from the above link:

"Internally, even though the instruction set structure may not be as elegant as Alpha's, POWER4 is close enough - in fact, its design philosophy bears the most resemblance to the dying performance leader. As it sheds the remaining baggage that blocks fast speed ramps and moves towards the mainstream (Apple?), this architecture could be well positioned for the final 64-bit showdown."

As for the Itanium, for one its not up to the Power4's performance and secondly most of them are running Linux, and HP-UX... Windows 64 is a baby and not ready for prime time. And Windows 64 for AMD is still an Alpha.

OS X has (a lot) more in common with HP-UX than Windows 2000/XP/64... does that make sense now?

Hate to break it to you pal but Alpha's dead, it's performance crown has been taken a while ago, Power4 is definitely better than Itanium but Itanium 2 right now is the performance leader among servers and workstations. Perhaps you should do a little more research before you post next time.

Power 4 1450 mhz:
SPECfp:1296 (peak)

Itanium 2 1000 mhz:
SPECfp:1422 (base and peak)

Itanium 800 mhz
SPECfp:645

[UnixMac]

Originally posted by CubeBoy:
Hate to break it to you but Alpha's dead, it's performance crown has been taken a whileago, Power4 is definitely better than Itanium but Itanium 2 right now is the performance leader among servers and workstations.

Power 4 1450 mhz:
SPECfp:1296 (peak)

Itanium 2 1000 mhz:
SPECfp:1422 (base abd peak)

Alpha may be dead, but the technology isn't, and a lot of it is found in the Power4 (as well as the Itanium) as for the performance numbers you quote, they are not real world, you don't run a CPU in a vacuum, and with the upwards of 512MB of L3 avaiable with a Power4, and the scalability of the system, its well ahead of the Itanium, again read the article that I reference, it's only a couple of months old... not much has changed. Otherwise, if the Itanium is superior, maybe you should write your congressman and ask him to yank that multi-million dollar contract for IBM to build a 100 Teraflop supercomputer for the DOD based on the Power4.... oh wait, never mind, the Itanium can't scale that far..

I will agree however, that at this very moment, the Itanium 2 is the fastest in a stand alone system.. but that will continue to change back and forth as Ghz increase.... the important thing is will EPIC replace RISK or will Intel listen to the guys who designed the Alpha (a winner in my book) and keep the RISC based instruction set.

[CubeBoy]

Originally posted by Simon:
OMG, this last post is going nowhere. Jee wiz.

Absolutely no serious and sane human being is going to compare a 970 which has not yet been placed on an Apple board and tested in public against a chip from Intel that is out on the market today. Even your always quoted SPEC marks are so close togther when you look at these chips that you should admit we should wait till the chip is on an Apple board till we judge it by comparison with what Intel has to offer at the time. And do we have a bus yet? Has Apple shown to you what their architecture for the 970 is going to look like? Have they told you what it's going to cost? No? Ah, right, because they haven't even said that they are going to use the damn chip!

Exactly. If Apple is working with tomorrow's 970 that has a 5% lower SPEC than today's Intel P4, Apple should start selling PowerMacs for 50 bucks a piece and give away OS X for x86 for free. Yada yada. I've heard stuff like this before. The last time it brought Apple right onto the verge of chapter 11 and take-over. I sure hope Steve puts suggestions like these where they belong - into the trash.

Would it comfort you if they sold the PowerMac for 400$ and just added a 1600$ to the price of the OS? You'd have to say "well the hardware's slow but the price is fair, but damn, why is the OS so expensive?". Would that make you feel any better about it?
Holy cow.

Simon, PPC970s "preliminary" SPEC benchmarks were done by IBM itself, likely with it's most ideal conditions and best hardware. SPECint is a set of benchmarks that benches CPUs on 12 aspects of UNIX performance, the results are weighted to establish a score that accurately represents CPU performance. The tests are NOT synthetic, they are actual UNIX operations. Having a Apple motherboard or OS will NOT change the score, it will probably lower it. Obviously your speaking out of complete ignorance as to what "preliminary SPEC scores" are. Now does that mean that it's not fast, absoloutely not, my G4 cube which is ancient by today's standards is fast enough for me. Well we mac users be the "envy of all pc users" as Krove suggested, unfortunately no.

[CubeBoy]

Originally posted by UnixMac:
Alpha may be dead, but the technology isn't, and a lot of it is found in the Power4 (as well as the Itanium) as for the performance numbers you quote, they are not real world, you don't run a CPU in a vacuum, and with the upwards of 512MB of L3 avaiable with a Power4, and the scalability of the system, its well ahead of the Itanium, again read the article that I reference, it's only a couple of months old... not much has changed. Otherwise, if the Itanium is superior, maybe you should write your congressman and ask him to yank that multi-million dollar contract for IBM to build a 100 Teraflop supercomputer for the DOD based on the Power4.... oh wait, never mind, the Itanium can't scale that far..

I will agree however, that at this very moment, the Itanium 2 is the fastest in a stand alone system.. but that will continue to change back and forth as Ghz increase.... the important thing is will EPIC replace RISK or will Intel listen to the guys who designed the Alpha (a winner in my book) and keep the RISC based instruction set.

I can't say this enough times, SPEC marks are not synthetic, it runs real world applications. in fact their the most accurate, trusted, and reliable tests in a UNIX background, read my last post. The both Itanium 2 and Power4 systems were equipped with the best hardware and periphals. Also, their is no new Power4 machine over the horizon while Itanium 3 is expected to come out soon.

[UnixMac]

"The modern RISC processors generally have a clock frequency that is lower than that of the Intel Pentium 3/4 processors or the corresponding AMD Intel look-alikes. However, they have a number of facilities that put them ahead in the speed of floating-point oriented applications. Firstly, all RISC processors are able to deliver 2 or more 64-bit floating-point results in one clock cycle. Secondly, all of them feature out-of-order instruction execution, which enhances the number of instructions per cycle that can be processed (although the newer AMD processors also have 2-way floating-point instruction issuing and out-of-order execution, they are limited by their adherence to the Intel x86 instruction set). Thirdly, the bandwidth from the processor to the memory, in case of a cache miss, is larger than that of the Intel(-like) processors. Notwithstanding these commonalities between the various RISC processors, there are also differences in instruction latencies, number of instructions processed, etc., "

-The above quoted from a Computer Engineer who will remain nameless as I don't have permission to name him.



Will the Itanium 2 replace traditional RISC processors with hybrid EPIC, time will tell. My bet is not.

[Ken_F2]

As to 970 performance, I think the most important point to emphasize is that it will be a VAST IMPROVEMENT over what we have now. Today, many Mac users tend to overstate/exaggerate G4 performance, while the IBM 970 will actually deliver on real, competitive performance. IBM's figures show that the 970 will deliver UNIX performance comparable to a P4 2.8GHz, and that is without Altivec (as far as we know). In Altivec-optimized applications, the 970 should deliver even greater performance. IBM's figures also show that the 970 will deliver anywhere from double to triple the performance of today's G4 processors under UNIX.

Moreover, the 970 has a much better implementation for SMP, so Apple will almost certainly continue its use of dual processors. Even if a single 970 only delivers 85% to 90% of the best P4 (3.4GHz?) performance at the time of release, remember at it also has Altivec (big % improvement in optimized apps), and Apple will likely be using dual processor systems. In contrast, consumer PCs are not generally made available with dual processors (P4 doesn't support multiprocessing, only P4 Xeon does).

Alpha may be dead, but the technology isn't, and a lot of it is found in the Power4 (as well as the Itanium)

Of course this is absurd. The only thing that the Alpha and POWER4 have in common is that they are RISC designs. Their implementation shares about as much in common as the Intel Pentium4 and AMD Athlon.

As for the ALPHA itself, the processor is dead. It is no longer being developed (as Intel acquired its engineers and assets). The ALPHA EV9 design that was completed jyst before Intel took over was easily the most advanced processor ever conceived. It would have laid waste to both Itanium2 and POWER4. SPECfp predictions for the EV9 were well in excess of 2000. But Intel will not be bringing the ALPHA EV9 processor to market, for obvious reasons, so it will never see the light of day.

But you are right, the technology itself isn't dead. When Intel bought out Compaq's ALPHA division two years ago, it acquired all but one person of the ALPHA design team from Compaq (the person that decided not to work at Intel was a pregnant women who was preparing to take a leave of absense anyway). It is already known publicly that the ALPHA engineers are working on a future generation Itanium design that will combine the best elements of the Alpha EV9 and Itanium processors; this processor is expected circa 2006. Many expect it to be a monster, performance-wise.

Otherwise, if the Itanium is superior, maybe you should write your congressman and ask him to yank that multi-million dollar contract for IBM to build a 100 Teraflop supercomputer for the DOD based on the Power4.... oh wait, never mind, the Itanium can't scale that far..

Each vendor routinely issues press releases about new, large-scale orders for their products. IBM's reputation, as well as the support they are able to provide, obviously played a role in that buy. Moreover, at the time the sale was made, Itanium2 was just recently announced, not widely available, and certainly not proven.

However, if you think Itanium2 can't scale, you are quite wrong. Indeed, SGI's Itanium2 boxes are currently the fastest for-sale systems on the face of the planet. Please see this January 7 news. Or for discussion of this news, see this page. Some quotes:

MOUNTAIN VIEW, Calif., Jan. 7 /PRNewswire-FirstCall/ -- SGI (NYSE: SGI - News) today announced that its new SGI� Altix(TM) 3000 family of servers and superclusters, the first high-performance Linux� environment capable of scaling to hundreds of processors with global shared memory, has swept the competition in a host of high-performance system benchmarks. Consistently delivering industry-leading performance at every level of scalability, the new SGI� systems significantly outpaced competing systems from IBM and HP.

World-Record Floating-Point Performance

In recent SPECfp�_rate_base2000 tests, a 1GHz SGI Altix 3000 system generated world-record floating-point performance for a 64-processor server with a score of 862. The closest 64-processor single system image competitor was the HP Superdome(TM) server (at 875 MHz) with a score of 267 -- less than a third the performance of the SGI system. Comparing 32-processor systems, SGI Altix 3000 performed 1.8 times faster than the IBM� eServer(TM) p690 system (at 1.3 GHz), and 3.5 times faster than the HP Superdome system (at 750 MHz).

The 32-processor SGI system returned a score of 443, compared to 251 for the IBM eServer p690 system and 128 for the HP Superdome system. The 32-processor SGI Altix 3000 results equate to a 300 percent boost in price/performance over IBM eServer p690.(1)

And these systems are running at only 1.0GHz. A number of prominent universities, including the University of Tokyo and University of Queensland in Australia, have purchased several of these 64-processor Itanium2 systems from SGI.

The next Itanium (Itanium3?), coming later this year, will run at least 30% to 50% faster than the current Itanium2, according to Intel. Whereas the current Itanium2 runs at 1.0GHz on the .18um process, the Itanium3 will be pin-compatible and run at 1.5 to 1.8GHz on the .13um process. Of course, for the forseeable future, Itanium and POWER4 will continue to be used in industrial "big iron" systems costing many thousands of dollars; thus, it is really rather pointless to compare them to 970 and P4 systems.

[UnixMac]

Well, I think its fair to say this subject is closed for me, I am wrong, the Itanium 2 is top dog... I was missinformed.

[Simon]

Originally posted by Un-Inferior:
simon you're blind.

Actually, I am color-blind.

you've just used the stock 'good enough' excuse

Ah com'on. It's not about excuses - since there's nothing to excuse for. Or do you have a poorly performing 970 at home?

perhaps you're just pissed that macs really are slower.

No, actually you are. That's why you're in a Mac forum wetting your pants because of the P4. Get a life.

you've completely IGNORED the graphic design market.
...
many of my peers switched to pc's just because they render faster and are a lot cheaper. rendering time literally is money.

First of all, I don't care about the graphic market. Nothing more to it - it's just not my market. And second, I'm glad for your friends. If their PCs are faster, cheaper and work better then go for it. I've never opposed to that. Check my post record.

you're a prime example of a complacent mac user. good for you.

No, you lack some serious education and manners my friend.

You're insulting and obviously not interested in what the point of the post was. I'll repeat it one last time: Don't try to decieve people by talking as if you were doing scientific comparisons when you are actually talking about something you haven't been able to test. That's it. And if your unpolite stubbornness doesn't allow you to accept that point, so be it.

[OreoCookie]

Originally posted by UnixMac:
Well, I think its fair to say this subject is closed for me, I am wrong, the Itanium 2 is top dog... I was missinformed.

No, it's not. Because it is availble in small quantities only. It doesn't have an impact yet.

On the other hand, the Power5 will be avaible next year (and yes, its little brother the PPC980).

[Simon]

Originally posted by CubeBoy:
Simon, PPC970s "preliminary" SPEC benchmarks were done by IBM itself, likely with it's most ideal conditions and best hardware. SPECint is a set of benchmarks that benches CPUs on 12 aspects of UNIX performance, the results are weighted to establish a score that accurately represents CPU performance. The tests are NOT synthetic, they are actual UNIX operations. Having a Apple motherboard or OS will NOT change the score, it will probably lower it. Obviously your speaking out of complete ignorance as to what "preliminary SPEC scores" are.

I'm sorry to tell you, but I know what SPEC marks are. And I know what preliminary work is sometimes worth - my business is full of that stuff. But, I also know that theoretically you're not wrong. I have just too often seen CPUs being touted as this or that and, upon being used in real-world environments, turn out to be something else.
I've seen processors being produced and advertised as low-cost/entry-level and finally turing out to be real killers. On the other hand I've seen processors being produced "for the next generation" and supposed to be "a revolution" - when they actually came out they couldn't compete with their predecessors in many tasks.

I'm not asking for much. It's only about fairness. Why are people here trying to deceive others? You can't test a chip in real-world environments when you haven't even seen it yet. That's all.

[UnixMac]

Originally posted by OreoCookie:
No, it's not. Because it is availble in small quantities only. It doesn't have an impact yet.

On the other hand, the Power5 will be avaible next year (and yes, its little brother the PPC980).

This is what I figured, I don't think IBM is gonna sit still while Intel one up's them in the high endn market, although with the entire Alpha team working for them (Intel), IBM is gonna have a tough time keeping up... Compaq/HP/Intel vs IBM....

[DrBoar]

It all boils down to the real and percived speed of the system not the CPU in a lab test. The 603 CPU was not as bad as the performas it was often used in. It got a reputation for being slow. Well the Performa 6200 was dog slow as a system. But with a slow and narrow system bus a horrible IDE implementation and slow graphics it was bound to be. The fact the the OS and applications was running mostly in 68k emulation did not help at all.

CPU X might be 20% faster than Y. But the percived and real speed of th 970 will depend on the rest of the hardware as well as the OS. Now that Apple use the same IDE disks and controllers as other do and the same AGP and AGP card and so on the OS will be an important factor.

What are the SPEC marks for the G4? If we are complaining that the 970 is not higher in SPEC marks than the current P4, how does the G4 fare?

[UnixMac]

Originally posted by DrBoar:
CPU X might be 20% faster than Y. But the percived and real speed of th 970 will depend on the rest of the hardware as well as the OS. Now that Apple use the same IDE disks and controllers as other do and the same AGP and AGP card and so on the OS will be an important factor.

Other than a faster CPU, Apple still need to get ATA133, and a faster FSB with DDR (which it will with the 970)... not to mention OS X is still a new OS, there is still a lot of optimization of the Aqua, and other layers that sit ontop of the Mach... In time this will all happen.....

I think that speed will not be a problem in systems to come.

[DrBoar]

The figures releases by Apple show that they have lost 50% of the tower sales during the last two years. They really need to get it right with the next generation of Towers. And they have done it twise before first wiht the PCI generation and a second time with the Blue & White.

The low end should have a ATI 9500 pro card, Hard disk with 8 MB cache and so on and the mid and high end should have even better stuff! Time for Apple to show that they are at the front line after a break since 1999 when the B&W G3/450 could take on any PC.

Simply adding 970 and its bus to the current towers would be stupid, with the arrival of the 970 it is time so show some radical stuff in a radical tower. Still a tower with the motherboard on a door and still Apple but more indistrial along the lines of the X servers and powerbooks. The curvy towers and the toilet seat iBooks were an era but it is high time to eliminate the curvy towers.

IBM and Apple the future can be really bright

[Ken_F2]

What are the SPEC marks for the G4? If we are complaining that the 970 is not higher in SPEC marks than the current P4, how does the G4 fare?

As far as I know, Motorola stopped releasing SPEC results for the G4 because they were so bad relative to modern processors from Intel, AMD, IBM, etc. However, a year ago, c't (a German magazine) did run the SPEC benchmarks on Apple's first G4 1.0GHz system. These tests were run under OSX 10.0x and used GCC 2.95.3.

The result for the G4 1.0GHz was a SPECint of 306, and a SPECfp of 187. For reference, the IBM 970 is quoted at a SPECint of 937 and SPECfp of 1051. Based on these results alone, the 970 @ 1.8GHz is more than three times as fast as the G4 1.0GHz in integer apps, and more than five times as fast for floating point apps.

Is the G4 1.0GHz really that slow in comparison to the 970 or the P4? Most certainly not. The G4 results were run under OSX 10.0x, and everyone remembers that version of OSX was a "dog" in terms of performance. And they used GCC 2.95.2 to compile the benchmark, which was the standard back at the time, but the newer GCC 3.1 compiler (included with OSX 10.2.x) produces faster PowerPC code. IBM uses its own highly-optimized Visual Age 6.0 compiler for its results. Moreover, in SPEC, a processor's performance result tends to be hurt if it has a slow bus; the G4 1.0GHz has/had a 133MHz processor bus, whereas the IBM 970 has a 900MHz bus (and P4 has a 533MHz bus).

We don't have results for the 970 and G4 under the same OS and using the same compiler, but when you account for all of the above, I expect the 970 1.8GHz will be between two and three times as fast as the G4 1.0GHz, depending on the application.

[Eug]

Originally posted by DrBoar:
The low end should have a ATI 9500 pro card, Hard disk with 8 MB cache and so on and the mid and high end should have even better stuff! Time for Apple to show that they are at the front line after a break since 1999 when the B&W G3/450 could take on any PC.

That's of course absurd. The 8 MB cache drives and ATI 9500 as the low end would be an incredible waste of money. The benefits for most people would be minor.

[UnixMac]

Originally posted by Eug:
That's of course absurd. The 8 MB cache drives and ATI 9500 as the low end would be an incredible waste of money. The benefits for most people would be minor.

I totally and respectfully disagree...

The iMac is a consumer Mac, no question about it... but the POWERMac is a Prosumer-Pro Mac... and the bottom end machine needs to be on par with the Wintel offerings, while the top end machine needs to have (dream setup):

-4 X PPC970 at 1.8Ghz (By then maybe even 2.2)
-900mhz bus
-1.0GB RAM standard, with many GB expansion, say 4 or more...
- ATA 133 or faster with RAID (two drive system, 2X120Gig) = 240GB
-2 X USB 2
-3 X Firewire 2
-fastest AGP bus available with Radeon 9700pro or fastest available with 128MB or more DDR
-and stuff I haven't even though off..

The point is, it needs to kick a$$ and cost should be what ever it takes. I predict it can be done for less than $4000.

So, you'ld have a $2000, $3000, & $4000 system

with the low end being a dual 1.8 970 and a few less bells and whistle.

[UnixMac]

BTW

If some one told me 20 years ago that IBM and Apple would be working to gether on a single Personal Computer, I would have laughed in your face.

The Days of Apple vs IBM are over.

[Eug]

All you need for a PowerMac is reasonable speed for the hard drive and a reasonable speed for the video card, for Quartz Extreme, over two monitors with spanning.

There is no big need for instance for a web designer to have a Ti 4600 or an 8 MB cache drive to run Dreamweaver and Photoshop.

I'd rather have a low end PowerMac be affordable to the short-on-cash customer who needs more than the limited iMac, than make all PowerMacs chichi machine for no real reason.

As for being on par with Wintel offerings... There is no law that states that work computers need ultra-high-end parts all around. They need to be functional for the task at hand. This is true in the Wintel world too. I'm no graphics designer, but when I was asked to set up the AMD system for work for 2D graphics, I got a SLOW dual monitor video card and a medium-speed hard drive. Yeah the card is slow in 3D, but it has ultra clear 2D support. It's not as if we're running Doom III and running HDTach all day on the machine. OTOH, I spent big bucks on other peripherals - money much better spent that way.

[cowerd]

-4 X PPC970 at 1.8Ghz (By then maybe even 2.2)
-900mhz bus
-1.0GB RAM standard, with many GB expansion, say 4 or more...
- ATA 133 or faster with RAID (two drive system, 2X120Gig) = 240GB
-2 X USB 2
-3 X Firewire 2
-fastest AGP bus available with Radeon 9700pro or fastest available with 128MB or more DDR
-and stuff I haven't even though off..

Gamer mentality. Sure people doing film and 3D might need that, but probably not. Someone doing PS will need RAID, but not a Radeon 9700 or 4x970. What Apple needs is more choices in the Store though.

[dfiler]

Yep, we just ordered $15k worth of server... it comes with a 4MB Rage (non-pro)

There are other criteria than just blistering 3d performance!