PowerMac 970 Benchmarks?

[MacRumors]

MacBidouille posts (http://www.macbidouille.com/niouzcontenu.php?date=2003-05-05#5440), what it claims to be, early benchmarks from PowerMac 970 machines.

The site has hinted that it has received such benchmarks, but has not published them until now. English translations are provided on their site. In brief, the site claims to have both early benchmarks as well as graphs depicting production machines running at single 1.4GHz, Dual 1.6GHz and Dual 1.8GHz.

The preliminary benchmarks reportedly reflect that the single 1.4GHz PPC 970 is anywhere from 87% to 254% faster than a current Dual 1.42GHz G4 machine depending on the specific test (Photoshop, Final Cut Pro, Alias|Wavefront Maya Render).

The second series of benchmarks (supposedly on production machines) shows different benchmarks (Cinema 4D-XL, Photoshop, Bryce 5) as comparison, which again shows significant (thought perhaps not as dramatic) improvements of the 1.4GHz 970 vs the Dual 1.4GHz G4s, and also graphs Dual 1.8GHz 970 machine perfromances.

Looking closely to these results we understand why Apple didn't need to wait for a 64 bits OS to launch the PPC 970. We'll take advantage of a 50% gain of performance between he up-market Pro G4 and the first PPC 970. We can imagine the difference with the top level PPC 970. It will be the best evolution ever between two Mac generations.

[mkaake]

hope it's true...

matt

[wallinbl]

Well, if they made this up, no one will ever pay attention to them again.

I do find it hard to believe that they have benchmarks of this.

[NitzerX]

Awesome! :D

[kylos]

Nice! Now if they trounce Intel's offerings I'll be thrilled.

[WannabeSQ]

I WANT ONE!!!

cool, if these are real, then the mac future seems very bright. Even if they cost more than our current G4s who cares, cause it'll be able to beat a 3ghz p4. Id still like to see some other benchmarks, like games etc.

[Megaquad]

ME WANTS!!!!

if this 1.4 ghz is so fast, then 1.8 will eat little children!!
but will this not create too giant leap between consumer,laptop vs. powermac line?

[Xerov]

Awesome! can't wait!

[jayb2000]

Thats fast!
I really hope to have a 970 based machine by the end of the year! :D

[PaisanoMan]

Those benchmarks look [suspiciously] impressive to me. If those machines are really that fast, they will definitely have a lot of customers -- I can see the supply problems already. :)

[Mr.Hey]

How interesting

[NitzerX]

Originally posted by wallinbl
Well, if they made this up, no one will ever pay attention to them again.

I do find it hard to believe that they have benchmarks of this.

No one believed they had pre-release pictures of the mirror door power macs either.

I'm usually quite skeptical, but these last few days the rumors just seem right...

[Escher]

Wow! That's nutterbutters! Imagine the new low-end 1.4Ghz SP 970 decimating the old high-end DP 1.4 like that. Everybody's gonna wanna upgrade. Maybe I'll finally abandon PowerBooks and iBooks and move to a PowerMac.

BTW: The fact that MacBidouille is providing their own English translations that they know the impact of their rumor reports. Their translations also help speed the spread of their news beyond the Francophone world.

I wonder how soon, if at all, Apple will start sending Cease and Desist letters to MacBidouille. I almost feel like calling the attorneys at Arendt Fox to ask whether they've noticed. ;)

Escher

[pgwalsh]

Originally posted by Mr.Hey
How interesting Hey! - A man of intrigue and few words. :D

[mkaake]

<edit>

after 15 minutes of waiting, i told it to submit again (After a connection error) and what do ya know? it submitted 15 minutes ago...

matt

(my original message was posted twice...)

</edit>

[cc bcc]

I don't believe these benchmarks. Cinema 4d is normaly about 1.7 to 1.8 times faster on a dual cpu system, not here. And it clealy shows that Bryce 5 benifits from dual cpu's, which is not true in real life. Sorry, these are fake. :(

[e-coli]

content edited because there's a few pricks on this board

[AntoineG]

This would be really great news if it proves to be right, I hope they don't announce them 4 month ahead of release date...

[suzerain]

(1)
The single 1.4 970 is slightly better the 3 Ghz Pentium IV in *all* tests. That means it ought to be about equal with the Pentium IV that's current at release...and that's at the LOW END. (!)

(2)
None of the software here is, apparently, optimized. All of the apps are 32-bit, and the OS was Alpha. In other words, we can expect the performance of the initial machines to even increase over time, as the software developers (Apple included) take better advantage of the 970's idiosyncrasies.

Ummm...I am pleased with these two revelations.

If these are even close to true, I will be buying one of the low-end desktops, I think, as soon as they are released, and then waiting to drop more serious cash on one of the PowerBooks whenever they come out with them.

[Mr. Anderson]

Originally posted by PaisanoMan
Those benchmarks look [suspiciously] impressive to me. If those machines are really that fast, they will definitely have a lot of customers -- I can see the supply problems already. :)

That might be one of the biggest understatements I've seen on these boards in a while.

If those numbers are true, it will usher in a new age of Mac Computing - although, even though I really need a new machine, I'd definitely wait for the duals to come out.

D

[vitrector]

There is a misquotationin the origianl post! The correct text (perhaps corrected since the post here) reads:
"Looking closely to these results we understand why Apple didn't need to wait for a 64 bits OS to launch the PPC 970. We'll take advantage of a 50% gain of performance between he up-market Pro G4 and the first PPC 970. We can imagine the difference with the top level PPC 970. It will be the best evolution ever between two Mac generations."

i.e. they say that Apple does not have to wait for the 64 bit OS to launch the 970 based computers (which makes sense since the 64 bit deal affects memory addressing, and not really performance so much)

[Trimix]

Have VISA, have AMEX, and ready to splurge !
:D

[type_r503]

Originally posted by e-coli
I know the 970 is coming VERY soon. Hold on tight. I can't say how I know, but I know. Don't bother prying for more info. I'm not going to lose my job or jeopardize any else's.

Maybe you know because it is completely obvious. Oh I hope I don't lose my job.

[Frobozz]

Originally posted by cc bcc
I don't believe these benchmarks. Cinema 4d is normaly about 1.7 to 1.8 times faster on a dual cpu system, not here. And it clealy shows that Bryce 5 benifits from dual cpu's, which is not true in real life. Sorry, these are fake. :(

I don't think your criteria is complete to pass these off as fake. Although you are correct if you were comparing today's machines, the software and hardawre combination that these benchmarks are being run may indeed see increases that older software and hardware combos do not. Keep in mind that the motherboards talk to each cpu better, and to the memory faster.

I'm not saying a disgree with you. I'm only saying that these could very well be accurate numbers.

Maybe Macbidouille is making up numbers and rumors for traffic related revenue.... but it seems weird for them to intentionally do that. I'll give them the benefit of the doubt.

[JtheLemur]

Yeah? YEAH?! Well I know that DUAL 970s are coming! So there! What ya gonna do about that?

Also, Apple's new keyboard will indeed dispense Tab when the tab key is pressed. Refills will be quite affordable because, after all who the hell drinks Tab other than ballerinas and the like?

The Robot Monkey A/V prototypes are looking great too... :D Imagine 50+ hours of MPEG-4 video in the space of only one monkey! Go ahead, imagine it! Bah, I know you won't!

[dekator]

It seems to be all about avoiding bottle necks, how great!. Yes, finally we'll have high-end Macs (he up-market in MacBidouille diction) again.
If the PPC 970 is really to have such a crushing performance, I wonder who's gonna buy single processor iMacs anymore (allegedly with a G4). After all, iMacs are used for video stuff as well. Will Apple make pricing such that iMacs will be a bargain ? Or will the Pro models just be super-expensive. I hope the latter will not come true because then Apple would bump into the same car again, so to speak, and possibly miss a wonderful opportunity to attract switchers.

[beatle888]

Originally posted by suzerain
All of the apps are 32-bit, and the OS was Alpha.

in previous macrumors threads its been discussed that current applications, even writen in 64 bit code wouldnt see performance increases with 64bit computing. this is because the applications that are out there dont demand a 64bit computing experience. in other words they dont need it. i think the performance increases will come from other facets of this chip. meaning that the 64bitness is just a plus but wont be actually used to its fullest potential.

im not sure but i BELIEVE apps like photoshop might not even really benefit from 64bit computing.

[Shaktai]

Ye gads! If these are fake, they will be forever damned, but if even close to true, they will be the rumor site of all time.

I expect the 970's in late June or early July, but this performance gain is far beyond anything I would have thought possible.

We will know soon.

[aasmund]

This whole thing smells like a perfume from Gehenna. Regards

[Tobsen]

Sorry, but Bryce does not support more than 1 processor, so the numbers can't be real :-(

[herocero]

i am typically quite the dreamer and optimist (ie still holding on buying macopener for my mac formatted iPod/dell for a nice new PB 15), but this report seems almost too good to be true.

where are they getting this info? wouldn't apple try to shut this down ASAP? i mean, let's assume this info is true . . . they have build numbers on an alpha of the next big osX release. seems like some pretty high up the ladder source. also, slightly convienent that the old PM is slower than the pentium 4, yet every incarnation of the 970 systems is faster . . . hope our mac-frenchy comrades have a day job, cause this could bite em in the butt.

for the first time in a long while, i hope i am super super wrong!
cheers

[dekator]

Originally posted by Tobsen
Sorry, but Bryce does not support more than 1 processor, so the numbers can't be real :-(
What if the design is such that it pretends to be one processor ? I'm not sure but can you exclude such a possibility ?

[Grimace]

How big is the Apple market in France anyway?? Europe??

[Nutzoids]

I am all ready to buy my new IMac this week. I was so excited...Now what do I do? Wait and save up for a 970. Or just move ahead as planned? What to do???? What to do?

[AppleMatt]

If this is true I can't wait to see Adobe crawl back with their tail between their legs.

I'm a little gutted considering that I've just gone and bought a 12" powerbook, :(

AppleMatt

[Remus]

Originally posted by dekator
What if the design is such that it pretends to be one processor ? I'm not sure but can you exclude such a possibility ?

From what I have seen, I do not belive this feature will make into this processor, but will be seen in the upcoming "980" PPC chip.

Of course this is only an opinion.:p

[cc bcc]

Originally posted by Frobozz
I don't think your criteria is complete to pass these off as fake. Although you are correct if you were comparing today's machines, the software and hardawre combination that these benchmarks are being run may indeed see increases that older software and hardware combos do not. Keep in mind that the motherboards talk to each cpu better, and to the memory faster.

I'm not saying a disgree with you. I'm only saying that these could very well be accurate numbers.

Maybe Macbidouille is making up numbers and rumors for traffic related revenue.... but it seems weird for them to intentionally do that. I'll give them the benefit of the doubt.

I can understand that you want these numbers to be right, but Bryce uses only 1 cpu.
Cinema 4D is 1.7 to 1.8 times faster on a dual cpu machine, both on the mac and the pc. And these machines are quite different in how the mobo communicates with the cpus.

Of course, they could have accidentaly switched the Cinema 4D and Bryce 5 results.. ;)

[NitzerX]

Originally posted by dekator
What if the design is such that it pretends to be one processor ? I'm not sure but can you exclude such a possibility ?

More likely (to me anyway) is that they have a pre-release patch that supports dual processors.

Most likely of all, they faked the results with the single 1.4 score and scaled to 1.8 with a fudge and halved the time for the second processor.

I still think these things are going to rock and they do seem to be coming soon!!!

:cool:

[wheezl]

In addition to the obivious bryce 5 statements already made. It seems odd to me that all of the benchmarks resulted in complete seconds. Sure if you were doing really long processes such as video encoding this makes sense. But some of the benchmarks yeilded results of only 7 seconds, and in a tie at that. You would think they would have at least gone out one decimal place on the results. Also they seem overly optomistic based on the performance of the parent Power4 processor.

I'm definetely looking forward to the 970 Macs but these numbers seem fishy.

[rog]

I think this is total BS. Why? Check out their Bryce graph. Why would there be dramatic gains in a DP 1.8 vs. a SP 1.4? Bryce has no DP or Altivec optmizations! so the benchmarks are bogus.

[illumin8]

Originally posted by beatle888
i think the performance increases will come from other facets of this chip. meaning that the 64bitness is just a plus but wont be actually used to its fullest potential.

im not sure but i BELIEVE apps like photoshop might not even really benefit from 64bit computing.
You are 100% correct. There are two main benefits to a 64-bit OS:

1. The ability to address more than 4GB of memory.
2. The ability to use 64-bit registers.

Every day it seems that I read a post by someone that thinks a 64-bit OS will magically double the speed of their computer. This is just not the case. You will usually only see a marginal (5-10%) speed increase, and usually only on applications that do a lot of heavy floating point type number crunching. 64-bit does nothing for your memory bandwidth or throughput, which is, IMO the most important benchmark you should be looking at.

The biggest problem with the G4 right now is not low clock speed, but low bus speed. The P4 is enjoying a healthy 533 mhz. (soon to be 800) FSB. G4 is stuck at 167 mhz. I also don't understand why Apple thinks sticking 333mhz. DDR memory on a 167 mhz. bus will increase performance.

Apple engineers must have been on crack when they thought they could double memory bandwidth without increasing FSB to match. It makes all those that went out and upgraded their PowerMacs just to get DDR memory look like fools.

I think they did it just for marketing reasons, because too many PC switchers were saying: "It doesn't have DDR memory like a PC so it is half as fast."

The new benchmarks are very encouraging indeed! I don't put much stock in the rumors, but I hope that the new 15" AlBook has a low-clocked (1 ghz. or 1.2 ghz) 970. I am waiting to buy one.

[madforrit]

I don't know about these numbers...hope they are true because my G4 450 would like to retire!

Hmmm, let's see, how could Apple screw this up?

1) Offer speed bumped G4's to 1.6ghz in July, intro 970 in Jan, OR have the new 970's performance be equivalent to a speed bumped G4.

2) Price the low end 970 at $4000, or even 3,000 for that matter.

3) Somehow retain that memory bottleneck that's been screwing us for so long.

4) Have 64-bit Panther be incompatible with most current OSX software.

5) Make the 970 so ugly that we all have to put paper bags over it...though I'm not going to mind a paper bag if the thing has performance numbers like MacBidouelle has posted!

Can anyone else think of how Apple could screw this up?

[illumin8]

Originally posted by Remus
From what I have seen, I do not belive this feature will make into this processor, but will be seen in the upcoming "980" PPC chip.
Correct you are. If the 980 processor is based on the Power 5 architecture, it should inherit a new technology that both IBM, Sun, and Intel have been making, multiple cores:IBM disclosed in an interview with EE Times that its next-generation Power5 chip would support two cores, each running two simultaneous threads. "We have the chip back and we are in early testing of the processor. It is performing exactly as we hoped," said Papermaster.
Read this article (http://www.eetimes.com/issue/fp/OEG20030228S0017) for more details.

[oldMac]

My vote would be for the classic...

1) Announce in June with availability in July

2) July comes and goes with delays until August

3) First machines ship in August with manufacturing delays which limit significant supply until September

4) Meanwhile, everybody stops buying G4s in May in anticipation of a June launch.

Yup... That sounds about right. :)

[gwuMACaddict]

between these new rumors and the new ipod, all my summer money is as good as spent- and i haven't even made it yet...

:D

[madforrit]

Originally posted by oldMac
My vote would be for the classic...

1) Announce in June with availability in July

2) July comes and goes with delays until August

3) First machines ship in August with manufacturing delays which limit significant supply until September

4) Meanwhile, everybody stops buying G4s in May in anticipation of a June launch.

Yup... That sounds about right. :)

I was asking for real screw-ups. This is standard! :D

[beatle888]

Originally posted by AppleMatt
If this is true I can't wait to see Adobe crawl back with their tail between their legs.

I'm a little gutted considering that I've just gone and bought a 12" powerbook, :(

AppleMatt

thats what sorta has me stumped. i thought apple provided their main developers with test systems. so wouldnt adobe already have a 970 machine if it was in the works? and if they did, wouldnt they be excited about the speed and look toward the near future? so why would they post the remark that windows based systems are preferrable for their video applications when they know that apple has a gem right around the corner?

[siberian]

I have never seen apple release a dramatically new product like a 970 PPC and not radically trim inventory early.

Look at the iPod. 2 months before it came out they started restricting availability. Selling them off, getting out of a bad situation.

If Apple is still making G4 based powermacs there is no way they are introducing an entirely new architecture based on the 970.

Call me once the G4 desktops stop rolling off the line and they start depleting inventory.

[silvergunuk]

ok i've just read the specs and i think i've soiled my pants! sign me up for a render farm. Quality nice!

[strider42]

Originally posted by illumin8
Correct you are. If the 980 processor is based on the Power 5 architecture, it should inherit a new technology that both IBM, Sun, and Intel have been making, multiple cores:
Read this article (http://www.eetimes.com/issue/fp/OEG20030228S0017) for more details.

you are aware that the power4, on which the 970 is based, has always been dual cores. Dual cores is not new technology. And its existence on a high end chip has absolutely no bearing ona deriviative chip, as evidence by the fact that the 970 is only a single core chip.

[Fukui]

Sorry, but Bryce does not support more than 1 processor, so the numbers can't be real :-(
Not unless the next OS X is using pervasive multithreading like BeOS.

[macrumors12345]

Originally posted by cc bcc
I don't believe these benchmarks. Cinema 4d is normaly about 1.7 to 1.8 times faster on a dual cpu system, not here. And it clealy shows that Bryce 5 benifits from dual cpu's, which is not true in real life. Sorry, these are fake. :(

That is exactly my take as well.

Sorry guys, but these benchmarks are FAKE. Bryce 5 does not take advantage of dual processors, so there is no way that the DP 970 1.8 is so much faster in Bryce 5 than the SP 970 1.4.

Now, I wouldn't be shocked if in general the real benchmarks look like this when the machines are actually released (they are not totally ridiculous, it's just that the person who made them clearly knew so little about the respective applications that s/he didn't even bother to figure out that the Bryce rendering engine isn't multithreaded.)

The "bad" new is thus not that these benchmarks are a completely inaccurate protrayal of the 970's potential, but rather that EVERYTHING that MacBidouille has told us (including schedules, etc.) is HIGHLY suspect. The smart money would bet heavily against MacBidouille's predictions now.

To see how little effect multiple processors have on Bryce performance, you can check out this BareFeats article http://www.barefeats.com/pm1000.html

[trilogic]

Originally posted by wallinbl
Well, if they made this up, no one will ever pay attention to them again.

I do find it hard to believe that they have benchmarks of this.

exactly how I think. why would they make this up, but how come they have access to future machines. :confused:

[zach]

Actually, my money is on these predictions if MacBidouille actually DOES somehow have acess to these machines. If they do or have data about them, it would not be unlikely to also have a Bryce patch that can take advantage of DP. If you think about it, these patches would be made for apple people so they can check these types of benchmarks.

However, going back to whole original debate, i dont really believe these numbers are real. I don't, however, think they can be passed off as fake because of these Bryce issues. I just have a feeling, however wrong it may be, that 970's would not be the types of things that apple would let rumors get through about. Think about it. This would be a HUGE revalation in processing, so don't you all think that it would be kept under heavy secrecy? I think everything that we "think" we know is totally wrong.



Uh... sorry for the long confusing rant. :-)

[fpnc]

I say slow down people. These PPC970 introduction rumors and benchmarks from MacBidouille are so far fetched that they verge on complete science fiction. Come back in about a year and they may be closer to the truth.

[post edited by moderator. Please refrain from attacking other nationalities]

And, of course, all of these rumors they are posting sure doesn't hurt their page views and they've been wrong before but people still keep coming back.

[tjwett]

[post edited by moderator. Please refrain from attacking other nationalities]

[Goekeli]

I read the whole thing. If any of that is ture, apple legal should be stepping in soon. Heads will roll. Or this is an intentional leak. Not very apple!
I'd like to think it's real but, hmmm. All those numbers sound really great but?hmm. We'll see.

:)

[macrumors12345]

Originally posted by dekator
What if the design is such that it pretends to be one processor ? I'm not sure but can you exclude such a possibility ?

Yes, you can exclude such a possibility.

The benchmarks are fakes - end of story.

By extension, most of the other stuff that MacB has been posting is probably also fake.

You guys are just way too optimistic.

But on the plus side, I wouldn't be THAT shocked if the real 970 were to perform like that when it is released (at least at the higher clock speeds).

[macrumors12345]

Originally posted by illumin8
Correct you are. If the 980 processor is based on the Power 5 architecture, it should inherit a new technology that both IBM, Sun, and Intel have been making, multiple cores:
Read this article (http://www.eetimes.com/issue/fp/OEG20030228S0017) for more details.

Argh. No. The Power 4 already has multiple cores, so a multicore Power 5 is nothing new. What the Power 5 brings to the plate (among other things) is simultaneous multithreading (SMT), which allows each CORE to appear to the OS to be two or more (in IBM's case, 4, I believe) processors.

So SMT is actually the OPPOSITE of what the guy who started this discussion was talking about. He wanted a technology that makes two processor cores appear as one single core (no such technology exists). What Power 5 (and possibly the "PPC 980", if such a beast comes into existence) will do is make one processor core appear as two cores.

But at the end of the day, Bryce 5 still wouldn't run any faster with SMT, because it still doesn't talk advantage of multiple processors!

[barkmonster]

wow, I saw one little message about the benchmarks a few hours ago, made a quick webarchive of the whole page just incase it's taken down and within minutes it made it to the main page on here!!

That's really quick, I havn't been able to post for an hour or so because the site has been so slow and I was watching a cool show about spy technology and stuff!

I don't think the results are that far off what you'd expect given the fact that on paper a PPC970 is around twice as fast as a G4 of the same clockspeed, add the huge increase in memory bandwidth (an obvious big speed increase the case of the dual 970 system and Altivec performance) and it seems quite realistic.

The bryce result seems very suspect. Even though the dual 1.8Ghz system CAN allocate an entire CPU to the task, it certainly seems over the top to expect

I don't actually think a 1.4Ghz PPC970 would be either equal to or faster than a 3Ghz P4. We need more info about the specs.

How much RAM did each system have installed ?

Is the RAM in the 970 systems dual channel DDR3200 giving 6.4Gb/s or is it some other configuration only supplying 3.2Gb/s or maybe even less ?

What kind of P4 system ?

There's the newer 800Mhz FSB (200Mhz x 4) versions now, you'd assume the tests would be against the older 533Mhz FSB (133 x 4) version unless they specifically mentioned it.

I just hope we get some seriously cool systems coming out that really do perform that much faster, we need it, apple needs it, not to mention those ads with the burning intel geeks are just plain funny!!

:D

[ewok-MB]

Originally posted by Frobozz Maybe Macbidouille is making up numbers and rumors for traffic related revenue.... but it seems weird for them to intentionally do that. I'll give them the benefit of the doubt.

Hi all, a few words from the MacBidouille's team.
- We don't make any money out of the number of visitors.
- The tests results are not invented, they were sent by a very good source.
- However if this source was 100% reliable it wouldn't be a rumor but a news.

[macrumors12345]

Originally posted by fpnc

I wonder, maybe MacBidouille (being French) is trying to get back at the Americans after the Iraqi war dispute. I can see these guys now, laughing their heads off and saying, "Boy, these Americans sure are gullible!"


Uh, okay, the MacBd rumors are clearly false, but your whole French conspiracy theory is equally ridiculous. Pot, kettle, black.

[Bengt77]

[post edited by moderator. Please refrain from attacking other nationalities]

[DharvaBinky]

Originally posted by illumin8

The biggest problem with the G4 right now is not low clock speed, but low bus speed. The P4 is enjoying a healthy 533 mhz. (soon to be 800) FSB. G4 is stuck at 167 mhz. I also don't understand why Apple thinks sticking 333mhz. DDR memory on a 167 mhz. bus will increase performance.


They didn't. But, in order to get any benefit out of a 167MHz FSB, they would have to get SDRAM that was faster than PC133. PC 166 *does* exist, but is wildly expensive and rare. The solution, then, is PC2700 (DDR333), which would remain synchronous (on every other RAM clock pulse) with the FSB on the G4. Notice that on the models that have a 133MHz FSB (such as the original XServe or the 12" Powerbook) and includes DDR, that they run PC2100 (DDR266). Again, synchronous on every other RAM clock pulse.


Apple engineers must have been on crack when they thought they could double memory bandwidth without increasing FSB to match. It makes all those that went out and upgraded their PowerMacs just to get DDR memory look like fools.


This is the other reason they put DDR in there... Marketing. (like you said) Peecees have it, so you sound like a dinosaur if you're not sportin'. But as I understand it, having your memory synchronous (or evenly divisible by) with your FSB is optimal for throughput.

I think, though, we'll see some interesting RAM configurations for the new 970 machines. Since there is no "Front Side Bus" per se (Hypertransport isn't a bus), we'll be hearing more and more about troughput speeds (gb/s) than MHz now. HT offers a max thruput of 6.4GB/s of dedicated bandwidth to each processor (since HT is point to point and not a bus, the P4 *shares* it's bus bandwidth). 6.4GB/s is a lovely high number... and just happens to be the bandwith provided by Dual channel DDR400.

hrmmm... sounds like a no-brainer to me.

;)

Dharvabinky

[AmbitiousLemon]

Originally posted by ewok-MB
Hi all, a few words from the MacBidouille's team.
- We don't make any money out of the number of visitors.
- The tests results are not invented, they were sent by a very good source.
- However if this source was 100% reliable it wouldn't be a rumor but a news.

My general impression is that MacB has some good sources. Past experiences have proved this.

This being said it is not completely out of question that these results are false (either faked by MacB or their source). SpyMac started its empire by concocting a similar fraud. Despite everyone knowing SpyMac created this fraud (iWalk) it has still grown into a very valuable site. The Mac web has put out the impression that you can lie and cheat and spit in our faces and we will love you for it. The recent entry of MacWhispers (created to funnel traffic into a fraudulent mac company) and their very poor rumor record has shown that people have learned that they can make money lying and cheating Mac enthusiasts and rumor followers.

I do not get this same feeling from MacB (they seem like good guys) but considering the other creeps out there, and the success of their schemes, i would not discount the possibility that yet another person (either MacB or their source) is lying to us for fun, fame, or profit.

[macrumors12345]

Originally posted by zap23
Actually, my money is on these predictions if MacBidouille actually DOES somehow have acess to these machines. If they do or have data about them, it would not be unlikely to also have a Bryce patch that can take advantage of DP. If you think about it, these patches would be made for apple people so they can check these types of benchmarks.


That would really makes no sense at all (for Corel to have MP optimizations for Bryce 5 that they never bothered to release). But to your credit, at least you don't believe (in the end) that the MacBd benchmarks are correct.

I GUARANTEE that these benchmarks are fakes. Why? Well, first of all, the Bryce issue that many of us have already noted.

Second, the results for the DP 1.42 G4 and the PIV 3.06 are just COPIED DIRECTLY from results on BareFeats. Just look here, http://www.barefeats.com/pentium4.html
Even if the MacBd testers managed to run the exact same test suite that BareFeats did, there would still be a testing deviation that would make the numbers not exactly the same (e.g. MacBd might get 35 sec or 32 sec for the same test that Barefeats times at 33 secs). But instead, in a feat of bizarre statistical coincidence, they get the exact same numbers for all of the tests! Plus it is infinitely improbable that the MacBD testers would choose the exact same tests as BareFeats...it's not like they were running a standardized test suite or anything.

Basically, what has clearly happened here is that whoever made this up went over to the BareFeats site, copied the results for the DP 1.42 G4 and the 3.06 P4, made up their own numbers for the 970 processors, and then sent these files off to MacBd.

If you still think that these benchmarks are real at this point, you are smoking some seriously strong stuff. Get over it - they're fakes.

[drastik]

I haven't made final judgement on these specs, but I do have a couple of comments here:

1. Regaurdless of what you think of MacB, even if you think ridiculous nonsense about them being French. Ruors do not typically make it to the front page of this site unless there is some coioberating evidence, that's why Macrumors is the best of all the rumor site.

2. It took awhile for the leaked DD powermac pictures to get slapped down, that would explain a delay from apple legal, it is international afterall.

PS. There is a tornado about 15 miles from downtown here in Nashville, and I am now offically stuck at work.

[DharvaBinky]

We've all seen some of our last G4s from motorola... wave bye bye... *grin*

Look at my pure genius. :cool:

http://forums.macrumors.com/showthread.php?s=&threadid=22170

You read it here first.

;)

Dharvabinky

[Freg3000]

5 words and 5 letters:

To Good To Be True

B-R-Y-C-E

[arn]

Originally posted by drastik
1. Regaurdless of what you think of MacB, even if you think ridiculous nonsense about them being French. Ruors do not typically make it to the front page of this site unless there is some coioberating evidence, that's why Macrumors is the best of all the rumor site.


well, rumors posted on 3rd party sites tend to get posted with less or no corroborating evidence... you have to go by the individual site's reputation.

So, I have no evidence for or against this particular piece.

arn

[Snowy_River]

I find it mildly amusing that almost everyone who posts saying that these are fake benchmarks also say that they expect the real numbers to be somewhat similar. Hedging our bets?

As to the question of Bryce 5, I don't know much about this, but I'd like to chime in with the fact that I know of a piece of professional software which, in the latest beta, supports dual processors. Heretofore, this software has not, so the existence of some kind of dual proc support in Bryce 5 that isn't available to the public isn't an instant indication of fake.

Now, about the barefeats numbers, I could easily see the following situation. A guy at Apple who has been a source for MacB sees the numbers on BareFeats, and runs similar tests on 970 based machines that he has access to, using software that he has access to (including, presumably, a new beta of Bryce that supports dual processors). He then copies the G4 and P4 results from BareFeats, tacks on his results from his impromptu 970 test, and sends them off to MacB. Sure, this wouldn't be a great, perfectly accurate test situation, but it would get a general idea across, and I'm sure that would be his intention.

Now, all of that said, I have no opinion, one way or the other, about the legitimacy of these rumors. I certainly hope they are real, as they would be really great news for the Mac community - back to the days when Macs were the fastest personal computers around. But, I have no basis from which to judge these numbers. I must simply wait and see...

[macrumors12345]

Originally posted by arn
well, rumors posted on 3rd party sites tend to get posted with less or no corroborating evidence... you have to go by the individual site's reputation.

So, I have no evidence for or against this particular piece.

arn

Arn,

Honestly, you should consider moving this rumor to Page 2 (if not dropping it altogether). As I pointed out in an earlier posting, the benchmarks for the G4 and the P4 are clearly just copied from the BareFeats website. This stuff is worse than just made up - it's plagiarized. There is really no question at this point that the 970 numbers are faked.

(see http://www.barefeats.com/pentium4.html)

Mike

[noel4r]

that's it, i'm gonna start saving for a new powermac!!! i wont expect it too soon because i dont want to be disappointed so i say they'll have it by jan. 2004. i should have a few grand by then....

[XnavxeMiyyep]

Qutoe from Goekeli
I read the whole thing. If any of that is ture, apple legal should be stepping in soon.
If Apple did that, it would be obvious that it's true. Maybe Apple is ignoring it so we won't know whether or not it's true. If it is, I'll be happy because of the 970, but upset, cuz I just bought a G4 a couple months ago.:(

[JtheLemur]

[post edited by moderator. Please refrain from attacking other nationalities]

[beelzeben]

And what's the deal with the French having all this info? International corporate espionage, anyone? What's up with THAT?! Harrumph.

Revenge for iTunes Music Store being US only? :D

[GeneR]

I guess we're just going to have to wait and see. Hmmmm.

However, if they can always make FCP faster, I'd be rather happy. :D

[Wonder Boy]

Originally posted by wallinbl
Well, if they made this up, no one will ever pay attention to them again.

Hey geniuses at Macwhispers! Sound familiar?

BTW, anyone else having touble accessing macwhispers? I keep getting prompted for a password.

[Tagada]

Originally posted by carletonmusic
How big is the Apple market in France anyway?? Europe??

Apple had up to 20% market share in France. Now anywhere between 3 to 6% like in the US. Notice that Apple is not much in the education market in Europe and much more present in SOHO and medical segment.

France is/was the number 3 market for Apple after the US and Japan
Apple Europe IS in France.
Many, many French developpers are behind past and present MacOS, Nextstep and MacOS X system and major applications (bryce, 4D, revolution,...).
Apple expo in France is the largest Apple event IN THE WORLD (anywhere from 80000 to 100000 visitors)

I guess the average situation in Europe must be quite close to the above with 1 exception : Suisse, where apple as 12% market share (also a smaller market ;-)

[Frobozz]

Originally posted by beatle888
thats what sorta has me stumped. i thought apple provided their main developers with test systems. so wouldnt adobe already have a 970 machine if it was in the works? and if they did, wouldnt they be excited about the speed and look toward the near future? so why would they post the remark that windows based systems are preferrable for their video applications when they know that apple has a gem right around the corner?

Well, that one is simple: FinalCut, Shake, FinalCut Express.... all aimed at elminating Adobe from the video production market on the Macintosh.

[Frobozz]

Apple is doing something right. Their music service is obviously most, or all, of this HUGE gain... but perhaps there's more.

$1.64 is the largest single day increase in, perhaps, over a year. Today is a good day to own Apple stock... :-P

[Tagada]

Originally posted by fpnc
...
I wonder, maybe MacBidouille (being French) is trying to get back at the Americans after the Iraqi war dispute. ...
And, of course, all of these rumors they are posting sure doesn't hurt their page views and they've been wrong before but people still keep coming back.
I am offended by your comment about the fact of MacB being French.
They run the site on their personnal money, are a very nice group of passionate people ready to help, make great things on their free time (overclocking experiences/DVD dezoning patchs,...)
This is wrong, simplistic and frankly insulting.
After all the Anti French post in the political corner (where it belong) I would have appreciated that you refrein yourself from this type of "humoristic" post... or that you created a thread in the political corner of the forum. not here!

[Tagada]

Originally posted by tjwett
isn't it a little late for April Fool's jokes? this might be the biggest, steamiest pile of horse***** i've ever smellled. how long will it take for people to realize that you can't trust the French?
I am offended by your comment about the French people
If you still think that it is appropriate to paly the French Bashing tune, than create a thread in the political corner of the forum. not here!

[macrumors12345]

Originally posted by Snowy_River
I find it mildly amusing that almost everyone who posts saying that these are fake benchmarks also say that they expect the real numbers to be somewhat similar. Hedging our bets?

Not really. OF COURSE the numbers will be fairly reasonable. If the benchmarks showed the 970 to be 10 times faster than the Pentium 4, or to be no faster than the G4, then everybody would know that they are fakes. So whoever fabricated them tried to make them look at least somewhat reasonable, which by extension means that they are going to look around where many of us might be expecting them to be.

It's like if I were going to make up a release date for the PowerMac 970. I would say something like, say, August 14, 2003, because many people on this board might believe that. I would NOT say July 2006, because nobody would believe that (although sadly many might believe it if I said May 6, 2003...some might even believe it if I said May 1, 2003). But if you knew that in fact I did not have any inside information, you would say, "I know this guy is faking the August 14, 2003 date, but nevertheless I think that they might come out sometime around late summer, because that is a pretty good guess (which is why he chose it)."



As to the question of Bryce 5, I don't know much about this, but I'd like to chime in with the fact that I know of a piece of professional software which, in the latest beta, supports dual processors. Heretofore, this software has not, so the existence of some kind of dual proc support in Bryce 5 that isn't available to the public isn't an instant indication of fake.



Okay, this is getting absurd.

[post edited by moderator. Please refrain from making personal attacks]

Look, I have been a Mac fan since the mid-80s, and I have never thought of switching to the Dark Side for a second, and I want Apple to start shipping faster hardware as much as anyone else on this board, and I do think that we will see PPC 970 machines this year. But I don't let that totally distort my view of reality such that I am in a complete state of denial.

Now, the reason that what you are suggesting wrt Bryce is completely absurd is that you are somehow assuming that Bryce CAN take advantage of dual processors on the 970 machine but CANNOT take advantage of dual processors on the DP G4!!! If this new, unreleased version of Bryce could in fact take advantage of multiple processors, then the DP G4 would be faster than the single P4, because BareFeats made it quite clear that when testing using just one 1.42 Ghz G4, Bryce was almost as fast as the 3.06 Ghz P4. So a version of Bryce that used both processors would certainly have rendered that scene as fast on the DP G4 as it did on the SP P4, if not faster. So we KNOW that the version of Bryce being used on the G4 and the P4 does not use multiple processors, because otherwise the DP G4 would see some gain relative to the SP P4 vis a vis the BareFeats results, and it doesn't.

But let me guess - now you think that maybe MacBd got ahold of a special, MP capable Pentium 4 that Intel has not yet released or even announced, just to run this test? (and that is why the DP G4 is not as fast as the P4) I mean, in your world they already have the special pre-release versions of the 970, OS X, and Bryce - why not Intel hardware too? Heck, maybe they have a faster-than-light starship and a cure for AIDS sitting around in their lab too, that they just haven't bothered to announce to the world yet.

Now, about the barefeats numbers, I could easily see the following situation. A guy at Apple who has been a source for MacB sees the numbers on BareFeats, and runs similar tests on 970 based machines that he has access to, using software that he has access to (including, presumably, a new beta of Bryce that supports dual processors).

Wow, that is SUCH a realistic scenario. Because I am sure that NOBODY at Apple would EVER have access to a DP G4 to use as an actual control for this test. It is not like the DP G4 is PRODUCED BY APPLE COMPUTER or anything like that!! Clearly they would need to go to an outside source to get these results, even though that would totally compromise the validity of the test.

I can much more easily see the following situation: Somebody copies stuff from BareFeats and makes up the 970 results, but is too dumb to even make them consistent or change them slightly so they are not an exact copy of BareFeats. Thus most of us are not fooled in the slightest, although some are so desperate to believe that MacBd has been telling the truth over the past month that they concoct bizarre and implausible stories to justify the glaring inconsistencies in the data.

Sure, this wouldn't be a great, perfectly accurate test situation, but it would get a general idea across, and I'm sure that would be his intention.

In fact, according to your story, it would be a WILDLY INACCURATE test for comparing the machines, because he would be comparing the SP version of Bryce running on a DP G4 1.42 to the DP version of Bryce running on a DP 970 1.8. Of course the latter setup is going to be much faster - unlike the former setup, it takes advantage of the dual processors! (the SP 970 1.4 vs. DP G4 1.42 would be a somewhat better comparison...if it weren't faked, which it is...but presumably the new version of Bryce would have many optimizations, not just MP awareness, so it still wouldn't be a good test) So what you are really saying is that even if the results are real, then they are mostly garbage anyway, because other important factors, like the degree of optimization in the software, is not being held constant across the machines being tested!


Now, all of that said, I have no opinion, one way or the other, about the legitimacy of these rumors.

Great. This is like saying, "I have no opinion, one way or the other, about whether the world is flat or round." One shouldn't be agnostic about things when the evidence is quite clear. And the weight of the evidence on this one is such that it is very hard to NOT unamibiguously conclude that these are fakes.

[Tagada]

Originally posted by JtheLemur
And what's the deal with the French having all this info? International corporate espionage, anyone? What's up with THAT?! Harrumph.
Maybe it's because some of the most advanced r&d in CPU for both IBM (and Motorola starting in 2005) is done in France!

[-hh]

Originally posted by Trimix
Have VISA, have AMEX, and ready to splurge !
:D

You, me, and probably a lot of other people who aren't being too badly affected by the economy.


But now let's be realistic.

Assume Apple has the product in the pipeline.
Assume too that the Benchmarks are this good.

The $64K Question: exactly when are you going to announce?


Afterall, if you want people to buy an expensive discretionary item, they have to be ready to buy, which means higher Consumer Confidence - - not to be in fear that their last paycheck will be the last.


April 29, 2003

The Conference Board's Consumer Confidence Index, which had been on the decline for the past four months, improved sharply in April. The Index now stands at 81.0 (1985=100), up from 61.4 in March. The Expectations Index rose to 84.8 from 61.4. The Present Situation Index improved to 75.3 from 61.4.
- http://www.conference-board.org/economics/consumer.cfm


Sounds promising, and the Market was up last week. But last Friday's unemployment numbers were up again, which is a negative. Also it should be little suprise that the CCI was up again in April - - afterall, it hit a 9 year low in February.

In the meantime, how does Apple's logistics look? Specifically:

... how many days worth of Inventory is Apple currently carrying on the likely obsolete models?

....and what's been tagged as EOL'ed?


If inventories are high and nothing's been EOL'ed, how likely do we think it is that Apple's going to announce, even if the product is literally already "in the pipeline" on the way to US warehouses.

Overall, I'd say that the marketplace is warming up to think about buying this kind of product, but Apple's inventory is the current 'big unknown'. I'm sure someone who knows his NDA rules thoroughly can make a general comment on Apple Inventory levels...for example, I really doubt that they're in that "<7 days" range like they used to be.


-hh

[ambush]

Originally posted by fpnc
I say slow down people. These PPC970 introduction rumors and benchmarks from MacBidouille are so far fetched that they verge on complete science fiction. Come back in about a year and they may be closer to the truth.

I wonder, maybe MacBidouille (being French) is trying to get back at the Americans after the Iraqi war dispute. I can see these guys now, laughing their heads off and saying, "Boy, these Americans sure are gullible!"

And, of course, all of these rumors they are posting sure doesn't hurt their page views and they've been wrong before but people still keep coming back.

[post edited by moderator. Please refrain from making personal attacks]

In the Apple rumor industry, there are 3 big sites. Macrumors ( basically a big site that report daily rumors ), thinksecret who's always right and macbidouille, who is also always right. (mdd pics, mdd board pix), etc.

Trust me, they would not deceive us like that.

[macrumors12345]

Originally posted by ambush

Trust me, they would not deceive us like that.

Okay....you want to bet on it?

[Freg3000]

Originally posted by Tagada
I am offended by your comment about the fact of MacB being French.
They run the site on their personnal money, are a very nice group of passionate people ready to help, make great things on their free time (overclocking experiences/DVD dezoning patchs,...)
This is wrong, simplistic and frankly insulting.
After all the Anti French post in the political corner (where it belong) I would have appreciated that you refrein yourself from this type of "humoristic" post... or that you created a thread in the political corner of the forum. not here!
Originally posted by Tagada
I am offended by your comment about the French people
If you still think that it is appropriate to paly the French Bashing tune, than create a thread in the political corner of the forum. not here!

Calm down everyone. I think that all of those comments were made in jest-they are just making a very mundane connection for a joke. I don't think that they were made in anger. However, I think it is important that people understand many Americans do feel a great deal of animosity towards the French; whether it be justified or not.

Anyway, I think we have pretty much dismissed these "benchmarks" in part because of the Bryce thing and also because of the duplication of the barefeets numbers. (And not not because MacB is French! :))

[thibsweb]

Hello all,

I'am french but I found interesting to notice that when this rumor began to spread widely (I mean, US Mac Rumor websites, Slashdot...) only a few minutes after the AAPL share went up 10% with an impressive amount of shares on the market (28M)

You can see a graph here :

http://www.boursorama.com/cours.phtml?symbole=AAPL

if this rumor is wrong, it still has made it to the NASDAQ.

Wtach for APPL in the near future ^_^

[jelloshotsrule]

Originally posted by ambush
I bet you're american?

In the Apple rumor industry, there are 3 big sites. Macrumors ( basically a big site that report daily rumors ), thinksecret who's always right and macbidouille, who is also always right. (mdd pics, mdd board pix), etc.

Trust me, they would not deceive us like that.

why do you sink to his level? as an american, let me say that most people here aren't as dumb as to be all about hating france.. though what's up with "idiot cheese"?? and yes, this all belongs in political forum or not on the site at all


as for thinksecret and macbidouille *always* being right.... i find that hard to believe. i'm not going to go back and find specific rumors they predicted that didn't pan out, but let's be honest... no one's "always" right.

[arn]

Originally posted by thibsweb
Hello all,

I'am french but I found interesting to notice that when this rumor began to spread widely (I mean, US Mac Rumor websites, Slashdot...) only a few minutes after the AAPL share went up 10% with an impressive amount of shares on the market (28M)

You can see a graph here :

http://www.boursorama.com/cours.phtml?symbole=AAPL

if this rumor is wrong, it still has made it to the NASDAQ.
\

I think this is more likely the reason:

http://www.apple.com/pr/library/2003/may/05musicstore.html

but who can tell

arn

[ktlx]

Originally posted by Freg3000
Anyway, I think we have pretty much dismissed these "benchmarks" in part because of the Bryce thing and also because of the duplication of the barefeets numbers. (And not not because MacB is French! :))

Although everyone seems focused on the Bryce numbers, the Photoshop numbers are suspect also. The numbers are the same as BareFeats' "SP Actions" benchmarks. These are actions that are chosen because they are not multiprocessor aware.

However, according to the MacB results, the "Actions" benchmarks are significantly helped by a second processor. That doesn't square.

But I do have to hand it to MacB on this one. They are putting the reputation on the line with specific rumors.

[j33pd0g]

I hope this rumor is true 'cause I would love to have a dual 1.25 gig g4 at an even lower price. That's enough power for me.

[deepkid]

Originally posted by Frobozz
Apple is doing something right. Their music service is obviously most, or all, of this HUGE gain... but perhaps there's more.

$1.64 is the largest single day increase in, perhaps, over a year. Today is a good day to own Apple stock... :-P

:D :D

I told you guys last week to buy AAPL, when it was around $13.

I hope you listened. :p

[Rai]

I bought 5,000 apple stock, a few weeks ago, when the rumors, about the music service . So i watch the stock price like a Hawk.

In the morning the news about 110,000 orders for ipods where in, and the info about selling 1mil songs in the first week.

At 2.45 i called my office and found out, the fedex had dropped of my new 30gig, drove to office and came home around 3.30.

At 2.45 when i left the stock price was up around 40 cents

At 3.30 when i returned the stock was up 2 dollars.

Around 3:15 something caused this stock to sky rocket, what is was i'm not sure, no more finicial news of apple was released to finicial sites, and the info about ipods and music service where known since the beginning of the day.

What time did the story get posted ?

But it could have been just market flucations

[iSmell]

OK, I have to post on a few things here:

1. Apple Legal wouldn't go after this even if it were true. It's just benchmark numbers. That's not trademarks, intellectual property or corporate secrets. The ONLY time Apple Legal's actions should be used to judge the validity of a rumor is when they actually swoop down and pull something. Then you can guess that it was real.

2. No rumor site is ALWAYS right. Thinksecret made a bunch of 100% predictions just before last macworld and a bunch of them were wrong. MacRumors has been on a hot streak lately, especially since the last minute Powerbook call that nobody else had, but they've been wrong before too. MacB has a pretty good record, indicating that they probably have some genuine inside sources, but that doesn't mean somebody can't give them false information.

3. It seems to me that MacWhispers had a pretty good call on the iPod USB/FW cable. They didn't know what it was, but they knew it existed. That means they (he) was talking to somebody that knew something. Jack Cambell may suck at life, but I'm not going to take macwhispers off my bookmark page until he guesses wrong a few more times. And conversely, it will move up on my list if we see 30" lcds, aluminum fronts on the next powermacs, or the 7" tablet thing. Even if he's making stuff up, at least it's fairly interesting stuff, and he hasn't suckered me into buying anything from his shill stores, so what do I care?

4. I can't remember the fourth thing.














Stonermonkey

[deepkid]

Originally posted by Rai
I bought 5,000 apple stock, a few weeks ago, when the rumors, about the music service .

That's quite a substantial investment. Do you intend to hold it for a year or more? Or live with capital gains penalties?

[BaghdadBob]

Originally posted by iSmell
2. No rumor site is ALWAYS right. Thinksecret made a bunch of 100% predictions just before last macworld and a bunch of them were wrong.
Really? I read ThinkSecret's 100% predictions just before the expo and as I recall every one of them was on the money. I don't know if they're always right, but I personally don't remember the last time they were wrong. On the other hand, they only seem to post rumors that they consider to be fact as far as I can tell, so they're unfair to compare to other sites.

[Freg3000]

Originally posted by BaghdadBob
Really? I read ThinkSecret's 100% predictions just before the expo and as I recall every one of them was on the money. I don't know if they're always right, but I personally don't remember the last time they were wrong. On the other hand, they only seem to post rumors that they consider to be fact as far as I can tell, so they're unfair to compare to other sites.

I think all of their 100% predictions were correct, but their 0% predictions (essentially 100% no predictions) were wrong. Most notably the 0% chance of new PowerBooks.

[chetwilliams]

Originally posted by Rai
I bought 5,000 apple stock, a few weeks ago, when the rumors, about the music service

I bought 40 call options (equivalent to 4000 shares) with a $15 strike price a week and half ago. I have already more than doubled my money. Not a bad return. Now I have to decide whether to cash out or hold on for the long haul. ;)

[macrumors12345]

Originally posted by Rai


At 3.30 when i returned the stock was up 2 dollars.

Around 3:15 something caused this stock to sky rocket, what is was i'm not sure, no more finicial news of apple was released to finicial sites, and the info about ipods and music service where known since the beginning of the day.

What time did the story get posted ?

But it could have been just market flucations

Well, according to the times at MacRumors, it got posted at the same time as the iTunes story (at little after noon...I assume Eastern Time?).

But I doubt that most investors read MacRumors.

What might be more significant is that the 970 story got posted at 2:50 or so on Slashdot? That is the "biggest" site I have seen it posted on.

Since the 1 million tracks sold PR came out at 8:30 am, it really shouldn't be a response to that. Plus they already knew there would be big sales based on the 275k songs sold in the first 18 hours. So it wasn't that much in the way of new information.

It therefore seems at least conceivable that the stock runup was a response to the 970 rumors. This would be very disturbing, as it would suggest that these results were not just faked as a prank (which is what I had been assuming the motivation was), but rather there were part of a plot to game the market. I wonder if anyone bought a bunch of AAPL call options....

Does anyone recall whether there were similar movements in the stock price back in 2001 with the Moto G5 rumors? I have no recollection at all.

[freemidnight]

Hey it would be time for MacRumors hire a translater:D

[mac15]

wow, well these were tested with only 32bit apps, it apple made someway for developers to easily bridge 32 to 64bit. It would wipe the floor with the pentium (in which it already does)

Well apple seemlessly moved from 68k to PPC, I"m sure this move will be the same, difficult but indeed not hard at the same time.

Wouldn't the apps need to be compiled at 64bit? then thats it

[dguisinger]

Originally posted by mac15
wow, well these were tested with only 32bit apps, it apple made someway for developers to easily bridge 32 to 64bit. It would wipe the floor with the pentium (in which it already does)

Well apple seemlessly moved from 68k to PPC, I"m sure this move will be the same, difficult but indeed not hard at the same time.

Wouldn't the apps need to be compiled at 64bit? then thats it

They would only need to be recompiled if they wanted to be 64-bit.....which 99% of applications would not need to be, so its pretty pointless to recompile for the 970.

[twil13]

I WANT a 970 PB this summer, so I hope that MB is partially wrong, but mostly right :) They said that no 970 in PB's till march, which is too long for me to wait :( On the other hand, I hope we are getting the 970's this summer, even if it is only in powermacs. Anyways, here's to hoping for 970 PB with Mobilitiy radeaon 9600 :)

Tony

[Snowy_River]

Originally posted by macrumors12345

Okay, this is getting absurd. No offense, but what is wrong with you people?!? Are you so drunk on the Kool Aid and so desperate for hope of a faster Mac that you will believe anything that anyone tells you, no matter how absurd? ...

You know, I find it hard to swallow that you say 'no offense' but then proceed to make personal attacks.

Now, the reason that what you are suggesting wrt Bryce is completely absurd...

Wow, that is SUCH a realistic scenario. Because I am sure that NOBODY at Apple would EVER have access to a DP G4 to use as an actual control for this test. It is not like the DP G4 is PRODUCED BY APPLE COMPUTER or anything like that!! Clearly they would need to go to an outside source to get these results, even though that would totally compromise the validity of the test.

I can much more easily see the following situation: Somebody copies stuff from BareFeats and makes up the 970 results, but is too dumb to even make them consistent or change them slightly so they are not an exact copy of BareFeats. ...

I was simply trying to present a plausible scenario for this kind of information coming forward. The fact that Bryce didn't show the benefits of the DP G4 was accounted for by the fact that these figures were taken from the BareFeats article. Now, let's see, if I was an informant for a rumor site, would I want to make a big production about getting together a bunch of benchmark tests, or would I prefer to do something quick and dirty that would get the point across? I'd probably go with quick and dirty for any number of reasons, not least of which being that it would be much harder to trace it back to me.


...In fact, according to your story, it would be a WILDLY INACCURATE test for comparing the machines, because he would be comparing the SP version of Bryce running on a DP G4 1.42 to the DP version of Bryce running on a DP 970 1.8....

Yes, this method of comparison (particularly the Bryce comparison) would have some inherent inaccuracies. But that wouldn't invalidate the overall point of the bench marks. Nor would it invalidate the comparison between the SP 1.4GHz 970 and the DP G4. The DP 1.8GHz 970 would only provide true comparison to the SP 1.4GHz 970, not properly to the DP G4.


...
Great. This is like saying, "I have no opinion, one way or the other, about whether the world is flat or round." One shouldn't be agnostic about things when the evidence is quite clear. And the weight of the evidence on this one is such that it is very hard to NOT unamibiguously conclude that these are fakes.


No, it's not. I'm saying that I DON'T KNOW, and no matter how much you wine, snivel and puel about it, neither do you (unless you actually work at Apple, and/or are the person who provided these benchmarks to MacB). I can see a plausible situation that would lead to these numbers being presented. I can also see that it is quite plausible that they have been faked. I don't know, so I won't pass judgement. I'll wait and see.

So, please get off your high horse, and stop insulting me.

[dekator]

Originally posted by siberian
I have never seen apple release a dramatically new product like a 970 PPC and not radically trim inventory early.

1) Mac staff is getting a 30% rebate on current G4 machines AFAIK.
All those machines are on short supply anywhere else in the world.
And anyway, we're likely to see a dual (or triple) approach with the G4 remaining in the low end line and the portables for some time.
2) As for the numbers: It might be that these are rounded, internal figures that are supposed to give a rough idea of the performance gains.
3)The European market accounts for about 35-40% of Apple's share.
4) The reason why this is published on a French site may be exactly *because* it's far away from Cupertino. Supervision is probably not as tight in Europe as it is in the States. I could well imagine some of Apple's top brass in Europe enjoying a little leak here and there.

[dguisinger]

Originally posted by dekator
1) Mac staff is getting a 30% rebate on current G4 machines AFAIK.
All those machines are on short supply anywhere else in the world.
And anyway, we're likely to see a dual (or triple) approach with the G4 remaining in the low end line and the portables for some time.
2) As for the numbers: It might be that these are rounded, internal figures that are supposed to give a rough idea of the performance gains.
3)The European market accounts for about 35-40% of Apple's share.
4) The reason why this is published on a French site may be exactly *because* it's far away from Cupertino. Supervision is probably not as tight in Europe as it is in the States. I could well imagine some of Apple's top brass in Europe enjoying a little leak here and there.


Agreed. Plus throwing one or more false numbers in for good measure would make it look like a guess, therefore keeping Apple legal at bay.

[pgwalsh]

Originally posted by dekator
Apple's top brass in Europe enjoying a little leak here and there. I dunno, you'd think they'd wear some Depends.

I'm sorry, I couldn't resist.



------
I imagine that the 970 will have similar performance to the Opteron, with the addition of whatever buss, ram and other technologies Apple implements.

[AmbitiousLemon]

Any posts that include personal attacks or attacks on any nationality will be edited or deleted. Please respect one another. If you can not refrain from posting these sorts of comments you will be banned.

[BaghdadBob]

On the 970s, I think we can all agree:

http://www.sylloge.com/powerads/sluggo.jpg

YEAH! (Sorry to anyone who's seen this a billion times. And to the moderators if they snuff it)

[steve53e]

Originally posted by chetwilliams
I bought 40 call options (equivalent to 4000 shares) with a $15 strike price a week and half ago. I have already more than doubled my money. Not a bad return. Now I have to decide whether to cash out or hold on for the long haul. ;)

Whether you hold or not is always a personal call. In any case, I'd set a protective stop and not give back too much should this market want to retrace. You can always get in again later when the market pulls back.

I bought on the rumors a few weeks ago as well. I bought May 12.5 Calls at .90 and Jan 12.5 LEAPS at 2.15 on the 16th or 17th of April. Both positions are sitting quite nice. I'll dump the May calls on the slightest pullback and take the profits. The LEAPS I'm going to keep in the bin for the long haul.

Best of luck in your trading!

[Booga]

Originally posted by mac15
wow, well these were tested with only 32bit apps, it apple made someway for developers to easily bridge 32 to 64bit. It would wipe the floor with the pentium (in which it already does)

Well apple seemlessly moved from 68k to PPC, I"m sure this move will be the same, difficult but indeed not hard at the same time.

Wouldn't the apps need to be compiled at 64bit? then thats it

Please don't perpetuate this myth. Making something 64-bit does not make it faster-- usually the opposite. It CAN be faster if you're doing certain operations, such as large bitfield operations that can't be vectorized, or memory copies if you're not already using the FPU, or if you need the extra bits and had been doing hacks previously... but in general it slows things down. Why? Twice the data to pump around, most of which is going to be unused in an app that is designed for 32 bits.

64 bits increases computational accuracy and/or range, not speed.

[dguisinger]

Originally posted by Booga
Please don't perpetuate this myth. Making something 64-bit does not make it faster-- usually the opposite. It CAN be faster if you're doing certain operations, such as large bitfield operations that can't be vectorized, or memory copies if you're not already using the FPU, or if you need the extra bits and had been doing hacks previously... but in general it slows things down. Why? Twice the data to pump around, most of which is going to be unused in an app that is designed for 32 bits.

64 bits increases computational accuracy and/or range, not speed.

Right. For example, a 64-bit add using 32-bit instructions takes about 5-6 instructions to carry out.....a 64-bit divide is much worse....so 64-bit only improves speed when working with 64-bit numbers.....for example, in my situation, on my xserves running MySql...I use 64-bit keys for indexes in the database. My speed would greatly improve with the 970. Most applications would not.

[BaghdadBob]

Boy! It sure is good to get back to some technical discussions. Although being able to make a joke in good taste would be nice, but oh well...

So, Mr. 64 bits smartypants (Booga), would an application like Photoshop or Bryce benefit largeley from moving data in 64-bit chunks?

I mean, you say if a program was designed for 32 bits it wouldn't benefit, but what if it was totally rebuilt to be 64 bits? Would only really complex apps benefit from this?

BTW, I havn't gone really processor-deep in my techno since I read "Birth of a New Machine", so riddle me this -- does this mean 64 bit instructions are to be used?

[dguisinger]

Originally posted by BaghdadBob
Boy! It sure is good to get back to some technical discussions. Although being able to make a joke in good taste would be nice, but oh well...

So, Mr. 64 bits smartypants (Booga), would an application like Photoshop or Bryce benefit largeley from moving data in 64-bit chunks?

I mean, you say if a program was designed for 32 bits it wouldn't benefit, but what if it was totally rebuilt to be 64 bits? Would only really complex apps benefit from this?

BTW, I havn't gone really processor-deep in my techno since I read "Birth of a New Machine", so riddle me this -- does this mean 64 bit instructions are to be used?

Depends.....most people say 32-bit color is enough, so in that sense no....you cant use a 64-bit add to add two pairs of 32-bit numbers, the carry bits would mess up the results. However, NVidia and ATI have been making the move to 64-bit color in their GPUs, and I beleive rendering engines for things like Maya, are 64-bit, however they utilize vector engines and FPU for decimal precision, so they already have that power.

Most applications have no need for 64-bit numbers. Memory access is only needed in rendering and database applications. And if the OS is 64-bit....and the applications are 32 bit, each app can access 4GB of memory....since the OS provides protected memory regions, the app gets a dedicated 4GB memory map. Thats where the largest benifit wil be seen! Most people dont even realize this.

[fpnc]

As for my post suggesting that:

MacBidouille ... was probably laughing their heads off and saying, "Boy, these ... sure are gullible!"

Yes, I made that comment and I truly apologize if it offended anyone. However, it was only a JOKE! In fact I made it somewhat ridiculous (but, I thought, completely inoffensive) only to try to emphasize how over-the-top I believe these rumors are concerning the impending release of a PPC970-based Mac. In any case, let's all try to remain polite and keep a little sense of humor concerning these posts, after all, given the current unknowns that's about all these rumors are worth. So, I urge everyone to calm down a little, and no need to comment further on this issue.

[beefstu01]

One more thing... A rumor this far reaching can either be very very good for Apple, or very very bad.

Let me explain.

With all these tech sites reporting this rumor (Macrumors, OS News, Ars, Slashdot, etc..), it's been quite far reaching. Needless to say, by now a bunch of people have heard about it, mainly the guys who will need such power later down the road. So now, even though many discount the rumor, there may be an expectation of the chip to perform a certain way. If it proves to be less than what was "leaked," people may thumb their noses at Apple and buy the next Pentium. If the rumors prove true, well, then a lot of people may then switch over to the Mac side of things. After all, if the machine is decently priced, and a bunch more powerful than a PC, why not go with it?

[jfw]

I worked on the KSR 64-bit supercomputers circa 1992. Simply converting most applications from 32 to 64 bits generally hurts performance, depending on the architecture of the hardware. If 32-bit reads and writes are reasonably efficient, then you can adopt a C programming model where "int" is 32 bits and "long" and pointers are 64 bits and not suffer much of a penalty; on the other hand, that usually has a hidden penalty in the hardware which handles halfword memory writes, which slows down full 64 bit mode. That might be a reasonable tradeoff for a consumer machine, rather than for a dedicated supercomputer; I don't know which way the 970 goes on that. The "I32LP64" model tends to upset a lot of poorly-written C code, due to inappropriate assumptions about the sizes of data objects.

There are some very specific application domains which can greatly benefit from a 64-bit integer size. Some cryptographic algorithms benefit from being able to do logic operations in 64-bit chunks (of course, many algorithms are designed not to need 64-bit arithmetic, and wind up not benefitting if it's available). It also turns out that the TCP/IP checksum benefits from wide word widths, even though it's technically defined as a 16-bit ones-complement sum. And it turns out there's a lot of perfectly ordinary code (filesystems, network protocols) which has been written assuming that 64-bit integers exist in order to simplify writing the code, at the cost of complicating the compiler's job; throw a 64x64 multiply in an inner loop on a 32-bit machine and you can seriously slow an application down -- in a way that real 64-bit integers can positively affect.

The leap from 16 bits to 32 bits was an important one, because there is a huge array of problems interesting to typical computer users which need data sets bigger than 65536 bytes. (*) There are scarcely any applications with an urgent need to address more than 2 gigabytes of RAM, and relatively few which could even sensibly use even that much memory. However, no doubt new applications will be found once developers can count on even cheap systems having 8GB or so (and of course we can always count on programmers' laziness to bloat just about any software ;) ).

(*) I can remember the flame wars between Intel supporters and Motorola supporters back in the mid 80s, where very carefully chosen 16-bit 8086 programs would run faster than equivalent "32-bit" 68000 (**) programs -- as long as you kept the data set small enough that it would run at all, of course.

(**) And of course, there were the endless discussions about whether the 68000 was "really" 16 bits, 24 bits, or 32 bits. Or 16/32 bits. Fortunately, the mighty 68020 made all those arguments moot.

[mathiasr]

Originally posted by jfw
I worked on the KSR 64-bit supercomputers circa 1992. [snip]

There are some very specific application domains which can greatly benefit from a 64-bit integer size. Some cryptographic algorithms benefit from being able to do logic operations in 64-bit chunks (of course, many algorithms are designed not to need 64-bit arithmetic, and wind up not benefitting if it's available). It also turns out that the TCP/IP checksum benefits from wide word widths, even though it's technically defined as a 16-bit ones-complement sum.
You know were are in 2003 and a lot has changed since 1992, even a Pentium IV or a G4 can handle 128 bits data chunks in their SIMD units (SSE2 or AltiVec). eg a G4 can crunch RC5-72 keys 3 times faster than a G3 at the same clock speed:
http://n0cgi.distributed.net/speed/query.php?cputype=all&arch=2&contest=rc572

Some speed gains previously expected from 64 bits computing are available on 32 bits CPUs since a few years due to the introduction of SIMD units and their larger (64 or 128 bits) registers.

Nevertheless pure 64 bits integer arithmetic will definitely benefit from 64 bits CPU.

[herr_neumann]

Originally posted by jelloshotsrule
.. though what's up with "idiot cheese"??t.

SImple, if you think that it is actually cheese you are an idiot. It is a fitting name to me. I have never liked the stuff, not even when I was little. Oh, I guess I should say that I am an American before my countrymen attack me.

[mim]

Originally posted by jfw
There are scarcely any applications with an urgent need to address more than 2 gigabytes of RAM, and relatively few which could even sensibly use even that much memory. However, no doubt new applications will be found once developers can count on even cheap systems having 8GB or so (and of course we can always count on programmers' laziness to bloat just about any software ;) ).


You will find that there are very many uses and needs for large amounts of memory (ie, more than 2-4 Gig).

The probelm is not that any one particular task requires that much memory (although I'm in architecture - and it would help us in 3d tasks greatly) - but by the time you let the system take as much as it needs, have some music playing, have internet and mail running, and then have 2 or 3 applications that you are using concurrently open....well, you aren't left with that much.

Again, as an architect using 3d I often hit 100% cpu and memory usage. Sure, you might say I could use a pen and paper to do what I'm doing - but it would so close to impossible as to be pointless.

My point - it should be my imagination that limits what is possible, not my computer.

BTW, what would the effect of 64bit processes have on speech recongition and AI? There are two pretty fundamental functions that are bound to benifit from 64bit (or more) computation.

a.

[edit - fixed grammer, damnit]

[Abstract]

Yes, but did you know that a single 970 at 1.4GHz is faster than a speeding bullet? Please............please pass this news on to other rumour sites and have them post it. If you/they don't believe me, I can easily throw together a few Excel graphs. ;)

[rog]

In the Apple rumor industry, there are 3 big sites. Macrumors ( basically a big site that report daily rumors ), thinksecret who's always right and macbidouille, who is also always right. (mdd pics, mdd board pix), etc.

Trust me, they would not deceive us like that. [/B]

Except they are by no means "always right" and they, nor any site that I can recall has been right about benchmarks on an unreleased chip months ahead of its introduction. What you mention are things that came out not very long before the MDDs were released, and everyone knew an Xserve like powermac was due any day.

These are fake, fake, fake. Total waste of time.

[rog]

Originally posted by mac15


Well apple seemlessly moved from 68k to PPC, I"m sure this move will be the same, difficult but indeed not hard at the same time.


It wasn't seamless at all. I remember that the first PPC machines ran 68k code at speed slower than the 68040 for over a year until newer machines came out. The move to PPC took well over a year before there were a lot of PPC programs, just like X took about 18 months for it to become mainstream and out of beta (despite the 10.0 and 10.1 releases).

Anyway, I actually think the move to 64 bit and PPC 970 will be seamless because there won't be any speed penalty for 32 bit apps in moving to the updated OS and chip.

[jelloshotsrule]

Originally posted by herr_neumann
SImple, if you think that it is actually cheese you are an idiot. It is a fitting name to me. I have never liked the stuff, not even when I was little. Oh, I guess I should say that I am an American before my countrymen attack me.

the funny thing is... i totally agree. what struck me is that in places that actually have cheese (ie, france), why would they even know about american "cheese"??? ha.

by all means, american cheese isn't real cheese. i like the taste, but i can accept the facts. heh


all this 32 vs 64 bit stuff is driving me nuts.

[BaghdadBob]

So...does Wisconsin have cheese?

[mccoma]

Not that I actually believe these benchmarks, but wouldn't Bryce get an incidental speed boost on dual processors from (if nothing else) the OS running threads on the other processor while Bryce is running. Bryce might not use dual processors but OS X does.

[Dave Marsh]

I've heard several people say that 64-bit apps would run a bit slower on the 970 than comparable 32-bit apps, which I find a bit confusing. If the processor was built to run 64-bit apps, and isn't just a juried-up 32-bit design finagled to run 64-bit apps (which we know ISN'T the case, since it's based on the POWER4 design), I don't understand why this would be the case.

What I mean is, hasn't the processor been designed to handle PURE 64-bit apps, with all the appropriate circuitry to accommodate that? And, when running 32-bit apps, doesn't that just mean that some of that additional circuitry just isn't being used (or perhaps is just being filled with leading zeroes)? If that's the case, and the processor is just moving data according to its clock rate, why would an app using only part of the design run faster? I'm envisioning a truck moving down the road with half a load vs a full load...60 MPH is 60 MPH. The processor might run hotter because more of it is in use, but slower? How so?

What am I missing in this analogy?:confused:

[LordNadroj]

I just ordered a PowerMac G4 Dual 1.42 Ghz with EVERYTHING including two 17 inch monitors. Should I cancel my order? Please respond with what you would do.

[Mudbug]

If it were my money, I wouldn't have ordered the two 17" monitors, and instead gone with the 20" - for what little amount of desktop space you'd gain with the dual 17's, you'd make up for in not having the break between the two in the 20.

My 2¢.

As for on topic - I just find it very hard to believe that these numbers are real and true. What I mean to say is... why on earth would benchmark test results be leaked to MacBidouille (or MacRumors or anyone else for that matter) long before anyone knows of any true existence of such a product. This has nothing to do with the country where the website is located, or the validity of past rumors. This has to do with the fact that this would be a HUGE infraction of the non-disclosure agreement between whoever's got the info and Apple.

I really enjoy reading about what you guys and girls think may be coming down the pike, but not at the cost of the legal system in any country. I just can't imagine that anyone would be willing to go out on a limb this far just to give some rumor mongers like us something to chit-chat about for a month or two (or 6 :)).

That's my second 2¢.

Last but not least - we're above name calling here. If you want to pick on someone for their ethnicity or their country of origin/present location, then pick on yourself. And do it quietly. Somewhere else. I hear slashdot is nice this time of year.

I'm now out of 2¢.
Continue on.

[nichrome]

LordNadroj, I would point and laugh at MacBidouille for using BareFeats's benchmarks without changing them to correspond to the claims they make in their article. I would then proceed to enjoy the new 1.42 GHz.

:)

Let's face it, the alleged 87-254 % speed increases with one less processor at the same clock speed (and with similar SIMD units and the same apps, even) is completely unrealistic. And as noted before, the likelihood of the benchmarks exactly matching those of BareFeats is nonexistent.

Besides, even with such speed increases, the effect it would have on your everyday computing would be quite small, assuming you don't always have your comp run mile-long Photoshop Actions...

[eric67]

Originally posted by Nutzoids
I am all ready to buy my new IMac this week. I was so excited...Now what do I do? Wait and save up for a 970. Or just move ahead as planned? What to do???? What to do?

You do like everyboy, you wait and see.....


the fact that Apple is offering large discount (around 40%, but more for some configuration) on the current dual G4 PM is probably a good info that a new PM is coming really soon.


For those who are thinking that Macbidouille might have some financial interest in publishing such news based on traffic rate, they are wrong, Macbidouille has been and still is mostly a hardware related discussion forum (upgrading, overclocking,...) and news are just there to keep the french-speaking community up to date with our favorite machine.....
I am also surprised by those date, I think they might be a bit overestimated, but you should also know that Macbidouille has already received letters from Apple lawers regarding some previous rumors which turned out to be true.
I think it is clear that the 970 is coming, it is also clear that it has to be better than the current G4 otherwise Apple would not have move from Motorolla to IBM, and the current PC-version of MacOSX could have just been an escape project in case Apple could find a better processor to replace the G4, then Apple could have used AMD processor.
In addition, one should not forget that not only the 970 is new in this future PM, but the DDR-RAM is finally going to be used as it should be, maybe even we will have dual channel DDR-RAM. the 970 support multiplier factor for its FSB compare to the frequency of the board, so it could also have some important impact on the performance.
Regarding the statment that Bryce does not support dual processor; it is true on the current version... nobody knows what version they have been testing. If Apple took the decision to go 64-bit, targeting the animation/video/image edition with new high end platform, do you really think that they did not talk to software developper????

let's wait and see, if it is true it will be just GREAT.

[NicoMan]

Originally posted by LordNadroj
I just ordered a PowerMac G4 Dual 1.42 Ghz with EVERYTHING including two 17 inch monitors. Should I cancel my order? Please respond with what you would do.

I don't think you should cancel it. The only thing I am not sure about is the monitors. Why on earth would you want 2 Apple 17in?... Drop those and get yourself one Formac 2010. As for the comp itself, I dunno if you think the extra (hypothetical at this point) speed is vital for you and you CAN wait (for at least a couple of months) then maybe it's worth waiting. But let me tell you, the Powermacs are already VERY fast, forget what you have read on these forums, and for 98% of the apps out there they are more than enough.
I am always of the opinion that you should get the machine that is available now (unless the new one is coming out in 2 weeks time, and that's not the case).

NicoMan

[nichrome]

Originally posted by Mudbug
[B]If it were my money, I wouldn't have ordered the two 17" monitors, and instead gone with the 20" - for what little amount of desktop space you'd gain with the dual 17's, you'd make up for in not having the break between the two in the 20.
Simply not true. Two 17" monitors mean you can move pallettes, download managers etc. to a secondary monitor and not have them get buried under other windows. A single 20" doesn't let you do that. Additionally, if you do graphics, you can have one screen have a window with the image zoomed in and the other at 100 %. On a 20" screen you'd have to stack the windows and switch between the two. Very inconvenient.

[Mudbug]

Originally posted by Dave Marsh
What am I missing in this analogy?:confused:

I think you're missing the weight of the trucks. Half the load at 60 mph takes less gas to get moving, while 60 mph with the full load takes more gas. Therefore more heat.

Better yet, ask an engineer.
I've got a degree in psych, and I'm a creative director in an ad agency. What do I know? :)

[Mudbug]

I've got to agree with NicoMan - drop the 17's and get TWO Formac 2010's - I'll uphold my statement about the 20 over the two 17's, but two 20's for almost the same price as the two 17's makes even more sense.

ALTHOUGH - make sure your desk is wide enough to hold that - two of those bad boys side by side takes up some lateral room, if not depth on the desk.

[MacBandit]

Originally posted by Dave Marsh
I've heard several people say that 64-bit apps would run a bit slower on the 970 than comparable 32-bit apps, which I find a bit confusing. If the processor was built to run 64-bit apps, and isn't just a juried-up 32-bit design finagled to run 64-bit apps (which we know ISN'T the case, since it's based on the POWER4 design), I don't understand why this would be the case.

What I mean is, hasn't the processor been designed to handle PURE 64-bit apps, with all the appropriate circuitry to accommodate that? And, when running 32-bit apps, doesn't that just mean that some of that additional circuitry just isn't being used (or perhaps is just being filled with leading zeroes)? If that's the case, and the processor is just moving data according to its clock rate, why would an app using only part of the design run faster? I'm envisioning a truck moving down the road with half a load vs a full load...60 MPH is 60 MPH. The processor might run hotter because more of it is in use, but slower? How so?

What am I missing in this analogy?:confused:

I too don't understand why someone would say that a 64bit processor can run 32 bit integers faster then 64bit. This is simply untrue.

A car analogy won't really work here but the fact is that even though the PPC970 has a really good on chip 32bit compatibility it will still run apps written for 64bit tasks faster. This is because it can not take to 32bit integers and run them simultaneously as if it were one 64bit integer so if the app is written in 64bit it in affect can nearly double the processing speed of the task. The problem is that not all tasks can take advantage of 64bit processing and therefore show a speed advantage. The real speed advantage of the PPC970 is in the bus speed. It allows Ativec enabled tasks to be performed much more quickly. Currently Altivec is bottlenecked by the FSB on the 7455s.

[mathiasr]

Originally posted by Dave Marsh
I've heard several people say that 64-bit apps would run a bit slower on the 970 than comparable 32-bit apps, which I find a bit confusing. If the processor was built to run 64-bit apps, and isn't just a juried-up 32-bit design finagled to run 64-bit apps (which we know ISN'T the case, since it's based on the POWER4 design), I don't understand why this would be the case.

What I mean is, hasn't the processor been designed to handle PURE 64-bit apps, with all the appropriate circuitry to accommodate that? And, when running 32-bit apps, doesn't that just mean that some of that additional circuitry just isn't being used (or perhaps is just being filled with leading zeroes)? If that's the case, and the processor is just moving data according to its clock rate, why would an app using only part of the design run faster?

Here is an example on IBMs site:
http://www-106.ibm.com/developerworks/linux/library/l-ppc/

To initialise a register with a pointer you need 2 instructions (PowerPC 32 bits) vs 5 instructions (PowerPC 64 bits).
This is a side effect of the instructions size that remains 32 bits in 64 bits mode, thus quantity of immediate data you can store inside is still the same (you'll find more explanations on IBMs page, just scroll to its bottom).
Using more instructions does not only take more time, it also takes L1 instruction cache space.

PowerPC 32:
lis 4,msg@ha # load top 16 bits of &msg
addi 4,4,msg@l # load bottom 16 bits

PowerPC 64:
lis 4,msg@highest # load msg bits 48-63 into r4 bits 16-31
ori 4,4,msg@higher # load msg bits 32-47 into r4 bits 0-15
rldicr 4,4,32,31 # rotate r4's low word into r4's high word
oris 4,4,msg@h # load msg bits 16-31 into r4 bits 16-31
ori 4,4,msg@l # load msg bits 0-15 into r4 bits 0-15

Another point is that you need more memory to store pointers (since they are 64 bits and no more 32 bits), this means you can keep only half as many in the L1 cache for instance, this could lead to more memory access.
If you have an array of 2000 pointers you want to copy, instead of moving 8.000 bytes, you have to move 16.000 bytes.

Inside the CPU working on 32 bits or 64 bits chuncks once they are in registers has no impact, but when you store/load 64 bits pointers in main memory or in the caches, they will take more space and put more stress on the memory subsystem.

[NicoMan]

Originally posted by Mudbug
I've got to agree with NicoMan - drop the 17's and get TWO Formac 2010's - I'll uphold my statement about the 20 over the two 17's, but two 20's for almost the same price as the two 17's makes even more sense.

ALTHOUGH - make sure your desk is wide enough to hold that - two of those bad boys side by side takes up some lateral room, if not depth on the desk.
I think the Formac 2010 is a little more expensive than the Apple 17", but the extra money is well worth it. So if you got money (which LordNadrog looks like having) and space then get the Formacs, they are fantastic (I have at work a dual 867 + 3 2010s - 1 running on a ATI 7000) with very good brightness and people have mentioned very good colours for graphic apps...)

NicoMan

PS: it looks like this dicussion has sidetracked a bit

[dekator]

MacBidouille has posted a commentary à propos their rumors:
1) They used a beta of Bryce 6 (availbale through P2P, name: Corel Bryce beta 6.sit, 91,2MB). This version is multi-processor optimized.
2) Similarity with Barfeats: Apple has allegedly used their P4 and Dual G4 1,42 benchmarks and run the PPC 970s through the same test routine.
3) Lionel addresses the question why they posted the rumors. First, a balance:
No money. Banners or similar don't yield more money due to increased traffic. (Well, they get money per buy, not even clic).
If the rumors are correct, they'll have probs with Apple, if not with their readers. Lastly, their server was under heavy stress. They just wanted to share information that "lets you dream".
Lionel says that he still believes the rumors to be true but will probably not post anymore rumors [of this kind] whether they'll be true or not.
Note that he said so before on similar occasions.

[deanbo]

A low end PPC 970 chip that has a clock speed equivalent to Apples current high end G4? What more do I have to say.

[MacBandit]

Originally posted by dekator
If the rumors are correct, they'll have probs with Apple............

Not necessarily true. Apple has taken a different tactic lately. Just look at ThinkSecret.com and it's prerelease information and renderings of the iPod. They were exactly spot on.

[dekator]

Originally posted by MacBandit
Not necessarily true. Apple has taken a different tactic lately. Just look at ThinkSecret.com and it's prerelease information and renderings of the iPod. They were exactly spot on.

Yes, agreed. Shares going up can't be wrong for Apple :-)

[NicoMan]

Originally posted by deanbo
A low end PPC 970 chip that has a clock speed equivalent to Apples current high end G4? What more do I have to say.
That Moto are rubbish? And that we hate them for the performance lag between Macs and PCs? No, you don't have to say it.

NicoMan

[NicoMan]

Originally posted by MacBandit
Not necessarily true. Apple has taken a different tactic lately. Just look at ThinkSecret.com and it's prerelease information and renderings of the iPod. They were exactly spot on.
I tell you what: them posting rumours on hardware coming out in 2 months or more CANNOT BE GOOD for Apple. How about PowerMac G4 inventory? Apple will sell their PowerMacs 970 no matter what. The problem is their existing inventory. And that's good money going to waste...

NicoMan

[deanbo]

Maybe I should requote my quote (ooohhh tricky). If you spent several thousand dollars on Apple's current high end G4 (dual 1.4 ghz), only to have Apple release a 1.4 ghz PPC 970 later several months(?) later as a low end machine (even if it is a single no dual processor) what would you be saying.
Doh!! and not likely.

[Dave Marsh]

Thanks all for the comments/explanations. I think I actually understood some of it.:)

[Dave Marsh]

I'm sorry to hear they're now a bit skittish. I've enjoyed their offerings and hope they reconsider.

Concerning the benchmarks inconsistencies, it's easy for me to see where an employee under NDA would be a bit rushed in putting something like this together. Also, it's not inconceivable that they would simply grab the Barefeats data for other platforms they either didn't have readily available, or didn't want to take the time to redo. The old data would only be for a relative comparison with a similar test on the new 970 platform after all...a taste of what may be on the way. We'll get real benchmarks soon enough.

I, for one, have enjoyed the hours I've spent this evening reading all the comments on this posting today. I've also read some very insightful comments from a few of the forum's more well connected members, which has added to my understanding of what to expect when the new Macs are released.

Now, I wonder, will we really get new eMacs tomorrow?:D

[NicoMan]

Originally posted by deanbo
Maybe I should requote my quote (ooohhh tricky). If you spent several thousand dollars on Apple's current high end G4 (dual 1.4 ghz), only to have Apple release a 1.4 ghz PPC 970 later several months(?) later as a low end machine (even if it is a single no dual processor) what would you be saying.
Doh!! and not likely.
Ok I got it. Yeah that's a DOH! A massive one btw. But the dual 1.42 has been out for a few months so if the 970 come out in 3 months or so, that's a normal upgrade cycle isn't it? The fact that it would be a massive upgrade doesn't really matter they are switching to a new chip provider, they are going to pace themselves with the chip production, the previous machine are pretty much irrelevant at this point.

I don't know if I'm really clear...

NicoMan

[Bonte]

Originally posted by nichrome
Simply not true. Two 17" monitors mean you can move pallettes, download managers etc. to a secondary monitor and not have them get buried under other windows. A single 20" doesn't let you do that. Additionally, if you do graphics, you can have one screen have a window with the image zoomed in and the other at 100 %. On a 20" screen you'd have to stack the windows and switch between the two. Very inconvenient.

Aye, I totally agree.
Worked for years with a 17 and 15' side by side and this works far better than a single 20'. Next will be definitely 2 apple 17" if I can spare the space.

[barkmonster]

Re: Requote


quote:
------------------------------------------------------------------------
Originally posted by deanbo
Maybe I should requote my quote (ooohhh tricky). If you spent several thousand dollars on Apple's current high end G4 (dual 1.4 ghz), only to have Apple release a 1.4 ghz PPC 970 later several months(?) later as a low end machine (even if it is a single no dual processor) what would you be saying.
Doh!! and not likely.
------------------------------------------------------------------------

Ok I got it. Yeah that's a DOH! A massive one btw. But the dual 1.42 has been out for a few months so if the 970 come out in 3 months or so, that's a normal upgrade cycle isn't it? The fact that it would be a massive upgrade doesn't really matter they are switching to a new chip provider, they are going to pace themselves with the chip production, the previous machine are pretty much irrelevant at this point.

I agree, it's happened a few times, infact every time apple have had signficantly faster G4s over the past few years, the entry level has ended up being the same clockspeed as the previous top model. They won't have the ability to artifically cripple the entry level model by eliminating L3 cache this time either, we don't need it anymore.

Powermac G4 timeline (gigabit models onwards)

400Mhz G4 as the low end and a dual 500Mhz G4 as the high end.

466Mhz G4 as the low end and 733Mhz as the high end. With the added bandwidth and other motherboard improvements, the 466Mhz G4 was actually faster than previous single 500Mhz G4s.

This is pretty much like going from dual 1.42 high end to single 1.4Ghz lowend, except the dual G4 would have obviously been faster than the single G4 for a lot of things even without the benfit of the digital audio G4s more efficient motherboard and memory system.

733Mhz as the low end (minus L3), 867Mhz mid range and Dual 800Mhz high end.

800Mhz as the low end (minus L3), Dual 1Ghz as the high end

The 'low <-> high' pattern breaks as apple realise they need breathing room for a speed bump over 2 product cycles

Dual 867Mhz as the low end (1Mb less L3 & DDR266), Dual 1.25Ghz as the high end.

This was actually the biggest speed bump they've ever done if you're running software that really screams on dual cpus

1Ghz (1Mb less L3 & DDR266), dual 1.25 as mid range (1Mb less L3), dual 1.42Ghz as the high end.

Now we get to the alleged future PPC970 lineup...

1.4Ghz PPC970 low end, dual 1.8Ghz PPC970 high end.

seems quite reasonable to me, we might know on paper that the systems are signficantly faster than the models they're replacing, for people less spec savy, they'd just see an entry level model that looks 40% faster than the previous one and some impressive bar graphs on an apple ad showing it kick a 3Ghz P4 into touch in a photoshop or maya bake off.

The fun thing is, Canterwood, even with the faster FSB over Northwood, isn't going make a 3.4Ghz P4 by Q4 of this year be super fast compared with the 3Ghz chips they have out now.

It's common knowledge that intel arn't planning anything quicker than 3.4Ghz by the end of the year, I can't see the increase of 1Mb on-die L2 and 200Mhz x 4 FSB making it anything more than a minor speed bump over the current P4 performance.

[Trimix]

Great thread, so forgive me for posting my thoughts in here, but only in this thread have I seen so many of you trying to make money on apple stock -

I feel that we shareholdes in Apple may finally be right after all -
great hardware, great ideas, great managmenet and the right vision in which direction to go.

Today, after an equally stunning start of the stock, our friendly analysts at Merrill came out bashing the stock, reiterating a sell recommendation - may I suggest that they do not truly know what they are talking about ?
Or even worse could they have shorted the stock and are trying to get out of a squeeze ?

In any event, and please Arn, don't kill my message, would it be not an interesting thought to maybe reconsider one's own allegiance to Merrill and their outstanding service ? Why, I may ask, should we shareholders support a broker with our retirement funds, our nesteggs and our orders, if he so chooses to torch the company, whose products we love dearly... and write endlessly about in forums like this ??
Sorry but I am slighly miffed...
:mad:

[Dave Marsh]

Wasn't this the company that last week was fined for their stockbrokers telling their customers to buy stock, that internally they were telling one another to sell? Something about conflict of interest with their investment bankers?

[Cubeboy]

Originally posted by barkmonster
The fun thing is, Canterwood, even with the faster FSB over Northwood, isn't going make a 3.4Ghz P4 by Q4 of this year be super fast compared with the 3Ghz chips they have out now.

It's common knowledge that intel arn't planning anything quicker than 3.4Ghz by the end of the year, I can't see the increase of 1Mb on-die L2 and 200Mhz x 4 FSB making it anything more than a minor speed bump over the current P4 performance.

Nah, .13 nm Pentium 4 pretty much reached it's maximum potential 1 year ago at 3.06 ghz. The only real reason the Pentium 4c (Canterwood) was released was to keep the 3200+ Athlons which are to be released this month from getting the performance crown.

The thing is Prescott, the Pentium 5 is slated for release in the second half of this year, in the same timeframe of the PPC970 and Athlon64.

It's will scale to 5.20 ghzs, have at least a 800 mhz bus, 1 mb L2 cache, 16 kb L1 cache (twice the amount on the Pentium 4), SSE3, 2nd Generation Hyperthreading, Advanced Branch Predictor, Lower Latency, to say a couple of the most visible features. Chip-Architect has a pretty good article about the possibility of 32/64 bit computing in Prescott as well as a very thorough analysis of all the features of Prescott based on the blueprints.

www.chip-architect.com (http://www.chip-architect.com)

[adamfilip]

I sure hope they release the 970's in june or sooner..
but im expecting apple to drop prices on current Towers before the new systems are released.

otherwise. there gonna have a much harder time selling the current series. when they release the 970's

im saving my pennies..
Dual 1.6 970's with a new enclosure.. common $1999 usd!

[i_am_a_cow]

Originally posted by Megaquad
if this 1.4 ghz is so fast, then 1.8 will eat little children!!


geez, sure hope it doesn't eat me :( . . .

[XnavxeMiyyep]

The fastest benchmarks (real or fake) only test the 970 against computers up to a Dell 3.06 GHz P4. How would it compare against a Dual 3.06 GHz Xeon?

[Mustafa Monde]

The vaunted 970 will come. Like many things rumored, it may fall short of the hype posted to this board. Building these things and integrating them with software correctly is challenging. I'm prepared to wait until some significant product surfaces.

Bear in mind, Apple consistantly wows with a new goodie every six months or so. Hopefully the next one will appeal to the pros.

[Shaktai]

Originally posted by XnavxeMiyyep
The fastest benchmarks (real or fake) only test the 970 against computers up to a Dell 3.06 GHz P4. How would it compare against a Dual 3.06 GHz Xeon?

Well I posted this before, but probably in another thread or even on another board. Barefeats has a comparision of the current PM dual 1.42, A P4 3.06 and a dual Xeon 2.4 ghz. You can probably extrapolate out from there to a dual Xeon 3.06. http://www.barefeats.com/pentium4.html

The results are interesting to say the least and based upon verifiable hardware. It is more difficult to extroplate out the possible performance of a dual 1.8 ghz 970. However I think that 2.5X faster then the dual 1.42 tested, would be a good overall conservative starting point.

[XnavxeMiyyep]

Originally posted by Shaktai
Well I posted this before, but probably in another thread or even on another board. Barefeats has a comparision of the current PM dual 1.42, A P4 3.06 and a dual Xeon 2.4 ghz. You can probably extrapolate out from there to a dual Xeon 3.06. http://www.barefeats.com/pentium4.html

The results are interesting to say the least and based upon verifiable hardware. It is more difficult to extroplate out the possible performance of a dual 1.8 ghz 970. However I think that 2.5X faster then the dual 1.42 tested, would be a good overall conservative starting point.
Thanks for the link. With that info in mind, the 970 will probably beat them all.:D