No 3GHz Soon and No G5 Laptop

[nightcap965]

A Long Hot Summer

Steve has messed up big time here. Didn't his parents ever tell him you shouldn't make promises you can't keep? Apple will take heavy losses over this; people like me who need all the processing power they can get who won't buy until the 3G target is achieved, and many G5 owners who won't now upgrade to rev.2 machines.

Bad for reputation and disasterous for sales - a PR double whammy!

You're absolutely right. I'm a G5 owner. I won't upgrade to the new revision, and I'll bet most other G5 owners won't, either, because the platform is only nine months old! I know computers age quickly, but I think my dual 2 GHz G5 still has a few months left in her. Sheesh!

 

[Hiroshige]

but yes G5 iMac

I have been trying to read most of this massive thread but has anybody realized that Steve would definitely keep something up his sleave for WWDC and it is going to be more than just displays. If they are confident of yields on 2.5 Ghz 970fx's, then they must be even more confident of yields at, perhaps 1.6 Ghz=G5 iMac. If they can make the 2Ghz fit in that slim X-serve, they can do even better on the 1.6.

 

[~Shard~]

I have been trying to read most of this massive thread but has anybody realized that Steve would definitely keep something up his sleave for WWDC and it is going to be more than just displays. If they are confident of yields on 2.5 Ghz 970fx's, then they must be even more confident of yields at, perhaps 1.6 Ghz=G5 iMac. If they can make the 2Ghz fit in that slim X-serve, they can do even better on the 1.6.

Yes, but don't forget the xServe essentially has a cold air intake on the front and huge airflow provisions - this would still need to be addressed for the G5 iMacs. Not saying it's impossible, but it's still something that needs to be considered and addressed.

 

[Downdivx]

imac form factor change

Seems to me that if Apple is upgrading the Imac to a G5, you would change the form factor - for better airflow and a new look.
W

 

[AidenShaw]

How can it be a big success if it's never been put into production?

VT didn't "forget" ECC. The guy behind it discussed that issue early on, including the fact that certain calculations would need to be run twice and that the speed of Big Mac was STILL better per dollar...

Not "certain calculations", but *every* calculation. Even good memory gets random errors, so partial checking would be useless. Every calculation would need to be done twice and compared, and recalculated several times and "voted" if the results differed.

And what do you do when the error causes a panic, or if the error happens when comparing?

One could argue that the Top500 rules should demand ECC, but they don't.

No, the rules shouldn't specify technology.

The rules should specify that you get the correct results, however. And obviously VATech didn't use the "do it twice and compare" logic while running the benchmark.

"Obviously" because they got over 50% efficiency, and "obviously" because the system was too unstable to run real science (although the benchmark apparently completed normally).

VT didn't "forget" ECC.

True, but they've realized what a big mistake they made and tore the machine down even before going into production use.

The "down" time is scheduling, timing, and funding reality, with an excellent outcome.

No, the down time is because the first machine failed. It was not usable, and they had to take the risk to completely replace the machine.

Had the PowerMac cluster been a success, it would still be running. VATech would have shut down a single 12-node rack, upgraded it to XServes, tested it offline, and then introduced the new rack into the cluster.

Then, as XServes arrived from Apple, one-by-one the PowerMac racks could be shutdown, upgraded, and restarted.

The cluster could have been upgraded without downtime - but the problem was that it was already "down" due to instability.


The VATech cluster is a "success" only to Mac fans, the big new clusters are being built with Linux on Opterons, Itaniums and Xeons. Even Dr. Varadarajan's company is headlining its new 20 TFLOP Itanium cluster (built in only 5 months) on its home page. (http://www.californiadigital.com)

 

[JFreak]

Yes, I agree completely; I think that the only people who really have use for even the lowest-end G5s are video professionals / people doing a lot of rendering-intensive tasks.

or audio professionals like me. i have hard time accepting the performance of a 1.25GHz G4 powerbook when i do real-time 24bit audio streaming for about 30 audio channels simultaneously - in other words, when i do live mixing in front of the house. i can really stretch this powerbook to the maxx and would KILL if i could have the performance of a mid-powermac portable enough.

and for those who complain: please, stop complaining because this was a good upgrade. the 3GHz talk was only hype that you stupid people took as a word from god. or maybe steve is a god for some people? hard to say. but anyway, this is good. if you need the speed, then buy the fastest. if you don't NEED it, stop looking at the specs and benchmarks. just use your machine. or if you don't have to use it, you could save some serious money by not buying one at all.

 

[maverick13]

The VATech cluster is a "success" only to Mac fans, the big new clusters are being built with Linux on Opterons, Itaniums and Xeons. Even Dr. Varadarajan's company is headlining its new 20 TFLOP Itanium cluster (built in only 5 months) on its home page. (http://www.californiadigital.com)

So ? As I mentioned earlier there is a 40 teraflop G5 supercomputer being built . That's better than current top 1.

http://www-306.ibm.com/chips/products/powerpc/newsletter/jun2004/ppc_process_at_work.html

 

[nagromme]

Originally Posted by nagromme
The "down" time is scheduling, timing, and funding reality, with an excellent outcome.

No, the down time is because the first machine failed. It was not usable, and they had to take the risk to completely replace the machine.

You quoted exactly the relevant statement I made, and yet, somehow, you don't understand it.

The first cluster made the deadline--THAT was all the "success" they needed for the NSF funding. Apple made that happen when no other choice could--VT was negotiating with Dell and others, but nothing else could give them the benchmark they needed for such a low price. The first phase of the cluster didn't NEED to be ideal for science--it needed to meet other people's rules by a certain date. If only it could have done both--made the list AND offered the power they are getting from Xserves! But nothing, at that budget, could have done either one.

And now, they do need to upgrade it for good performance on "real science." Otherwise they'd be running things twice, as they admitted from the start. No argument--the original cluster would have been far slower (but still cost-effective) for real research than the new one is. Thus the new cluster. Why not do it right the first time? Because the Xserve G5 didn't exist, and the deadline wasn't in VT's control. The choice of G5 towers just to make a deadline is utterly obivous. I say again:

The "down" time is scheduling, timing, and funding reality, with an excellent outcome.

Why? Because the "down" time is to replace a system that got needed funding with one that's twice as fast for research. It just can't be said any simpler. They are now getting FULL G5 power with ECC at the same original low price.

TWO huge successes--unless emotional needs lead you to ignore the realities of the situation. That's unreasoning and unconvincing.

VT is not as stupid as you make them out to be.

 

[~Shard~]

Seems to me that if Apple is upgrading the Imac to a G5, you would change the form factor - for better airflow and a new look.
W

Well duh - I think that goes without saying. I think the only thing we can say we know for sure with the new G5 iMacs is that they will have a new form factor - everything else is anyone's guess.

 

[yuphorix]

TWO huge successes--unless emotional needs lead you to ignore the realities of the situation. That's unreasoning and unconvincing.

VT is not as stupid as you make them out to be.

Couple of things i'd like to mention. First off, you've essentially proven my point that the original G5 cluster is just a publicity machine to get them on the Top500 list. You've also proven my point that $4million dollars is still within the range of the price a university can pay for a publicity event.

And trust me, calculating twice doesn't just solve the errors... it's alot more complicated than that.

I don't consider the original cluster a success at all becaused it failed to do anything but prove to IBM that they make one hell of a FLOP cruncher! And obviously, your Mac-ego seems to have taken you over because you claim there are two huge successes when only one Cluster has been completed so far.

 

[dr.Zoidberg]

for sure?

Well duh - I think that goes without saying. I think the only thing we can say we know for sure with the new G5 iMacs is that they will have a new form factor - everything else is anyone's guess.

dude, this is Apple we are speculating about, NOTHING is for sure.....and on that note....What if Apple decides to make the new iMac a dual G4?? That would distingiush it as a consumer machine for sure. Just a thought.

 

[pjkelnhofer]

Seems to me that if Apple is upgrading the Imac to a G5, you would change the form factor - for better airflow and a new look.
W

Now that Apple has introduced water cooling, has anyone thought it could become a part of the new redesigned iMac?

 

[AidenShaw]

That's not a G5 cluster....

So ? As I mentioned earlier there is a 40 teraflop G5 supercomputer being built . That's better than current top 1.

http://www-306.ibm.com/chips/products/powerpc/newsletter/jun2004/ppc_process_at_work.html

That's not a G5 cluster, it's an IBM BladeCenter cluster running Linux. You get two 2.2GHz PPC970fx chips and IBM chipsets in what is effectively a 0.5U system, with datacenter level reliability and remote management features built into the chassis itself.

It's funny - when the VA Tech cluster was first announced I said that it would be great publicity for the IBM PPC blades.... How true.

And, unless IBM's yields improve, it'll mean that 2300 PowerMacs won't be able to be built!

(It's also interesting that the page says that they'll use 4564 2.2GHz CPUs, when the top speed currently available is shown as 1.6GHz.)

 

[maverick13]

That's not a G5 cluster, it's an IBM BladeCenter cluster running Linux. You get two 2.2GHz PPC970fx chips and IBM chipsets in what is effectively a 0.5U system, with datacenter level reliability and remote management features built into the chassis itself.

It's funny - when the VA Tech cluster was first announced I said that it would be great publicity for the IBM PPC blades.... How true.

And, unless IBM's yields improve, it'll mean that 2300 PowerMacs won't be able to be built!

(It's also interesting that the page says that they'll use 4564 2.2GHz CPUs, when the top speed currently available is shown as 1.6GHz.)

that's not a G5 that's an IBM cluster running linux. Thanx for the laugh.

G5 is not a synonym to Mac u know. I didn't talk about a Mac Cluster I talked about a G5 cluster. You know PowerPC G5? That's PowerPC 970FX

 

[Hiroshige]

AidenShaw shows his true colors

And, unless IBM's yields improve, it'll mean that 2300 PowerMacs won't be able to be built!

He's really somebody who hates the Mac! He's a troll! Shrek is his first cousin!

 

[Macster389]

Apple just loves to piss me off this is why I am just going to Build a PC , **** Apple , atleast they could have put out something better in the lower models. a single/dual 2ghz and Dual 2.2 low and mid end would have been great. and come on $3000 for a comp and all we get is a 2nd rate soon to be outdated and replaced Radeon 9600XT that should be minimum . a 9800XT should be standard on a 3K machine , i'd love to see Dell , HP or AlienWare try to pull this Bull **** off. ALL APPLE HAS DONE IS UPDATE 1 MACHINE AND NOW IS TRYING 2 SELL US IT'S OVER STOCKED CRAP AT A DISCOUNT.

While i like the Liquid cooling alot i'm disappointed to see it's only on the High end. NOTE TO APPLE: AMD NOW HAS AN ATHLON 64 +3800. They will have a +4000 by Oct/Nov or sooner.

Also Why did they ever bother to put in a PCI-X Slot in. this is a complete waste as PCI-X will be used Exclusively for the Server Market, The PC Industry has already Picked the Better and faster PCI Express as the new Standard to replace PCI making PCI-X a total waste unless ur running an X-Serve. Doesn't Apple Realize i can Biuld me a Spanking new +3800/FX-53 system for under 2K , and don't let me jump on those Phoney Benchmarks where the new G5 2.5 beats the AMD 64 FX-53 by 93% .....LMAO yea right.

I said it before. Steve is greedy, He charges way to much for out dated product! This revision if you could call it that is lame. Why not sell the 1.8 G5 for $1200 and add a surcharge of $800 to keep Apple afloat. This is infact what he is doing! And counting on the loyalty of the Mac croud to pay the PRICE. A 5200 fx card are f...ing kidding. This time he is trying to see how far he can push it and still get us to pay the price! I hope all of you will hold off from ordering! We waited 1 year for this SH.T!! Why not wait 6 more months for something better and at a better price. If the G5 sales crash over the next 6 months Steve would be forced to offer something better and at a better price. Remember your cash is what Steve needs MAKE HIM WORK FOR IT. If you were a PC person and Dell or HP gave you an update like this one you wouldn't buy it!!!!!

 

[Mav451]

that's not a G5 that's an IBM cluster running linux. Thanx for the laugh.

G5 is not a synonym to Mac u know. I didn't talk about a Mac Cluster I talked about a G5 cluster. You know PowerPC G5? That's PowerPC 970FX

I guess maybe the question is on the distinction of the G5. Is a G5 because it runs OSX? Or is it a G5 b/c it is a PowerPC 970FX?

This is under the presumption that the 970FX wasn't necessarily designed for Apple (but for other IBM projects, servers, etc.)

 

[AidenShaw]

If you say "G5" - people will think Apple. This is not Apple, it's IBM.

G5 is not a synonym to Mac u know. I didn't talk about a Mac Cluster I talked about a G5 cluster.

Can you buy anything from IBM with "G5" in its name?

No.

It's an IBM cluster that happens to share one part with systems sold by Apple - the microprocessor chip (which itself happens to be designed and made by IBM).

If you say "G5 cluster", most people will think that it would be made with Apple hardware. This cluster is not a "G5 cluster" - it's an IBM BladeCenter JS20 cluster. No Apple hardware is involved.

And, BTW, you shouldn't really call it a "40 teraflop cluster" until it's been built and has run the benchmarks. While it will have a theoretical peak potential of 40 TF, in the real world your mileage will vary. "Up to 40TF" or similar wiggle-words would be appropriate.
_________________________

But, thanks for the link to the Spanish supercluster - it does counter my question about whether Xeon, Opteron and Itanium will continue to grow to fill most of the top cluster ratings.

At least one PPC970 cluster will be joining the other POWER-based systems in the ratings.

 

[AidenShaw]

it's marketing, not technology

Is a G5 because it runs OSX? Or is it a G5 b/c it is a PowerPC 970FX?

This is under the presumption that the 970FX wasn't necessarily designed for Apple (but for other IBM projects, servers, etc.)

It's a "G5" because that's the Apple marketing name for what Apple considers to be the 5th generation of PPC processors.

As far as I can find G1 and G2 didn't really exist at the time - but once Apple marketing coined the name "G3" for the third generation, references to G1 and G2 started to spring up for the PPC601 (G1) and PPC603 (G2). [Was there ever a Power Mac G2? No.]

And in the future, the G6 will be whatever Apple decides that it is.

http://www.cupertino.de/pages/archiv/CodeNames/Processors.html

G1 - PowerPC Processors
Remember, G1 and G2 wasn't used as a name per say[sic], until the G3 (3rd Generation) came about. They were the First and Second Generation Chips, and everyone knew that -- they just didn't call them G1 or G2.

(Lots of other PPC trivia on that page...)

 

[gloftis]

Wow, I just realized how pathetic these updates really are. Even apple thinks they're pathetic-they don't even get the main spot on the web-page, that's left to airport express. They had to make a new product announcment before the PM updates so they could hide those shameful updates behind something new, but not totally look like they knew the updates sucked, by not putting them up at all.

Don't blame all of Apple for screwed-up marketing copy. The weak link Apple's strategy is marketing.

Typically, copyrighters don't know crap about the products and the really competent communicators in the company (e.g., tech writers) who DO understand have little or no access to the messages being sent to the market. Apple exacerbates this situation by requiring computer engineering skills for tech writing positions.

Instead of trying to wow buyers with neon graphics, their message should be how buyers--especially new buyers and technophobes--could benefit from Macintosh's intuitive environment without getting the 10-year-old next door to do the setup and without spending hours in classes with titles like "Introduction to WIndows."

There are millions of people with X86 doorstops who might get excited by the prospect of not having to decypher "C: drive," Registry, and the Start menu. If Apple were addressing peoples' needs instead of trying to blow the minds of the .01% of the population who uderstand and give a rat's about cluster computing, they would be gaining market share.

You want to generate sales? There are 76 million baby boomers whose cumulative wealth is estimated at around $500 billion. No identifiable generation has every been so large with so much disposible income. A good percentage of them grew up before the PC boom. Get them interested in a Mac, and they will not only buy one, but give them to grandchildren, too, along with all the Digital Lifestyle peripherals.

Parsing processing speed will never make converts, but solid, well-written Apple marketing campaigns, aimed at the needs of people with money to spend, could make Bill Gates pause.

 

[jragosta]

Not "certain calculations", but *every* calculation. Even good memory gets random errors, so partial checking would be useless. Every calculation would need to be done twice and compared, and recalculated several times and "voted" if the results differed.

And what do you do when the error causes a panic, or if the error happens when comparing?




No, the rules shouldn't specify technology.

The rules should specify that you get the correct results, however. And obviously VATech didn't use the "do it twice and compare" logic while running the benchmark.

"Obviously" because they got over 50% efficiency, and "obviously" because the system was too unstable to run real science (although the benchmark apparently completed normally).




True, but they've realized what a big mistake they made and tore the machine down even before going into production use.




No, the down time is because the first machine failed. It was not usable, and they had to take the risk to completely replace the machine.

Had the PowerMac cluster been a success, it would still be running. VATech would have shut down a single 12-node rack, upgraded it to XServes, tested it offline, and then introduced the new rack into the cluster.

Then, as XServes arrived from Apple, one-by-one the PowerMac racks could be shutdown, upgraded, and restarted.

The cluster could have been upgraded without downtime - but the problem was that it was already "down" due to instability.


The VATech cluster is a "success" only to Mac fans, the big new clusters are being built with Linux on Opterons, Itaniums and Xeons. Even Dr. Varadarajan's company is headlining its new 20 TFLOP Itanium cluster (built in only 5 months) on its home page. (http://www.californiadigital.com)

So by your own admission, VA Tech got the right answer when running the benchmarks.

But, even if they didn't have the xServe plans underway, even if you cut the time in half, they'd still be in the top 10 - which no Windows computer ever managed.

As for the 20 TFLOP Itanium system, you're comparing Apples and oranges. They _deployed_ it in 5 months. Now, add in the time to order, receive, build, test, AND THEN deploy it and it's way more than 5 months.

As for your allegations that the Big Mac didn't work, how about a source? Computerworld gave it the top award for computing last year. Other magazines did, as well. Would they have given an award for a system that didn't work? Not likely.

Once again, we have the word of a no-name, anonymous nobody (you) against some of the most respected names in the industry. Sorry, no one believes you.

 

[jragosta]

I said it before. Steve is greedy, He charges way to much for out dated product! This revision if you could call it that is lame. Why not sell the 1.8 G5 for $1200 and add a surcharge of $800 to keep Apple afloat.

Looks like another teenager who doesn't understand how business works.

 

[kangaroo]

My thought too...

dude, this is Apple we are speculating about, NOTHING is for sure.....and on that note....What if Apple decides to make the new iMac a dual G4?? That would distingiush it as a consumer machine for sure. Just a thought.

I agree. It seems to me if a G5-iMac won't fly at this point-in-time, then a dual G4-iMac makes perfect sense. It would also fit into Apple's glacial-paced upgrade strategy. They could start with a dual 1GHz on the high end and then 6 mos. later offer a dual 1.25GHz and so on. 12 to 18 mos. from now they'd offer a G5-iMac and push the dual G4s to the lower end.

 

[Mav451]

Not to go off on a tangent, but a question to AidenShaw, who seems to be rather knowledgeable on this supercomputing stuff.

An article from X-bit Labs, quite a while ago (1/27/2004 actually): http://www.xbitlabs.com/articles/cpu/display/powerpc-g5_19.html

Points out that the G5's theoretical peak numbers were based on 32-bit precision, not 64-bit precision.

1) First of all, is that true?

2) Secondly -- Does this ALSO apply to the peak numbers for the G4's as well?

The hilarious "Who has the Power?" http://forgetcomputers.com/~jdroz/pages/09.html page uses the FLOP benchmark as the end-all comparison, making a 933Mhz G4 seem to be, in their opinion, as "powerful" as 2 x 2.0Ghz Opterons.

 

[Macster389]

Looks like another teenager who doesn't understand how business works.

I would love to have a business with margins like Apple. What is Apple's cost to build this new top of the line G5 1.8. I doubt more than $600-$800. The $1200 I refered to before should be the retail for this awsome G5 1.8. Unless of course the 5200 fx card tips the scales over $800. You sound very business savy. What do you think Apple pays for that card. Maybe $20 on a good day! Most mfg Co's like to make a 20-30 percent Margin. On the G5 1.8 and 2.0 what do you think Apple's Margin is. My guess would be 60-75 percent. OH Ya do you know what margin is. A 60% Margin is a 120% markup on cost, but you new that right. I would love to have a business with people like you as my customers. I could charge what ever I want, and you would simply except it. What a wounderfull business that woud be.