PDA

View Full Version : IBM's PPC 970 hits 2.5 GHz?!



Ed M.
02-27-2003, 01:14 PM
Looks like IBM is gearing up to stomp all over Intel.

The IMB STOMP-A-THON begins!

"Runs at frequences ranging from 1.8 GHz - 2.5 Ghz"

IBM CeBIT Press Room (http://www-5.ibm.com/de/pressroom/cebit2003/en/highlights/powerpcblade.html)

OR URL:

http://www-5.ibm.com/de/pressroom/cebit2003/en/highlights/powerpcblade.html

Translated from original (http://translate.google.com/translate?u=http%3A%2F%2Fwww.heise.de%2Fnewsticker %2Fdata%2Fciw-27.02.03-000%2F&langpair=de%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools)


OR URL:

http://translate.google.com/translate?u=http%3A%2F%2Fwww.heise.de%2Fnewsticker %2Fdata%2Fciw-27.02.03-000%2F&langpair=de%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

TheMacObserver coverage (http://www.macobserver.com/article/2003/02/27.11.shtml)

URL: http://www.macobserver.com/article/2003/02/27.11.shtml

--
Ed M.

mlinde
02-27-2003, 03:31 PM
I wouldn't call for a "stomp-a-thon" until you see a 64-bit OS running 64-bit applications on 64-bit processors that are significantly faster than comparable processors. 1.8-2.5 GHz 64 bit processors may be more than 2x faster than comparable 32 bit processors (I'm not a big fan of this kind of math), but that means the entire suite of applications/os/hardware has to come out NOW, not "in the future" to be as fast or faster than the current Pentium-class offerings. Megaherz myth or not, benchmarks in the real world show that current PPC offerings don't live up to the hype. I still sit next to my Quicksilver G4, but what I see in the 970 is a full suite of application and OS upgrades to join my hardware upgrade. If I have to upgrade everything (again) to be compatible, I'm tempted to look at the other side. Speed does matter in this business...

Ed M.
02-27-2003, 03:55 PM
You seem to have absolutely NO clue about the PowerPC ISA or what the 970 means in terms of the current software that's available on the Mac NOW.

Point of FACT #1:

The Moto PPCs out perform any Pentium or Athlon at EQUAL MHz. The 970 brings us *that* much closer in terms of 32-bit application performance.

Point of FACT #2:

The PPC 970 is TWICE as fast as the current G4 at the same clock speed. Also note the fact that the IBM chips will be 1.8 to 2.5 GHz. That's for a SINGLE PPC 970.

Now imagine what one of these things will do to a Pentium or Athlon. Then add another PPC 970 or *three* and you can see where Intel's offerings look absolutely pathetic.

Oh, and the PPC 970 has PLENTY of bandwidth. Were likely to see AltiVec performance that's off the scale, not to mention that second Floating Point Unit ;-)

So, in the end, this chip will run all your current 32-bit applications natively on this processor (IBM said so) and it will be just fine because that's exactly how the PPC architecture was designed from the start.

So please stop trolling.

--
Ed M.

Ed M.
02-27-2003, 04:09 PM
http://www.ambrosiasw.com/webboard/Forum64/HTML/001106.html

mlinde
02-27-2003, 04:19 PM
You apparently don't live in the world of real applications and utilization. I've heard the "stories" about performance, but the reality is that when you put a current pentium or athlon PC against a current Mac with the same set of tasks for 3D modeling and rendering, the PCs are FASTER than the Mac, sometimes up to 2x faster. Whether the raw processor benchmarks faster is irrelevant to me or my clients. What is relevant is how fast it works in a real computer.

I understand that the PPC is supposed to be faster, but in real-world applications it isn't what happens. You can pick and choose your favorite single task that the PPC will do faster (Steve Jobs does this often with his Photoshop games at MacWorld), but overall real world performance the PPC-based Macintosh does not hold its own against the Intel/AMD-based Windows CPU.

I also understand that on equal MHz footing, the PPC would (or should) outperform the Intel/AMD camp. In case you missed the last three years, they haven't been on equal footing in some time. The FASTEST PPC you can get is 1.42 GHz, thats almost the SLOWEST Intel/AMD machine you can buy today. Apparently there are P4 processors over 3GHz available today. That's TODAY. The 970 isn't even a shipping unit yet. I have faith in IBM's capability to ship their processor, but I also know that the competition isn't standing still (unless their name is Motorola).

I'm not "trolling" or looking for a fight. I'm looking at my 3k Mac that's just over a year old and doesn't clock at 1/2 the speed of a new 2k PC. I'm frustrated...

Beamtracer
02-28-2003, 01:50 AM
The news about the IBM 970 processor is very significant, and very exciting. There's no doubt that this will be a very very fast processor.

Currently Intel does have a temporary lead in megahertz. Ed is right when he says that the Motorola G4 is faster than Intel processors running at the same clockspeed. However Intel's clock speeds are faster. The difference is not as great as the clockspeed would indicate, though.

Also, don't take Motorola's current offerings as any indication on how the IBM 970 will perform. Because the 970 is a new architecture, you can't compare its clockspeed with anything from Intel, AMD or even Motorola. I'm still surprised how much emphasis people put on literal clockspeeds.

IBM will also be making their own branded boxes running the 970 processor with the Linux operating system. I'm not sure if applications running on Linux x86 will necessarily run on these IBM Linux machines, as the 970 may require programs to be written with a different byte order.

Anyone holding out to buy a Mac with a 970 processor should note that the machines will probably be very expensive when they debut, especially if they run as fast as predicted.

Newtek should continue to optimize Lightwave for Altivec, in the knowledge that this optimization will accelerate the new processors.

Julian Johnson
02-28-2003, 04:00 AM
Ed - mlinde a troll? Please, I don't know anyone who's championed and evangelised Mac Lightwave longer than Michael. I share his concerns.

I can see you and Beam are very excited about the new chip (I am, too) and I'm glad that Andrew Welch is also pretty pumped. The integer and FP performance looks stellar even without Altivec.

However, you have to have lived through the past four years where Apple's promises that the G4 would 'smoke' the Intel/Athlon duopoly look like some sick joke for Lightwave users. Now, before you embark on a perfectly reasonable tirade about how a) Newtek have failed to optimise Lightwave for the G4 and b) how G4s are far superior at similar clock speeds let me promise you that the arguments and sentiments are understood. Completely.

In practice, irrespective of megahertz, the current crop of Athlon/Xeon/Pentium 4 boxes has a hefty advantage in Lightwave over the G4s - both in terms of rendering speed and OGL performance. That situation doesn't look like changing in the next 6 - 12 months, other than the dark side's chips getting faster still.

What I need is some hard currency in terms of the new 970s. When will Apple actually confirm that it is going to use these chips in its machines such that I/we can make proper business decisions on the likelihood of their appearence? With the chips already being promoted by IBM what is Apple waiting for? If Apple does adopt the chip when will we see those machines? How much will they cost? Will they be in dual or single configurations? With no software rewritten, how fast will Lightwave be? How fast will the Intel/AMD chips be at that point and how much will they cost? What will be done to address the video/graphic/bus/video driver issues that seem to cripple OGL performance on the current Macs? Will there be a megahertz ceiling for two years as there was with Motorola and the G4 - how long will the 970 be stuck at 2.5? Will the roadmap give the 970s a processor advantage that's sustainable for longer than 6 months after launch or will the Intel/AMD chips ovecome that advantage within that period and restore their hegemony?

Idle theories, gossip, ivory tower benchmarks, putative speeds are all truly exciting but it's answers to the questions above we need and that Apple themselves should be evangelising right now if they have a genuine concern for their high end users (many of whom seem to be considering the other side). In theory, a dual 2.5gig PPC 970 machine that outperforms a dual 5Gig P4 looks great [even if you could cool such a beast ;-)]. By mid 2004? Who knows?

I know you cannot answer those questions with any certainty and probably no one will get any definitive numbers until the machines are released (when we find out all the real world compromises (the bus, the RAM, the video subsystems etc.) that have to be made to get the products shipped). So, for the moment it's a question of faith and trust in Apple - both those commodities are, I imagine, in fairly short supply with power users. The chips 'on paper' numbers speak for themselves.

My one abiding fear is that we all get sucked into a 12 month frenzy of anticipation (and stasis with our existing kit) only to be confronted with delays and sub-optimal real world performance - that by the time the machines appear the opposition has already eroded any theoretical advantage. Sound familiar? I'm hoping that isn't so...but I share Michael's frustration whilst my clients don't care.

This latest news from IBM is good. Let's hear something from Apple.

Julian

ingo
02-28-2003, 10:24 AM
>> In practice, irrespective of megahertz, the current crop of Athlon/Xeon/Pentium 4 boxes has a hefty advantage in Lightwave over the G4s...

But dont forget that they cost a lot more too. I'm looking for a fast render slave for LW but after i read the offers i got i see no reason why i should buy a Pentium (and dont forget that the Pentium is now at the end, there's nothing more like the normal 3 GHz or the overclocked 3,5 GHz). For the price of a dual 2,8 Xeon machine i can buy a dual 1,25 GHz Mac AND a single 1 GHz Mac.
A comparable PC to the dual 1,25 Mac is in case of renderspeed a dual 2 GHz P4, and if you look at Dell you will see that they cost exactly the same (and i bet Steve looked at Dell before he moved the prices down).
And in case of the 970 as a maybe Apple alternative we shouldn't forget that the PowerPc alliance doesn't exist anymore like a few years before. So if you want to have a 970 you have to buy a Linux box from IBM, say goodbye to Lightwave and you have a new 3D program on a new speedy machine ;-)

monovich
02-28-2003, 11:18 AM
Originally posted by ingo

For the price of a dual 2,8 Xeon machine i can buy a dual 1,25 GHz Mac AND a single 1 GHz Mac.
A comparable PC to the dual 1,25 Mac is in case of renderspeed a dual 2 GHz P4, and if you look at Dell you will see that they cost exactly the same (and i bet

I don't understand your math. I just priced a dual 2.8 Xenon box with Geforce 4800 and a gig of ram for around $2700 (yes, I'm looking at the dark side for LW work)

A stripped down G4 dual 1.25 + a single 1ghz G4 would cost $3500 minimum (as per the apple store).

Seems to me that PCs are a much better value.

Sure hope this IBM chip shows up soon, but, knowing Apple, I SERIOUSLY doubt wether or not it'll be out before 2004 (if at all). They just can't seem to keep up in the hardware department anymore. :(

ingo
02-28-2003, 01:42 PM
Huh, i'm at 3700 $ for a Dell dual 2.8, but i must admit i'm a bit lost in Dells crappy website shop. I think i should build a nice Mac PPC blade server for my renderings ;-)
And youre right, there is no sign to see any PowerPC chip from IBM in the future for the Mac, but anyway i still think the Moto's are better. Less power needed and less heat, with the 970 in its actual version you cant put it into a 17" powerbook, thats sad.

monovich
02-28-2003, 03:01 PM
ah, there's the discrepancy. I'm just taking about building up a generic PC box with part-by-part, and you were looking at a Dell. Just goes to show that Dell computers aren't a better value than Apple

Ed M.
02-28-2003, 05:41 PM
[[[And youre right, there is no sign to see any PowerPC chip from IBM in the future for the Mac, but anyway i still think the Moto's are better. Less power needed and less heat, with the 970 in its actual version you cant put it into a 17" powerbook, thats sad ]]]

uh.... Bullsh!t. At 1.2 GHz. on the .13 manufacturing process it only draws 19 watts... You were saying? Anyway, check your facts.


[[[Idle theories, gossip, ivory tower benchmarks, putative speeds are all truly exciting but it's answers to the questions above we need and that Apple themselves should be evangelising right now if they have a genuine concern for their high end users (many of whom seem to be considering the other side). In theory, a dual 2.5gig PPC 970 machine that outperforms a dual 5Gig P4 looks great [even if you could cool such a beast ;-)]. By mid 2004? Who knows? ]]]

Say what you want, The fact is that Intel and the Windows world are the ones that seem to be in a bind. There are several significant articles that you should be made aware of and then simply judge for yourself. The Win-tel-on platform seems to be in a pickle, and I'm not so sure we'll be seeing a 5 GHz. Pentium within the next year or maybe even longer, and besides, what's it going to get other than a bigger number to sticker onto a piece of silicon?

At this particular moment, It looks like a LOT of developers (and users) have hitched their wagon to a couple of "dead horses" so to speak -- the dead horses in this case seem to be Intel and Microsoft, BUT before you spark-up your flamethrowers, read the articles and my reasoning and then you be the judge.

These URLs all add up to the reasons that explain why:

Intel: No rush to 64-bit desktop
http://zdnet.com.com/2100-1103-985432.html
(This article kinda seals it...)

Intel signals the end of BIOS
http://zdnet.com.com/2100-1103-985600.html

(A little late with the BIOS stuff aren't they? Anyway, it just sounds like proprietary hack on OPEN Firmware... After all they NEED those DRM features. It also implies that this feature is going to be implemented on their 32-bit CPUs first because Palladium/Longhorn will require it)

Intel: Beyond gigahertz
http://techupdate.zdnet.com/techupdate/stories/main/0,14179,2902061,00.html
(Funny how they seem to be parroting what Apple has been saying for YEARS already -- typical AND it would seem that the MHz. thy've been preaching for so long will likely come around to bite them ;-)


I've hinted at all this before -- prior to any of those articles even being written. I've brought it up on various other message boards and now Intel goes and makes it *official* with their latest article. The fact is that the Windows world (with respect to it's users) is in a bit of a bind, and Windows doesn't seem to be as *portable* as Microsoft had boasted. It would appear that you guys were sold on a rather dubious idea over a long haul. In any event current users of Windows desktop configurations will be stuck at 32-bit for quite some time -- until the end of the decade or even longer. Imagine that.

That means 32-bit Pentium4's for the foreseeable future for anyone running ANYTHING on desk-top Windows (There is only a 64-bit SERVER version not meant for the desktop). Anyway, Let's look at the debacle a little closer.

Apparently 4 separate teams at Intel were given the task of coming up with a 64-bit desktop solution. After running a LOT of simulations they all decided that such a move would not be feasible -- for *them*. Here's the basics why:

There is no 64-bit desktop version of Windows (and it's unlikely that one is planned given Intel's latest announcement ),. Besides, it's already become too big and unwieldy in it's current 32-bit state (so let's add a few more ;-). So, the only 64-bit version of Windows is Windows Server 2003 and it only runs on the *Itanium* line and is only meant for *servers*. 32-bit apps running through emulation on the Itanium platform are EXTREMELY SLOW and anymore hacks to get it to run faster will likely increase the price and complexity and simply add to the gremlins that plague it, not to mention the additional expense/cost.

Itanium systems start at over $10,000. Not only that, but IA-64 is a completely different architecture from what's already out there. In effect, it essentially *orphans* a whole TON of applications already out and running on Windows 32-bit boxes. And here's where it gets even more squirrelly for Windows users...

AMD *is* planning on offering a 64-bit desktop processor, but it's 64-bit server line has already been ignored by Microsoft. For the desktop AMD-X86-64 would be yet another completely different architecture from Intel's IA-64 or even Intel's vaporware Yamhill-X86-64. Will Microsoft begin to support and develop *another* version of Windows another for the AMD x86-64 platform? It isn't clear that they will. It will more than likely be either one or the other... All while maintaining a whole sh!t-load of legacy 32-bit AMD and Intel boxes. In the end, it might leave AMD's platform to the LINUX camp. Sure developers could in theory port to LINUX, but then it would beg the question: if LINUX, then why not OS X? Then there is always Apple's X11. ;-) The point is... Offering yet another version of Windows for a completely different platform will just ADD to the confusion that a lot of endusers already face. Compatibility issues across each of those platforms will likely skyrocket as well. But this is only the start of the horrid situation. Consider that if it will be a task for Microsoft *and* Intel to deliver a 64-bit desktop version of their wares (at a reasonable price to desktop-endusers) then where does that leave developers? A: In an even worse situation...

Developers will be asked to support, maintain and ensure complete compatibility of their drivers and apps across 3 different versions of Windows from legacy to the new current architectures. That's a LOT of versions for the same "Windows" platform! Win32, WinIA-64 and WinX86-64-AMD and WinX86-64-Intel (if it ever makes it to silicon). What's more, none of these apps have even been dreamed up yet, and remember, emulation is out of the question because it yields poor performance for the $$ ;-) AMDs mode-switch doesn't seem to be that nice of a workaround either, so I doubt many developers will offer multiple versions of their applications for the same Windows platform.

This leaves Windows users (and developers) stuck at 32-bit for quite some time since the developers have effectively limited their options by supporting what seems to be an increasingly stagnating platform. Intel has stated it, and Microsoft has stated it. There will be no 64-bit desktop version of Windows for the desktop or an Intel chip aimed at the desktop for quite some time. Period. And AMD seems to be left with a processor but no apps, though a LINUX port is possible, but Apple seems to be making an argument for porting to OS X based on those LINUX ports too. Then there is IBM...

IBM is offering the PowerPC 970. A 64-bit chip aimed directly at desktop workstations in multi-proc configs that will run older 32-bit code natively and arguably even faster at the same clock-speed given the overall processor improvements with respect to pipelines and the bus - a chip that Apple is likely to adopt for *their* desktop workstations. Remember, Apple is ALREADY running a UNIX-based OS with THOUSANDS of commercial applications available. Still, you would imagine that IBM would want an applications running on their hardware and it's likely that IBM has far more clout with developers in that it reasonable to expect the apps to start rolling in. Then it gets back to the same question... If LINUX on IBM 970 then why not OS X on IBM 970? The point is that a good lot of you should have been sharp enough to foresee this sad situation. But it's not over yet. It gets even more horrid for the Windows desktop camp.

Suppose AMDs desktop CPU takes off and Microsoft endorses it with some as-yet-to-be-dreamed-up X86-64 version of desktop Windows OS and developers all line up with their applications all debugged and ready to go -- In less than 3 years? Where does that leave Intel, the company *hell-bent* on making Itanium succeed? It forces Intel to come up with ANOTHER HACK that keeps the company solidly bound to the *x86 specter* for quite some time. Now that's what I call ironic. In effect, Itanium would fail miserably and thus Intel's attempt to get away from that antiquated architecture known as X86,. and they will have to go back to the drawing-board this late in the game. If Microsoft decides to chance support for *multiple* versions of their already bloated OS for different platforms while attempting to secure, bug-fix and ensure compatibility across said versions, then developers and consumers will be rewarded with even MORE of that confusion.

All this said, it looks like Windows-users will be running exclusively on proprietary Windows 32-bit for quite some time. The bottom line is that Intel and Microsoft seem to have been caught with their pants down and not knowing what to do they simply stick to Win32 on Pentium for desktops, effectively missing the 64-bit desktop boat. It seems to be a nice way to reward all their loyal customers who were initially sold on the ideas that those companies were promoting. Isn't it time the industry moved on? Someone tell me why Windows on Intel was a good choice again for the long term?

--
Ed M.

Ed M.
02-28-2003, 05:42 PM
In conclusion, it doesn't seem likely that we'll see a 64-bit DESKTOP solution from M$ or Intel in the next 5 years. It would Just add to the confusion, chaos and compatibility issues that developers and endusers already need to consider and I'm sure that this factored in heavily into Intel's simulations. I mean what about all those nice brandy-new desktops that were just sold to all you guys this year? Those legacy systems aren't going to go away any time soon, ya know... Are people going to be required to upgrade *again* ? Will OEMs suddenly drop support for those systems in favor of moving the masses along? It might have considerable blow-back.. people might indeed update -- to a platform that's much more forward looking. Just some things to consider...

--
Ed

robewil
02-28-2003, 06:31 PM
Ed

While I don't agree that the Wintel world will wait 5 years or longer for the next significant jump in performance, I acknowledge that it is conceivable that the Apple world may leapfrog it someday.

The point is that day is not today, nor was it a year ago or even longer than that. I, personally have had the philosophy that every 3-5 years or so, its time to evaluate what my needs will be for the next 3-5 years. Thusly, I was an Amiga user for years and then migrated to Macintosh. But around 1996, I felt it was time to go to PC. Lightwave was a very big reason I made this decision. I have never regretted this. However, if in fact I see Apple being the leading way to go in the future, that's where I'll be. Companies and individuals alike have to re-invest in hardware and software every so many years anyways so I never really see why its this platform evangelism is such a big deal.

It is my theory that Mac people especially, are very defensive because they chose against the norm. Therefore, they are in a state of mind that they constantly have to defend this decision. What's going to happen on the day Apple comes out with a machine that actually does soundly beat the performance of a top of the line Intle or AMD-based machine? I suspect you and many of your Machead friends will be shouting "Nyah, nyah, nyah" because suddenly your decision to be a Machead has been validated and us Windows users were such fools all these years because we chose to make our livings on a platform that at some particular time and date is not at the top of the food chain.

Go ahead and keep believing that this day will come but if your expecting Windows users to bow their heads in shame or say "Gee, Mac people were right all along, what a doofus I am for going with Intel/AMD/Microsoft.", I think you're going to be disappointed.

Ed M.
02-28-2003, 06:58 PM
No, unless you can find some other articles quoting Intel execs about *not* sticking with 32-bit, then it's safe to assume that the above articles all add up to 32-bit desktops running Windows for quite some time. Hey, don't blame anyone but Intel and M$ for painting themselves into a corner (and screwing you guys)...

--
Ed M.

Beamtracer
02-28-2003, 09:14 PM
I actually think AMD will produce a viable 64-bit desktop processor within the next year. It's running late, as processors always do, but they have been able to exhibit crude prototypes running a 64-bit version of Windows. Intel seems to be in deeper trouble with their disappointing Itanium (Itanic) processor.

The IBM 970 processor will go into manufacturing production mid-2003. That's not far away. Apple has a pretty good record of getting new hardware technology into their machines very quickly.

Apple won't make official announcements until the last minute, as they don't want to disrupt the sales of their current machines. However test boxes have been seeded to various developers.

Lynx3d
03-01-2003, 07:27 AM
we'll see where the PPC970 really ends up...

The only speculations you can somewhat compare are Spec results, what i could find so far:
SPECFP Base:
1.8GHz PPC970 estimated: 1051***
Motorola G4 1GHz: Somewhere below 200
Athlon 64 1,2GHz: 719*
Pentium IV 3,06GHz: 1077**
Itanium 2 1GHz: 1431**
(so much for itanic...itanium2 definitely has its uses)


*measured by cīt, using SSE2, 32bit, estimated 1274 at 2GHz
see heise news (german) (http://www.heise.de/newsticker/result.xhtml?url=/newsticker/data/cp-29.12.02-000/default.shtml&words=Hammer%20SPEC)
**SPEC.org database:
http://www.spec.org/osg/cpu2000/results/cfp2000.html
*** found here (http://www.realworldtech.com/forums/index.cfm?action=detail&PostNum=986&Thread=1&entryID=10790&roomID=11)

Besides the fact SPEC is VERY dependant on compiler optimizations (guess the main reason there is no usable G4 results) it doesn't translate to rendertimes in any way, the Athlon XP has much lower SPECFP results than the Athlon 64 at same clock speed, but renders almost identically fast in POVRay...

It's all just hot air what's written here IMHO.
Wait for the day until you can use your PPC970 workstation with 64bit software and then we'll see how PC stack up against it.

(-edit- oops, looked wrong on the Athlon 64 results...and typos)

Ed M.
03-01-2003, 05:01 PM
[[[The IBM 970 processor will go into manufacturing production mid-2003. That's not far away. ]]]

Beam, How does March sound instead? ;-)

--
Ed M.

Ed M.
03-01-2003, 05:02 PM
Lynx3d...

[[[we'll see where the PPC970 really ends up... ]]]

We will indeed, but it is very important to keep in mind that this PPC 970 chip is only the FIRST in a series where there will likely be one more enhancement and manufacture on .09 process before moving to a POWER5 derivative (Aside: Power5 will be 4x the performance of the current Power4 and run vastly cooler). It's also very important to note the BANDWIDTH of this monster as well as the additional FP units and Integer units. The bandwidth alone should cause AltiVec ops to be off the scale! It's also great for SMP where bandwidth is needed. This thing is gonna slap the living crap out of any Pentium CPU that's out at the time. Perhaps you should read the above articles that I linked to so you can get a better understanding of where Intel and Microsoft want to take the Desktop market... 32-bit desktop line with new proprietary implementation of Open Firmware that implements a ton of DRM so it will be able to run the next versions of Windows. This will be implemented on 32-bit processors FIRST and isn't expected to debut until 2005... That's 2 years.. Then there is the laptops to worry about. Can't put an Itanic or AMD-64 in one of those so, again, a fork... I mean let's face it, if FOUR (4) separate teams at Intel ran simulations and delved deeply into when 64-bit Wintel desktops will be viable and ALL came to the same conclusion that it will happen by the end of the decade at the earliest, then it's safe to assume that it is indeed the case. As for the AMD comment and a rumored version of Win-64 running on it... All I have to say is that it isn't a DESKTOP version of Windows, it's the SERVER version. There's a difference and not only in price (which is also a major factor) Oh, and then there's those apps to worry about... Oh sure it can run 32-bit apps (supposedly), but you have to boot into the correct *mode* (64-bit or 32-bit).. It will be either one or the other. Furthermore, before you go and mention that LINUX is ready for that puppy, consider this...

Aside: I'm going to hack post together from a bunch of other posts I've written simply because it's easier and faster and the point will be well made... Bits and pieces have already been gathered from posts on his site as well. Anyway...

I don't think Linux will come to the *mainstream desktop* arena anytime soon. To be honest, Linux isn't developed enough (User Interface wise) and there are hardly ANY commercial apps that run on all of the various flavors of Linux that are out there... So, the big hurdle is convincing applications developers that Linux-PPC, Linux-x86 or Linux ANYTHING is viable to develop for, and two things point to it NOT being viable for mainstream work.

First, Linux has been around for about 10 good years or so; its even been on the Amiga. It has been around for x86 machines for a LONG time and probably has a good installed base already. The point is that the people who are using the OS aren't in the market that developers and OEMs need to be in for them make money... And you can forget about support.

Is there any *consistent* PC hardware or how about simple configs between any given OEMs? No. There are a LOT of people complaining about peripherals and drivers et. al. not working correctly or as expected, and this is on WINDOWS! Imagine what it would be like for Linux if major developers were interested in a port. Windows alone is chaotic and unpredictable simply because M$ has to support all kinds of arbitrary hardware in a single software release. Many developers call this a recipe for disaster, and indeed it is. Now note the fact that LINUX is *forked* Where is the consistency? How long before that's ironed out?

The people who use Linux are the "geeky types" and solderheads. Not people at home (which is the BIGGEST market) or people in the creative fields. So if you don't have mainstream apps like Office, Photoshop, Illustrator, Quark, Maya, AutoCAD etc... as well as apps and a User Interface that's consistent and familiar ALREADY (i.e. today), the outlook for the apps coming in the NEAR future is bleak. Linux has been around too damn long already and really hasn't amounted to anything but a really nice low to midrange server OS and is just now breaking into the desktop. What took so long?

Linux development is moving more slowly than people think... Again, it's been out in the open-source community for how long now?? A very long time.

So, the question still remains, who is going to bring all these new and wonderful apps to the LINUX platform? Which apps will they be? Which flavors of Linux will these apps be compatible with? How long will it take the apps to develop and mature on this new platform? Will there be consistency of interface through ALL flavors of Linux UI? the questions keep mounting... OS X on the other hand is a completely different story...

OS X has been out for a very short time and has ALREADY passed Linux in probably every functional way imaginable and its evolution is progressing at an incredibly fast rate. OS X is mature and consistent on the desktop TODAY and it will only get better. It has a UI that's far more robust, advanced and usable than ANYTHING on Linux AND it's more familiar to people AND it has a larger installed base (OS X is the #1 shipping UNIX in volume -- commercial or otherwise). It's had its shakedown period and more importantly... It already HAS A LOT OF GREAT COMMERCIAL APPLICATIONS as well as developer commitment. Can Linux even begin to say that? No.

And lets not forget this article:

http://www.theregister.co.uk/content/4/24060.html

Or this article:

http://www.zdnet.com/anchordesk/stories/story/0,10738,2773365,00.html

To quota a few of my friends on their views of LINUX...

[[[On the grand scheme of things, people just don't trust Linux enough to bet their entire livelihood on it. They toy with it, prod it, but in the end, they don't switch their entire computer base over to it. Without that kind of commitment, Linux will remain in the hobbyist arena (when talking home desktop). ]]] - Pamela J. Miles

[[[There's a lot more to a desktop platform than just the OS; it's the entire infrastructure that matters. A solid desktop OS needs all manner of support from font foundries, file conversion utilities, installers and a general ability to open and work with documents across all other platforms in friendly fashion.

In general, my personal experience has been fairly grim when it comes to these issues (regarding LINUX). OS X provides an answer to all of this that is so strong that the question of desktop LINUX has gone from "How?" to "Why?" Those who like X86 boxes will slide glacier-like to Windows while the independent folks will tend to Mac-OS X and a mainstream Unix with a robust interface and mature applications.]]] - Del Miller, Aerospace Engineer

[[[Linux, and Unix interfaces in general are designed by committee. Large groups of people hashing, arguing, testing ideas, until they get the one that they can all agree on.

Examples of this are: KDE CDE OpenWindows WorkPlace Shell Gnome Windows...

All of them horrid.

To design a good interface, you need talent, but you also need vision. Committees never have vision. They have meetings. ]]] - John C. Welch


[[[There's sort of a fallacy with LINUX. LINUX never has been, and never will be a viable consumer desktop. To be such, it would have to be designed for the consumer.

LINUX is what it has always been -- a reasonable implementation of a UNIX based operating system, that is not bad as a low-end server (that has grown into the mid-end -- and may someday grow higher). It is not a bad foundation to build a turnkey system for some enterprises to use (as turnkey solutions go. IBM?). But that is not the same as "consumer desktop".

The ultimate operating system is not a command line with a thin graphics shell on top. So, LINUX has never really been anything close to a desktop solution (let alone a consumer desktop solution) by anyone but the completely self-deluded. ]]] - David K. Every

[[[OSX is already the unit leader in terms of a UNIX distribution, BUT... Other people that rely on UNIX (corporate entities et. al.) are learning that IT IS ALREADY THE INTERFACE LEADER as well. And this is 12 months out of the gate. Jaguar looks to fix/improve many things... And in another 12-18 months, I think it will be the standard by which most UNIXXES are measured... OSX delivers today on what LINUX has been promising for 10 years... ]]] - Dave K. Every

[[[I'm one of the few people that believes that Linux is at a point where it (as an operating system) *is* worthy of desktop status... but the kicker is that people don't use operating systems... they use applications... until more consumer friendly software starts appearing for Linux... it will have little impact... and by consumer software... i don't mean products like word, a web browser, and quicken... (there are already open source products like that already... instead, there needs to be more kids software and stuff accessible from companies that issues software to use their hardware as well as many other application ... when these products come about ... then Linux will be able to start functioning in the desktop arena. ]]] - Kelly McNeill

So, the larger picture shows that people simply don't trust Linux as a viable solution. Sure, People will always mess around with it and there will always be open-source development and such, but in the end, I'm not convinced that these people and organizations are going to trust it enough to switch their entire computer base over to it. Simply put, it isn't what Linux was intended for and there are just too many additional hurdles to clear before any of this can even begin to sound viable. Will everyone be doing their own in-house, specialized development and support? I doubt it.

--
Ed M.

js33
03-01-2003, 06:33 PM
Hi Ed,

I read all your posts in this thread and you are obviously very excited about the 970 processor. But until Apple says anything publicly it's vaporware until then (at least on the Mac).

I hope Apple decides to use it. Unless Motorola can pull their heads out Apple has no choice but to switch to the 970. After all who else makes PPC processors? Nobody.

So do you think the 970 will eventually be the entry level processor and the full Power 4 or 5 will become the high end?

Also what do speculate the price of one of these wonders will be?

If Apple regains the performance crown but shoots themselves in the foot with outrageous prices then I don't see a mad flock to the machine except for the high end users and that market is pretty small on the Mac compared to Wintel.

I agree with your comments about Linux vs. OSX. OSX has throughly quicked Linux *** in the desktop market for Unix OSs.

Also all talk about 64 bit apps is going to vaporware for awhile so what good will a highend 64 bit CPU do you if the apps aren't ready for a couple years? Look at Quark. They still haven't released an OSX version yet and OSX has been out how long now (3 yrs?). So if it takes a major vendor that long to release a 32bit OSX version how long will a 64bit version take?

I believe LW will be ready alot sooner than most if the 64bit chips materialize soon.

I agree with most of your points about the Wintel 64bit roadmap but history will show that no matter what the current direction looks like they will be able to sort it out to stay competitive in price and performance. Just 10 years ago a top end PC was a 486 running at 50Mhz. Now it is a 3 Ghz machine. That's 60 times faster in 10 years.

Also if the 970 machines are say over 3 grand then someone can just get a dual 3 Ghz and still have more speed cheaper.

Cheers,
JS

Lynx3d
03-01-2003, 06:53 PM
Oh, and then there's those apps to worry about... Oh sure it can run 32-bit apps (supposedly), but you have to boot into the correct *mode* (64-bit or 32-bit).. It will be either one or the other.

Erm that's not true.
From what i understand a 64bit OS running on hammer systems in "long mode" still can execute 32bit code (in compatibility mode) without recompile.
It's only that the real and virtual mode are unavailable, so obviously an old OS or things like DOS programs wouldn't run in long mode, only in Legacy mode.

Actually i've just read that on a PPC970 system you won't be able to install the current OSes at all, because they need some changes. However applications don't need to be recompiled:
http://www.theregister.co.uk/content/3/27621.html

Ed M.
03-01-2003, 07:21 PM
[[[Also what do speculate the price of one of these wonders will be? ]]]

Cheap, given that IBM plans to use a LOT of them in blade servers that will compete directly with Intel (Dull..err I mean Dell?) Dell has to be ****ting knowing that IBM is going to be releasing some kick-*** servers using these puppies.

[[[If Apple regains the performance crown but shoots themselves in the foot with outrageous prices then I don't see a mad flock to the machine except for the high end users and that market is pretty small on the Mac compared to Wintel. ]]]

Doubt it, there is a LOTof pent-up demand and if the machines do perform as predicted, Apple's current pricing is MORE THAN JUSTIFIED.

[[[Also all talk about 64 bit apps is going to vaporware for a while so what good will a highend 64 bit CPU do you if the apps aren't ready for a couple years?]]]

That's not the point, who cares if the are no 64-bit apps to run on this thing initially, there will likely be a few. The point is developers will have a system that they can adopt to over time -- they can migrate with confidence overtime. Which developers will be ready for Wintel-64 bit desktops when (and if) it arrives later this decade? Developer tools? Applications development? Stability? By that time, the 64-bitt PPC market will have been well along.

[[[I believe LW will be ready a lot sooner than most if the 64bit chips materialize soon.]]]

I hope so. NewTek should look at this as an advantage.

[[[Also if the 970 machines are say over 3 grand then someone can just get a dual 3 Ghz and still have more speed cheaper. ]]]

Do you REALLY think a dual 3GHz P4 will be faster than a dual 970? Or even a single 970?? By that time Apple will be on the Power5 derivative and the Power5 is 4x's the performance of the Power4, so imagine the derivative chip (YIKES!) Remember, the 970 is only the FIRST in a series ;-)

--
Ed M.

js33
03-01-2003, 07:26 PM
Hi Ed,

I read all your posts in this thread and you are obviously very excited about the 970 processor. But until Apple says anything publicly it's vaporware until then (at least on the Mac).

I hope Apple decides to use it. Unless Motorola can pull their heads out Apple has no choice but to switch to the 970. After all who else makes PPC processors? Nobody.

So do you think the 970 will eventually be the entry level processor and the full Power 4 or 5 will become the high end?

Also what do speculate the price of one of these wonders will be?

If Apple regains the performance crown but shoots themselves in the foot with outrageous prices then I don't see a mad flock to the machine except for the high end users and that market is pretty small on the Mac compared to Wintel.

I agree with your comments about Linux vs. OSX. OSX has throughly quicked Linux *** in the desktop market for Unix OSs.

Also all talk about 64 bit apps is going to vaporware for awhile so what good will a highend 64 bit CPU do you if the apps aren't ready for a couple years? Look at Quark. They still haven't released an OSX version yet and OSX has been out how long now (3 yrs?). So if it takes a major vendor that long to release a 32bit OSX version how long will a 64bit version take?

I believe LW will be ready alot sooner than most if the 64bit chips materialize soon.

I agree with most of your points about the Wintel 64bit roadmap but history will show that no matter what the current direction looks like they will be able to sort it out to stay competitive in price and performance. Just 10 years ago a top end PC was a 486 running at 50Mhz. Now it is a 3 Ghz machine. That's 60 times faster in 10 years.

Also if the 970 machines are say over 3 grand then someone can just get a dual 3 Ghz and still have more speed cheaper.

Cheers,
JS

Ed M.
03-01-2003, 07:27 PM
[[[Erm that's not true.
From what I understand a 64bit OS running on hammer systems in "long mode" still can execute 32bit code (in compatibility mode) without recompile.
It's only that the real and virtual mode are unavailable, so obviously an old OS or things like DOS programs wouldn't run in long mode, only in Legacy mode. ]]]

LOL! This is EXACTLY what I'm talking about... A LOT of people aren't going to want to be bothered with all this "mode this" , "mode that mode" stuff. it should work transparently, otherwise it's a hack and will feel extremely *kludgey*to the enduser, and of course mom, pops and aunt Mary won't be running one of these things ;-) More confusion, more incompatibility... More chaos.

--
Ed M.

Ed M.
03-01-2003, 08:17 PM
js33...

You mentioned that it's all speculation whether Apple will be using this processor or not. Well, the one GLARING hint that should be staring you in the face is the fact that IBM actually licensed the *name* (not the technology, since they own it as much as Apple and Motorola) AltiVec. It's official. They are using AltiVec. That said, what 64-bit desktop/comercial LINUX-PPC OS's currently utilize AltiVec? For that matter, what 32-bit comercial LINUX apps utilize it? I'm talking outside the embedded space. I'm specifically referring to apps for servers and desktop..

A: NONE (or some very few obscure ones).

Then ask yourself... Why would IBM make a conscious decision to utilize AltiVec on these processors and furthermore, would they *need* to?

On the other hand, What OS would obviously fit the bill in this case? What OS currently makes extensive (with much more to come) use of AltiVec? What OS has the support for MANY full-blown, time-tested, mature, robust and *familiar* applications that utilize AltiVec to some extent ranging from *some* to a lot?

Yep, you guessed right... Mac OS X. When you sit down and think about it, all arrows point to it. It really is *that* simple.

--
Ed M.

js33
03-01-2003, 08:25 PM
Sorry for the double post...my bad.

Yeah Ed everything seems to point to Apple using the 970 and it will be interesting to see how it all plays out.

Cheers,
JS

Lynx3d
03-01-2003, 08:26 PM
What are you actually talking about?
Practically "real mode" is already unavailable for everyone out there using WinME or later OSes.
Everything Windows guys use today runs in protected mode, and that's exactly what a Hammer with 64bit OS can run, and the user has to do nothing different then he ever did.

And how do you think a PPC 970 will run 32bit applications? For sure not in 64bit mode, it switches modes aswell, and again the OS has to do it. Only difference is that it uses the same registeres and ignores the higher bits. A bit more elegant but no substantial difference.

Ed M.
03-01-2003, 08:59 PM
Wow, js.. This is a first in quite some time... When was the last time you actually agreed with any of my posts? lol

Anyway, yeah, It's going to.

Lynx3d... The Hammer isn't even out yet. What's more, it looks to be a long ways off. How long has it been delayed already? I hate to say it, but it looks like there will only be ONE Windows desktop CPU this time around. I mentioned the debacle they are in, above in my previous posts. Not only that, but ALL the Wintel processors will *NEED* to be running some variation of LaGrande in order to run Palladium/Longhorn. Yep, all the nice DRM stuff M$ is forcing Intel (and Windows users) to adopt and implement. So, either AMD is burdened with more hacks trying to compete against Intel or they will be relegated to the Linux world. Either way, it's unlikely that they will be able to take on IBM there... That's probably why they've formed a partnership with them.

http://www.geek.com/news/geeknews/2003Feb/wbc20030221018774.htm

http://news.com.com/2100-1001-957757.html
http://www.tech-report.com/onearticle.x/4230
http://www.infosatellite.com/news/2001/11/h121101amd_future.html
http://news.com.com/2100-1001-957757.html?tag=fd_top

--
Ed M.

js33
03-01-2003, 10:24 PM
[[[Wow, js.. This is a first in quite some time... When was the last time you actually agreed with any of my posts? lol

Anyway, yeah, It's going to.]]]

Hehehe,

Yes Ed I don't often agree with you but I agree the 970 is coming out and that Apple will probably adopt it as they really have no other choice at this point. The G4 is so long in the tooth now that if Apple doesn't do something soon they might vanish altogether.

Wintel has alot more time to get their 64 bit ducks in a row and processors than Apple does.

I think it will still be awhile before anyone is foaming at the mouth must have 64 bit now.

The Wintel camp also won't sit idlely by as far as performance goes. Also it will be awhile before we have realtime rendering for anything other than games and the 64 bit ability to address more memory will be good but it won't really push the envelope anymore than progress has in the last 10 years.

Cheers,
JS

Ade
03-01-2003, 10:39 PM
The G4 is soo long in the tooth that it wont be till the 7457RM revision that supports DDR properly!
They still havent used the 7457 chip yet (rumoured to be the 1.42), and thats not the RM version!

malanoski
03-03-2003, 02:49 PM
I sit here reading the post's and ponder my past in this dang buss. I started out apple went to sgi then to pc and now back to apple. I've used just about every 3d program out there to make a living except houdini. I have a widows machine at home and a mac at work. yes my pc machine renders circles around the mac but the operating system is trash, i spent the last weekend trying to figure out why I got the blue screen of death and not doing any work in lightwave. if you want to switch for the speed be prepared for maintance it's a regular thing. I have never had a problem like that with OSX.
I no I will not by another pc with windows on it
unless it is for a render farm can't beat the price/performance but for the actual work I will use the mac hopfully they will get the speed up but more important to me would be graphics. clients will always want the product yesterday.

I heard from a apple rep when i was sitting in on an OSX server show at work that he was wondering what intel was going to do when they put the new chips in. We all asked him questions but he just kind of chuckeld but would only imply that it would be this year (at least for the server.) I think you will see the new chip in the server first or possably at the same time as a new desktop if the hold the release. After the talk someone asked if the new chip would be called a g5 and made by motorola and when would we see it in a desktop. the answere was most likely g5, not motorola and he would be supprised if we did not see one before the end of the year but he was "speculating of course" sounds like there working the bugs out of the chip with the os running on it.

Beamtracer
03-10-2003, 07:06 PM
PowerPC 970 to be announced at Apple's World Wide Developers Conference, May 19, 2003. That's what rumor sites MacRumors and Macbidouille are saying. This isn't the release of 970 powered machines to the public, however it is the official presentation of the processor to developers.


From MacRumors:

Peter Sandon of IBM will be presenting the chip at WWDC... and certain ADC developers will be able to see a demonstration of a prototype 970 Macintosh.

Development tools are being finalized and there is rumored to be a beta of a 64-bit Mac OS X.

Macbidouille also claims that while the 2.5GHz 970's can reach 2.5GHz at 0.13microns, it requires a significant amount of power and generates too much heat.

Newtek: I hope you will be sending a delegation of your top Lightwave engineers to attend this event!!!

Beamtracer
03-24-2003, 06:41 AM
An interesting article from the British IT magazine, The Register, about mounting speculation that Apple will announce the IBM970 processor at the World Wide Developers Conference:

http://www.theregister.co.uk/content/39/29898.html