Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Community > MacNN Lounge > Official PPC 970 conjecture thread (or Eug's meaningless PPC 970 thread).

Official PPC 970 conjecture thread (or Eug's meaningless PPC 970 thread).
Thread Tools
Eug
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 06:12 PM
 
OK, some points and conjecture:

1) Motorola claims to have completed work on "low-k" technology for 0.13 um parts, which will supposedly allow up to 20% faster speed, and should allow them to lower power. They also claim that this has now been applied to 0.18 chips to increase speed.

2) Motorola G4 7455 1 GHz 0.18 um has a spec'd typical power output of 15 Watts without low-k. Thus, a low-k-ified 1 GHz 7455 would be even lower, say around 13 Watts typical I'm guessing.

3) Motorola G4 7457 1.3 GHz 0.13 um has a spec'd typical power output of 18.7 Watts without low-k. Thus a low-k-ified 1.3 GHz 7457 would be 16 Watts typical I'm guessing.

4) Despite Motorola's 1 GHz max speed for the G4 7455, 1.42 MHz versions are in current use and are technically NOT overclocked chips. One wonders if this "low-k" technology is already in these.

5) Motorola claims the 7457 will hit 2 GHz without huge problems. Since the 0.18 um 7455 is already at 1.42 GHz, I see no reason to doubt this claim.

6) Current PowerBooks, a brand new design, use the G4 7455 chips. However, due to the chip design, none of the CPUs can properly use the full DDR bandwidth.

7) G4 7457 chips are pin-for-pin drop-in compatible to 7455 motherboard designs.

8) IBM claims a 1.2 GHz PPC 970 has a typical power spec of 19 Watts.

9) IBM claims a 1.8 GHz PPC 970 has a typical power spec of 42 Watts.

10) There is no lower power technology yet announced for the PPC 970.

11) The PPC 970 should be faster on average clock for clock than the G4, but the G4 should hold its own on certain apps. In other words, for certain usage, while the PPC 970 is an awesome chip, the G4 will suffice, esp. when power and heat are a concern.

12) The biggest benefit of the PPC 970 is the requirement for a radical new bus design, to dramatically increase bandwidth.

13) As fast as the PPC 970 can be under optimal conditions, it is not as fast as the fastest P4 in most situations.

For these reasons I predict:

1) The new PowerBooks coming soon will include a new 15.2" AluBook, but all new PowerBooks will continue to use the G4, possibly the 7455 but preferably the 7457. There will be a speed bump across the board, but the 12" will continue to lack any L3 cache. Both the 15.2" and the 17" will have a 167 MHz bus, with the top speed being 1.17 MHz. However, all new PowerBooks will still be bandwidth crippled. (It's possible the machines would go 200 MHz FSB and 1.2 GHz top speed, but I have my doubts.)

2) The new PowerMacs will include a PPC 970. The speeds may only be up to 1.6 GHz to begin with, an incremental upgrade over the number of the G4 speed (ie. 1.42 GHz). However, both single (with slower chips) and dual machines (at 1.6 GHz x 2) will be introduced, with speeds overall significantly faster than current G4s. Dual machines will rival the fastest P4s overall and will hopefully overtake them in some apps. The single-processor machines will still be significantly slower however, for obvious reasons.

3) The iMacs will continue to mirror PowerBook speeds, and all will continue to use G4s. None will have an L3 cache.

4) The iBooks will continue to use the G3, because it would be difficult for Apple to slow down the MHz of the iBook to switch to a G4, and new PowerBooks will only be 1.17 GHz, so a 900 MHz iBook G4 would be too fast. Eventually, the iBooks will convert to a G4, but they will all be L3 cache-less.

5) The Xserves will be updated to PPC 970 either around the same time as the PowerMac or else later. They will not be PPC 970 before the PowerMac is.

6) Panthwire will be required to run a PPC 970 machine.
( Last edited by Eug; Jun 7, 2003 at 06:20 PM. )
     
Angus_D
Addicted to MacNN
Join Date: Mar 2000
Location: London, UK
Status: Offline
Reply With Quote
Jun 7, 2003, 06:22 PM
 
I predict that the 970 will come with an in-built kettle for making tea.
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 06:23 PM
 
Originally posted by Angus_D:
I predict that the 970 will come with an in-built kettle for making tea.
Water cooling?
     
paulscan
Fresh-Faced Recruit
Join Date: Jun 2003
Location: NC, USA
Status: Offline
Reply With Quote
Jun 7, 2003, 06:26 PM
 
Originally posted by Angus_D:
I predict that the 970 will come with an in-built kettle for making tea.
That's somewhat unsubstantiated. Nobody drinks tea anymore. The 970 will come with an integrated soda dispenser.
     
ironknee
Addicted to MacNN
Join Date: May 1999
Location: New York City
Status: Offline
Reply With Quote
Jun 7, 2003, 06:58 PM
 
Originally posted by Eug:
OK, some points and conjecture:


6) Panthwire will be required to run a PPC 970 machine.
so if that's true, then no PPC 970 machine available for sale during wwdc?

Is that a good thing? If one can run Jaguar on the machines, Apple can start selling them to the pro customers who probably have been waiting for a G5 machine...just a though
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 07:27 PM
 
Originally posted by ironknee:
so if that's true, then no PPC 970 machine available for sale during wwdc?

Is that a good thing? If one can run Jaguar on the machines, Apple can start selling them to the pro customers who probably have been waiting for a G5 machine...just a though
No PPC 970 for sale at WWDC. Indeed, I'm still not convinced it will even be announced there.

Now the question is, how long until Panther is shipping?
     
bradoesch
Professional Poster
Join Date: Jun 2000
Status: Offline
Reply With Quote
Jun 7, 2003, 07:30 PM
 
If the 970 is actually cheaper to produce, I hope Apple will lower the prices on the towers to reflect this.


edit: Removed what I probably never should have put in the first place.
( Last edited by bradoesch; Jun 7, 2003 at 08:52 PM. )
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 07:31 PM
 
That's somewhat unsubstantiated. Nobody drinks tea anymore. The 970 will come with an integrated soda dispenser.
I feel honoured that your first post was in my thread.
( Last edited by Eug; Jun 7, 2003 at 07:38 PM. )
     
ironknee
Addicted to MacNN
Join Date: May 1999
Location: New York City
Status: Offline
Reply With Quote
Jun 7, 2003, 07:35 PM
 
Originally posted by Eug:
No PPC 970 for sale at WWDC. Indeed, I'm still not convinced it will even be announced there.

Now the question is, how long until Panther is shipping?
but isn't panther's biggest feature is that it would be 64 naitive? and if they don't announce the 970, why would you make panther 64 then?

wondering....
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 07:37 PM
 
Originally posted by bradoesch:
If the 970 is actually cheaper to produce, I hope Apple will lower the prices on the towers to reflect this. I'd like to buy a tower one day, but I'd like the price to be close to a comparable Intel system.
Where did you get that? So far I've only seen a rumour that the 970 is going to be cheaper, but my guess is that it'd actually be more expensive to produce.

Remember that the G4 is 33 million transistors with a die size of 106 square mm, and the PPC 970 will be 52 million transistors with a die size of 121 squre mm.

Unless IBM plans on having a very low margin on their chips, it would stand to reason that the PPC 970 will be more expensive than the G4. This is especially true given that we'd be comparing the already shipping 1.42 GHz G4 to the brand new 1.6 or 1.8 GHz PPC 970.
     
Severed Hand of Skywalker
Addicted to MacNN
Join Date: Apr 2001
Location: The bottom of Cloud City
Status: Offline
Reply With Quote
Jun 7, 2003, 07:45 PM
 
Panther will come no later then September, they are previewing the darn thing in 16 days.

I think a slightly modded 10.2 will come with the 970 chips if they ship before september.

"Ahhhhhhhhhhhhhhhh"
     
BrunoBruin
Mac Elite
Join Date: May 2001
Location: Northampton, MA USA
Status: Offline
Reply With Quote
Jun 7, 2003, 07:59 PM
 
Originally posted by Eug:
OK, some points and conjecture:
All very reasonable, if you ask me. I'm pining for a 970 PowerBook but I'd be happy with a faster 7457 G4. Do we have to wait for the 7457 for a 200MHz bus? (I assume the 7455 tops out at 167 or we'd have seen faster buses in the towers, since Apple seems to have tried to eke out every last bit of performance within the contraints of the processor.)
     
King Bob On The Cob
Mac Elite
Join Date: Apr 2002
Location: Illinois
Status: Offline
Reply With Quote
Jun 7, 2003, 08:02 PM
 
Originally posted by Eug:
Where did you get that? So far I've only seen a rumour that the 970 is going to be cheaper, but my guess is that it'd actually be more expensive to produce.

Remember that the G4 is 33 million transistors with a die size of 106 square mm, and the PPC 970 will be 52 million transistors with a die size of 121 squre mm.

Unless IBM plans on having a very low margin on their chips, it would stand to reason that the PPC 970 will be more expensive than the G4. This is especially true given that we'd be comparing the already shipping 1.42 GHz G4 to the brand new 1.6 or 1.8 GHz PPC 970.
IBM has more at stakes with these chips and will use them in their own systems. They will do all they can to keep manufacturing prices down. Since more companies are using the chips, more chips will be made which means better pricing due to bulk ordering.
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 08:08 PM
 
but isn't panther's biggest feature is that it would be 64 naitive? and if they don't announce the 970, why would you make panther 64 then?
Well first off, the 64-bit thing is also a rumour, not fact. Second, even if they made Panther 64-bit and did it for the 970, I think it'd make sense to be developing a 64-bit OS anyway, because the future of computing OSes is 64-bit regardless if the 970 exists or not. The timing might just be a bit different.

More importantly however, remember that Panther won't be shipping until several months later (Sept. according to SHoS). Sure, it will be previewed in a little over two weeks, but that doesn't mean Apple has to release the PPC 970 at the preview. Indeed, if PPC 970 wasn't completely ready for sale at WWDC or within a month or two, I don't think it'd really be a good idea to preview the hardware there at WWDC. It would destroy any miniscule sales the G4 PowerMac still has.

Now many people are convinced the PPC 970 WILL be ready for order then, and that it will ship with a hacked X.2. Well, it seems to me it'd make more sense to sell the PPC 970 with an OS developed specifically for it, rather than spend a bunch of extra money and resources hacking X.2 to work this chip and motherboard architecture, only to have it supplanted by a completely new OS a couple of months later. (And I'm ignoring the 64-bit Panther thing at this time. My argument stands regardless if Panther is 32-bit of if there's a 64-bit version as well.)
     
Severed Hand of Skywalker
Addicted to MacNN
Join Date: Apr 2001
Location: The bottom of Cloud City
Status: Offline
Reply With Quote
Jun 7, 2003, 08:29 PM
 
Don't all apps have to be made 64 bit also? Apple would probibly release the 970's this month and tell developers to to get their butt in gear to have their apps ready by the time the 64 bit 10.3 ships.

"Ahhhhhhhhhhhhhhhh"
     
OldManMac
Addicted to MacNN
Join Date: Jul 2001
Location: I don't know anymore!
Status: Offline
Reply With Quote
Jun 7, 2003, 08:40 PM
 
Originally posted by paulscan:
That's somewhat unsubstantiated. Nobody drinks tea anymore. The 970 will come with an integrated soda dispenser.
Just a doggone minute, whippersnapper! I drink green tea every day, 'cause them there antioxidants is good fer ya!
Why is there always money for war, but none for education?
     
bradoesch
Professional Poster
Join Date: Jun 2000
Status: Offline
Reply With Quote
Jun 7, 2003, 08:51 PM
 
Originally posted by Eug:
Where did you get that? So far I've only seen a rumour that the 970 is going to be cheaper, but my guess is that it'd actually be more expensive to produce.

Remember that the G4 is 33 million transistors with a die size of 106 square mm, and the PPC 970 will be 52 million transistors with a die size of 121 squre mm.

Unless IBM plans on having a very low margin on their chips, it would stand to reason that the PPC 970 will be more expensive than the G4. This is especially true given that we'd be comparing the already shipping 1.42 GHz G4 to the brand new 1.6 or 1.8 GHz PPC 970.
Both our sources are rumors. What I meant to say instead of "If the 970 is actually cheaper to produce" is "If the rumours are true and the 970 is actually cheaper to produce." Hopefully that makes more sense.


Brad
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 7, 2003, 11:54 PM
 
Both our sources are rumors. What I meant to say instead of "If the 970 is actually cheaper to produce" is "If the rumours are true and the 970 is actually cheaper to produce." Hopefully that makes more sense.
Actually, mine wasn't a rumour. It was just my own guess that the 970 might be more expensive than the G4.

It is already well known (and not a rumour) that the 970 is a much more complex chip than the G4. We already have pictures of the actual chip posted on the web - and they're now being made into desktop backgrounds. The fact that it's more complex, that it will be so new, that it's faster, and that it's meant predominantly as a (low-end) server chip, suggests it is unlikely to have bargain basement prices. Sure, it might not be humungously expensive either, but I find it very difficult to believe that IBM is going to price this supposedly cutting edge 64-bit server chip at a significantly lower price than Motorola's chip for embedded applications.

They will do all they can to keep manufacturing prices down. Since more companies are using the chips, more chips will be made which means better pricing due to bulk ordering.
And of course, Motorola would strive to do the same thing. Apple is actually only a relatively small market for their chips in fact. They sell a lot of these chips.
     
ckohler
Grizzled Veteran
Join Date: Jan 2001
Location: Evansville, IN
Status: Offline
Reply With Quote
Jun 8, 2003, 12:47 AM
 
Originally posted by Severed Hand of Skywalker:
Don't all apps have to be made 64 bit also? Apple would probibly release the 970's this month and tell developers to to get their butt in gear to have their apps ready by the time the 64 bit 10.3 ships.
No, they don't have to. The 970 is fully 32bit compatible and all existing Mac apps will run on the 970 without modification or need for emulation.

Also, the real strength of the 970 isn't in its 64bit capability but its high bandwidth archetecture. Even today's 32bit apps will fly on the 970 and future apps compiled using 970 aware compilers will be even faster. Only apps that can take advantage of large memory chunks (OSX, Database, Rendering, etc) will get any benefit from the 64bit capability.
     
Severed Hand of Skywalker
Addicted to MacNN
Join Date: Apr 2001
Location: The bottom of Cloud City
Status: Offline
Reply With Quote
Jun 8, 2003, 12:57 AM
 
Originally posted by ckohler:
No, they don't have to. The 970 is fully 32bit compatible and all existing Mac apps will run on the 970 without modification or need for emulation.
That isn't really what I wanted to say. I mean don't all apps also have to be updated to 64 bits to get the speed increase, not just the OS.

"Ahhhhhhhhhhhhhhhh"
     
Dogma
Mac Elite
Join Date: Jan 2001
Location: Cumbria, England
Status: Offline
Reply With Quote
Jun 8, 2003, 02:00 AM
 
Good reasoning, nice research etc.

As for what I'm hoping for...

Well, for one. Just plain faster machines. I don't care how they do it anymore. Make them hit the 2Ghz barrier this side of 2004 at least. I'm thinking fingers out of butts Apple, and make machines that people really aspire to own again. It's not as if their advertising does anything, it's selling a lifestyle, and computers are tools.

Apple make the best tools for us computer guys. In my own opinion. However, sometimes, just sometimes, I would like to drive a TVR instead of a boring old BMW. It may be quality, it may be a smooth ride getting you there, but damn, it sure as hell can take a while to get there...
Hark, I hear a robin sig'ing in the trees!
Nae, there is no sog to be sug,
or am I wrog? Why can't I sig?
     
Gul Banana
Mac Elite
Join Date: May 2002
Status: Offline
Reply With Quote
Jun 8, 2003, 06:38 AM
 
Originally posted by Severed Hand of Skywalker:
That isn't really what I wanted to say. I mean don't all apps also have to be updated to 64 bits to get the speed increase, not just the OS.
Not really, since most applications won't get a significant speed increase out of just the move to 64-bit. It's the other features of the 970 that will make most things faster clock-for-clock.
[vash:~] banana% killall killall
Terminated
     
Angus_D
Addicted to MacNN
Join Date: Mar 2000
Location: London, UK
Status: Offline
Reply With Quote
Jun 8, 2003, 08:09 AM
 
Originally posted by paulscan:
Nobody drinks tea anymore.
A large number of Brits do.
( Last edited by Angus_D; Jun 8, 2003 at 08:57 AM. )
     
Severed Hand of Skywalker
Addicted to MacNN
Join Date: Apr 2001
Location: The bottom of Cloud City
Status: Offline
Reply With Quote
Jun 8, 2003, 08:15 AM
 
Originally posted by Gul Banana:
Not really, since most applications won't get a significant speed increase out of just the move to 64-bit.
Photoshop? iDVD? iMovie?

"Ahhhhhhhhhhhhhhhh"
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 8, 2003, 10:36 AM
 
32-bit --> 64-bit may not mean much for speed, except for a few specific apps AFAIK. The speed increase will come from just using the faster chip (and faster bus, etc.).

Anyways, what I'm saying about Panther isn't that it has to be 64-bit to run the 970, because it doesn't. It's quite possible that the 970 could ship with a 32-bit Panther initially. The reason I think it should be Panther though (whether 32 or 64-bit) is because adding PPC 970 support to an OS wouldn't be as easy as adding 1.42 GHz dual G4 support when the OS already supports a dual 1.0 GHz.

With the new architecture (eg. chip, bus, video, etc.), one would require a whole new set of drivers and potential OS design changes. Wouldn't it be easier to design an OS specifically to include those right from the outset (ie. Panther), rather than to try to tack them on later to an OS which is already shipping? Apple could do this because they could have had the luxury of knowing that the 970 and Panther would be released around the same time. I'm not saying it's impossible that the 970 would ship with X.2. I could be talking out of my @ss, but it just seems simpler if it shipped with X.3.

BTW, some other things I should point out:

1) The PowerBooks have had a significant price drop recently. It seems likely then that they are going to be updated soon. (And it's not just because of the crappy US dollar, because the US prices have dropped, too.)

2) The PowerMacs have not dropped in price. Before new PowerMacs are introduced, do the previous models always drop in price? I dunno. If they do drop in price, how soon in advance? WWDC is in 2 weeks.

3) Apple will still be at MWNY, which is 3 weeks after WWDC. They have a feature presentation on Wednesday July 16. If Apple demo'd the 970 at MWNY they could say that you could order it immediately after the presentation, for delivery at the end of August or beginning of September. This would be similar to the timing for the 17" AluBook back in January, and would allow the 970 to ship with Panther. (And of course, MWNY is a bigger venue for media coverage, and they could stream the MWNY presentation to the whole world. I don't think Apple plans to stream the WWDC presentation, although they might add the video to their QT site later.)

One problem with this possibility though is that Apple is only scheduled for a one-hour presentation, and there are other meetings scheduled later. Would one hour be enough? Has Apple gone overtime before? Another problem is that rumour has it that Jobs will not be there (which would suggest no new hardware), but who knows?
( Last edited by Eug; Jun 8, 2003 at 10:43 AM. )
     
hayesk
Guest
Status:
Reply With Quote
Jun 8, 2003, 11:26 AM
 
I'm still holding on to WWDC being a software-only show.

New Macs won't be announced until July, IMHO.
     
bewebste
Senior User
Join Date: Mar 2000
Location: Ithaca, NY
Status: Offline
Reply With Quote
Jun 8, 2003, 12:29 PM
 
Originally posted by Eug:
Anyways, what I'm saying about Panther isn't that it has to be 64-bit to run the 970, because it doesn't. It's quite possible that the 970 could ship with a 32-bit Panther initially. The reason I think it should be Panther though (whether 32 or 64-bit) is because adding PPC 970 support to an OS wouldn't be as easy as adding 1.42 GHz dual G4 support when the OS already supports a dual 1.0 GHz.

With the new architecture (eg. chip, bus, video, etc.), one would require a whole new set of drivers and potential OS design changes. Wouldn't it be easier to design an OS specifically to include those right from the outset (ie. Panther), rather than to try to tack them on later to an OS which is already shipping? Apple could do this because they could have had the luxury of knowing that the 970 and Panther would be released around the same time. I'm not saying it's impossible that the 970 would ship with X.2. I could be talking out of my @ss, but it just seems simpler if it shipped with X.3.
While 32 bit applications don't need to be modified to run on the 970, the core of operating system will certainly need to be 64 bit aware in order to run properly. Just as one example, the virtual memory system would absolutely require modification to work correctly on 64 bit address spaces. So, there's definitely no way that Jaguar would be able to run on a 970.

I agree though that, assuming that Apple is indeed planning on shipping the 970 soon (not necessarily in June/July, but say by the end of the year) then it would make sense either for Panther to ship at the same time or for it to ship out but have the support for the 970 machines even before they come out. 64 bit support definitely doesn't seem like something you'd want to do with a 10.3.1 updater package.
     
bewebste
Senior User
Join Date: Mar 2000
Location: Ithaca, NY
Status: Offline
Reply With Quote
Jun 8, 2003, 12:31 PM
 
Originally posted by hayesk:
I'm still holding on to WWDC being a software-only show.

New Macs won't be announced until July, IMHO.
It's very true that Apple typically doesn't announce new hardware at WWDC, with a few minor exceptions. However, I wouldn't be surprised if they were to announce that they'll be using the 970 chips so that they could have some preliminary sessions telling the developers what to expect from the new hardware, even if the hardware itself doesn't ship for another month or three.
     
kmkkid
Professional Poster
Join Date: Mar 2002
Location: Brantford, ON. Canada
Status: Offline
Reply With Quote
Jun 8, 2003, 12:59 PM
 
If Apple doesnt present MAJOR hardware upgrades for all lines this year, they are going to be committing suicide (yeah, I know you've heard this all before ). Everyone is holding out for a GOOD machine, and if they continue to just bump the specs slightly even though they have the speed in the 970, people are going to take their business elsewhere. I can almost guarantee that they are losing alot of sales as we speak because of people waiting for the 970's or equivelent, both consumer and pro.


Chris
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 8, 2003, 04:44 PM
 
While 32 bit applications don't need to be modified to run on the 970, the core of operating system will certainly need to be 64 bit aware in order to run properly. Just as one example, the virtual memory system would absolutely require modification to work correctly on 64 bit address spaces. So, there's definitely no way that Jaguar would be able to run on a 970.

I agree though that, assuming that Apple is indeed planning on shipping the 970 soon (not necessarily in June/July, but say by the end of the year) then it would make sense either for Panther to ship at the same time or for it to ship out but have the support for the 970 machines even before they come out. 64 bit support definitely doesn't seem like something you'd want to do with a 10.3.1 updater package.
OK, thx for the clarification. I guess perhaps it'd be better to say that to run 970 one may not need a complete 64-bit implementation, even if it is desirable. Either way, then that's even more evidence pointing to the conclusion that WWDC will not see the introduction of the 970. I'd say MWNY or later.
However, I wouldn't be surprised if they were to announce that they'll be using the 970 chips so that they could have some preliminary sessions telling the developers what to expect from the new hardware, even if the hardware itself doesn't ship for another month or three.
Well, that makes sense from the developer point of view, but from the business and marketing point of view it would seem ill-advised, no? 3 months is quite the lag time (even for Apple) for hardware introduction to shipment. That's a full quarter where Apple has virtually no PowerMac sales. Mind you if they could push up the release of Panther to August then I might see that happening. Otherwise no 970 at WWDC.

What software would actually benefit from a 64-bit implementation anyway? How much tweaking are people going to need to make their software work with a 64-bit aware Panther, if any at all? Could Apple not simply make the 970 tweaks effectively invisible to the average developer? (ie. If it works with Panther, it will work with the 970.)
If Apple doesnt present MAJOR hardware upgrades for all lines this year, they are going to be committing suicide (yeah, I know you've heard this all before ).
Few are suggesting that the 970 won't be released until 2004. I just don't think it will appear as early as WWDC (even though I hope it would).
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 9, 2003, 09:46 AM
 
P.S. Here are some tantalizing SPEC numbers, in case you missed previous mention of them:

Code:
CPU SPECint2000 SPECfp2000 Notes ----------------------------------------------------------------- Motorola G4 7455 1.4 GHz 418 248 From the web Pentium III 1 GHz 448 316 IBM Power4+ 1.45 GHz 909 1221 IBM PPC 970 1.8 GHz 937 1051 Estimated Intel P4 2.40C GHz 984 1081 Intel P4 3.0 GHz 1164 1213
Some people have different P4 SPEC numbers, but Intel has since upgraded their CPUs with Hyper-Threading.

If these numbers are to be believed, a PPC 970 @ 1.8 GHz should perform at about the level of a P4 2.3 GHz with Hyper-Threading on the fastest Intel motherboard.

So don't expect a single PPC 970 to be competitive with the fastest P4. However, even if PPC 970's performance is only 3/4 of claimed speed, under optimal conditions a dual PPC 970 should crush the fastest current P4. And the P4 cannot be used in an SMP configuration. (I am not factoring in the SMP-able Intel Xeons or Itaniums, because these CPUs are not used in desktops.)
     
Spliffdaddy
Posting Junkie
Join Date: Oct 2001
Location: South of the Mason-Dixon line
Status: Offline
Reply With Quote
Jun 9, 2003, 10:48 AM
 
well, technically the 970 is a low-end server chip. You kinda have to compare it with a Xeon - Intel's low-end server chip.

While plain P4's are not SMP-capable it still exists in their design. Only because Intel 'crippled' it during production does it not work. Previous generations of Intel desktop processors were SMP-capable (celeron, for example), although Intel rarely touted this fact.

What about dual Athlons? How does the 970 stack up against that?
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 9, 2003, 11:43 AM
 
Originally posted by Spliffdaddy:
well, technically the 970 is a low-end server chip. You kinda have to compare it with a Xeon - Intel's low-end server chip.

While plain P4's are not SMP-capable it still exists in their design. Only because Intel 'crippled' it during production does it not work. Previous generations of Intel desktop processors were SMP-capable (celeron, for example), although Intel rarely touted this fact.

What about dual Athlons? How does the 970 stack up against that?
While you're technically right, I didn't include the Xeon because it's not really a desktop chip. OTOH, pricing would be similar to a top-of-the-line dual PPC 970 PowerMac so the argument does have some merit. I dunno if I'd wanna run standard software though, since it'd be like running consumer software on an Xserve. It would work fine most of the time, but who knows what problems you'd run into. Anyways, AFAIK, a Xeon runs essentially the same as P4, but adds SMP capabilities. Itanium is a whole different kettle of fish.

Dual Athlons? They should do well. They're currently at 2.167 GHz (so-called 3000+), but are limited by a 333 MHz bus. The Opterons will be interesting too, but they're really competitors to Xeons and Itanium. I dunno what is happening with Clawhammer.

Here are some dual x86 vs. single x86 benches of some slower machines. A dual Athlon runs about 60% faster than a single in this bench.

And don't forget, IBM has said its blades should run at "1.8-2.5 GHz". My guess is a 2.5 GHz PPC 970 is roughly about the speed of a P4 3.2 GHz.
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 9, 2003, 05:42 PM
 
An eWeek article claims the new Q37 PowerMacs are just Round-The-Corner (TM). (August)

It will supposedly run a hacked Jagwire, called Smeagol.

"Sources said that the IBM chip will make its first appearance in a new Power Mac known internally as Q37. However, sources said, Q37 won't ship with a 64-bit version of Mac OS X, limiting OS performance gains in the initial release. Instead, Q37 will launch with a special build train of the current Mac OS X Version 10.2, a k a Jaguar."

"This build, code-named Smeagol, will run on the new chip but won't take advantage of many of its key features, including 64-bit support. Sources said Apple's goal for Smeagol is to deliver Mac OS X performance at least "on par" with what Jaguar could achieve on Motorola G4 chips running at the same speed; the move will allow Apple to ship the new hardware before Mac OS X 10.3, a k a Panther, can take advantage of all the new processor's capabilities."

"Apple's current plans call for wrapping up development of Smeagol within a month or so of WWDC, suggesting that Q37 may ship by August; however, sources were unable to confirm specifics of Apple's release schedule."
     
MindFad
Posting Junkie
Join Date: Sep 2001
Status: Offline
Reply With Quote
Jun 9, 2003, 06:31 PM
 
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 9, 2003, 09:55 PM
 
Originally posted by MindFad:
Heh. That's awesome.

     
I Bent My Wookiee
Mac Elite
Join Date: Apr 2002
Location: Chillin' at the back of the Falcon
Status: Offline
Reply With Quote
Jun 9, 2003, 10:46 PM
 
Originally posted by Eug:
An eWeek article claims the new Q37 PowerMacs are just Round-The-Corner (TM). (August)

It will supposedly run a hacked Jagwire, called Smeagol.
Just like I said.

Well minus the funny name.

"Barwaraaawww"
     
paulscan
Fresh-Faced Recruit
Join Date: Jun 2003
Location: NC, USA
Status: Offline
Reply With Quote
Jun 9, 2003, 11:49 PM
 
Originally posted by Angus_D:
A large number of Brits do.
Brits also pay substantially more for Apple products.
     
kupan787
Senior User
Join Date: Jun 1999
Location: San Jose, CA
Status: Offline
Reply With Quote
Jun 9, 2003, 11:54 PM
 
Originally posted by Eug:
4) Despite Motorola's 1 GHz max speed for the G4 7455, 1.42 MHz versions are in current use and are technically NOT overclocked chips. One wonders if this "low-k" technology is already in these.
Have you seen the fans and heat sinks on those towers. I seriously doubt they are the "low-k" chips...

5) Motorola claims the 7457 will hit 2 GHz without huge problems. Since the 0.18 um 7455 is already at 1.42 GHz, I see no reason to doubt this claim.
This assumes they work on this chip. I firmly believe Apple will never use them in a single Mac, there will be no need. Between the current G3 and current G4 (7455), GOBI, the 850, 970 (at .13 and then at .09), I think Apple will be set.

10) There is no lower power technology yet announced for the PPC 970.
I would say the drop to 90NM would be good. That will drop the wattage disipate by the 1.8 and less parts. So in a way, there is a plan, not a "special technology".

11) The PPC 970 should be faster on average clock for clock than the G4, but the G4 should hold its own on certain apps. In other words, for certain usage, while the PPC 970 is an awesome chip, the G4 will suffice, esp. when power and heat are a concern.
What application are you think of that the G4 could beat a 970 in? The 970 is said to run at twice the clock of the G4 (wording on that is bad). So a 1GHz 970 is the equivilant of a 2GHz G4. The 970 will blow away anything a G4 can do...

13) As fast as the PPC 970 can be under optimal conditions, it is not as fast as the fastest P4 in most situations.
Proof? And please don't quote SPEC, as that is one "benchmark" and the P4s compiler has already been heavily optimized to give good results. Until we have other, more realistic benchs, no one can make a claim one way or the other.

4) The iBooks will continue to use the G3, because it would be difficult for Apple to slow down the MHz of the iBook to switch to a G4
I would bet that the iBooks never see G4s. They will continue with the IBM G3, onto GOBI, and then onto the 850 (rapid/io, altivec, super G3). Mention of this was over at the appleinsider forums.

5) The Xserves will be updated to PPC 970 either around the same time as the PowerMac or else later. They will not be PPC 970 before the PowerMac is.
I bet they wont see 970s for a while. I have seen mentioend many places that IT heads don't like buying untested tech. Apple won't spring a 970 in those right off the bat. Maybe after rev 2 970 powermacs come out.

6) Panthwire will be required to run a PPC 970 machine.
Nah. An updated 10.2.X will be all that is needed. Why? Because the 970 can run 32 bit code at native speed, no performance hit. So a point release will need to be made so the new hardware is recognized (just like any other hardware update, they get a "newer" OS).

Panther will just "unlock" the 64bitness, and allow for all the features 64bitness brings (more ram, 64bit apps, etc).
( Last edited by kupan787; Jun 10, 2003 at 12:18 AM. )
     
kupan787
Senior User
Join Date: Jun 1999
Location: San Jose, CA
Status: Offline
Reply With Quote
Jun 9, 2003, 11:58 PM
 
Originally posted by Eug:
P.S. Here are some tantalizing SPEC numbers, in case you missed previous mention of them:
Interseting to note @ http://www.intel.com/performance/res...op/tables.htm:

P4 @ 3.0 GHz

specfp 1213
specint 1164

P4 @ 3.06 GHz

specfp 1057
specint 1109

Now both of these are listed as Intel® Pentium® 4 Processor with HT Technology and Intel 875P Chipset. So why is the 3.0GHz part so much better?
     
I Bent My Wookiee
Mac Elite
Join Date: Apr 2002
Location: Chillin' at the back of the Falcon
Status: Offline
Reply With Quote
Jun 10, 2003, 12:11 AM
 
Originally posted by Eug:

8) IBM claims a 1.2 GHz PPC 970 has a typical power spec of 19 Watts.

10) There is no lower power technology yet announced for the PPC 970.
Don't the current G4 chips in the powerbooks use 20 watts?

So a new 1.2 GHZ 970 in a laptop would be a nice change.

"Barwaraaawww"
     
scottiB
Professional Poster
Join Date: Jan 2000
Location: Near Antietam Creek
Status: Offline
Reply With Quote
Jun 10, 2003, 12:12 AM
 
What application are you think of that the G4 could beat a 970 in? The 970 is said to run at twice the clock of the G4 (wording on that is bad). So a 1GHz 970 is the equivilant of a 500MHz G4. The 970 will blow away anything a G4 can do...
I think you inadvertantly switched that: a 500MHz 970 = 1GHz G4
I am stupidest when I try to be funny.
     
kupan787
Senior User
Join Date: Jun 1999
Location: San Jose, CA
Status: Offline
Reply With Quote
Jun 10, 2003, 12:18 AM
 
Originally posted by scottiB:
I think you inadvertantly switched that: a 500MHz 970 = 1GHz G4
Oops, you are right, and I just fixed it
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 10, 2003, 12:23 AM
 
What application are you think of that the G4 could beat a 970 in? The 970 is said to run at twice the clock of the G4 (wording on that is bad). So a 1GHz 970 is the equivilant of a 500MHz G4. The 970 will blow away anything a G4 can do...
I'm talking if the G4 were at equal GHz. For instance RC-5 speeds of the G4 and PPC 970 would be similar, judging by IBM's own numbers. But overall, the IBM PPC 970 should have faster GHz for GHz numbers judging by SPEC, etc.

Proof? And please don't quote SPEC, as that is one "benchmark" and the P4s compiler has already been heavily optimized to give good results. Until we have other, more realistic benchs, no one can make a claim one way or the other.
SPEC isn't perfect, but judging by the posts on the various forums, it's the best there is.

I would bet that the iBooks never see G4s. They will continue with the IBM G3, onto GOBI, and then onto the 850 (rapid/io, altivec, super G3). Mention of this was over at the appleinsider forums.
A G3 with Altivec basically is a G4. It doesn't really matter whose chip they use actually. It just has to have Altivec, since so much of Apple is now built around it (most importantly iDVD and OS X).

Now both of these are listed as Intel® Pentium® 4 Processor with HT Technology and Intel 875P Chipset. So why is the 3.0GHz part so much better?
Because the 3.06 runs on a 533 MHz bus. The 3.0 runs on an 800 MHz bus.
( Last edited by Eug; Jun 10, 2003 at 12:29 AM. )
     
kupan787
Senior User
Join Date: Jun 1999
Location: San Jose, CA
Status: Offline
Reply With Quote
Jun 10, 2003, 02:29 AM
 
Originally posted by Eug:
I'm talking if the G4 were at equal GHz.
And if the 970 was at equal GHz to a P4, it would blow it away....


SPEC isn't perfect, but judging by the posts on the various forums, it's the best there is.
I wouldn't say it is the best. I also wouldn't say it is the worst. All I am saying is lets wait and see until some other "benchmarks" come out (photoshop, after effects, quake 3, etc, etc) before we decide who the winner is.

A G3 with Altivec basically is a G4.
The 850 is most definatly not a G4. It will be what the G4 should have been (rapidio, true ddr system bus, scalable fsb, etc).

It doesn't really matter whose chip they use actually. It just has to have Altivec, since so much of Apple is now built around it (most importantly iDVD and OS X).
Agreed.

Because the 3.06 runs on a 533 MHz bus. The 3.0 runs on an 800 MHz bus.
Ok, I didn't see that mentioend on the website, and don't follow the PC side that much. It just struck me as odd that a slower chip would run faster.

But, if I am not mistaken, its not really a 800MHz bus, but ratehr a quad pumbed 200MHz bus. Where as the 970's bus runs at 1/4 clock speed of the processor. So a 1.8GHz 970 has a 450MHz real bus (900Mhz double pumped).

So I am wondering, would a 4x200MHz bus be slower than a 2x400MHz bus (say for the 1.6GHz 970)? Would the P4 bus have more latency?
     
- - e r i k - -
Posting Junkie
Join Date: May 2001
Location: Brisbane, Australia
Status: Offline
Reply With Quote
Jun 10, 2003, 05:59 AM
 

[ fb ] [ flickr ] [♬] [scl] [ last ] [ plaxo ]
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 10, 2003, 08:47 AM
 
And if the 970 was at equal GHz to a P4, it would blow it away....
True, except there won't be any 3 GHz PPC 970 chips for a long time, if ever. (Maybe by that time it will be a PPC 980?) The reason I mention a similarly clocked G4 is because it is thought that PPC 970 was originally to debut at 1.4 to 1.8 GHz. A 1.4 GHz G4 has been out for months. Furthermore, eWeek claims that with Smeagol, a PPC 970 would not be much faster than a similarly clocked G4. (If true, hopefully, Panther will truly increase GHz for GHz speed of the PPC 970.) That said, I get the feeling that the top PPC 970 released PPC 970 will be at 1.6 or possibly 1.8 GHz, esp. if IBM's 1.8-2.5 GHz claim for their blade servers holds any water. Mind you, some people think that the faster 7457 chips might be released very soon too.

The 850 is most definatly not a G4. It will be what the G4 should have been (rapidio, true ddr system bus, scalable fsb, etc).
True, but that is just getting into semantics now. Either way, it ain't a G3 as we have come to known the G3. Indeed, I wouldn't be surprised if Apple simply marketed it as a G4 or G4 Extreme or something like that.

But, if I am not mistaken, its not really a 800MHz bus, but ratehr a quad pumbed 200MHz bus. Where as the 970's bus runs at 1/4 clock speed of the processor. So a 1.8GHz 970 has a 450MHz real bus (900Mhz double pumped).
Well, the PC is more like 2x400x2, since it's dual channel DDR400. Whatever the case, it's still damn fast. However, remember what's out there for memory that could be realistically used with the PPC 970:

RDRAM? No.
QDR? No.
Single channel DDR? Unlikely.

Dual channel DDR400 is the most likely. ie. The same as the Intel "800 MHz" motherboards. Not 900 MHz because the memory isn't easily available in large volumes, and it would be very expensive.
( Last edited by Eug; Jun 10, 2003 at 10:34 AM. )
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Jun 10, 2003, 09:53 AM
 
Look at macsurfer.com this morning. 2 websites with G5 at WWDC rumors have the good old:

[Removed at the request of Apple Legal]

where the article uset to be.

If that doesn't spell *CONFIRMED!* then I don't know what does...

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
Nonsuch
Professional Poster
Join Date: Jan 2001
Location: Riverside IL, USA
Status: Offline
Reply With Quote
Jun 10, 2003, 10:31 AM
 
Originally posted by bewebste:
It's very true that Apple typically doesn't announce new hardware at WWDC, with a few minor exceptions. However, I wouldn't be surprised if they were to announce that they'll be using the 970 chips so that they could have some preliminary sessions telling the developers what to expect from the new hardware, even if the hardware itself doesn't ship for another month or three.
Keep in mind too that the usual "WWDC isn't for hardware" matra is further upset by the brouhaha with IDG. Apple supposedly isn't attending Macworld Create this July (or at the very least isn't providing a Stevenote) and has long been chafing at the obligation to make product announcements on IDG's schedule. I think we're going to start seeing WWDC become the "whatever big announcement Apple is ready to make" show, in lieu of MWNY (now MWB).
Find out just what any people will quietly submit to and you have found out the exact measure of injustice and wrong which will be imposed upon them.

-- Frederick Douglass, 1857
     
Eug  (op)
Clinically Insane
Join Date: Dec 2000
Location: Caught in a web of deceit.
Status: Offline
Reply With Quote
Jun 10, 2003, 11:08 AM
 
Originally posted by Nonsuch:
Keep in mind too that the usual "WWDC isn't for hardware" matra is further upset by the brouhaha with IDG. Apple supposedly isn't attending Macworld Create this July (or at the very least isn't providing a Stevenote) and has long been chafing at the obligation to make product announcements on IDG's schedule. I think we're going to start seeing WWDC become the "whatever big announcement Apple is ready to make" show, in lieu of MWNY (now MWB).
MWNY 2003:

"Apple Feature Presentation
Wednesday, July 16
9:30 am – 10:30 am

Apple has a long-standing tradition of delivering high-powered presentations about the world's most advanced operating system. Apple and Mac OS X will continue to drive innovation and remain the platform of choice for creative professionals. Come hear the latest announcements and developments on the creative community's best tools and practices.
"

Doesn't sound like a keynote though, and there doesn't seem to be enough time for a PPC 970 launch. But Apple will be there.

Look at macsurfer.com this morning. 2 websites with G5 at WWDC rumors have the good old:

[Removed at the request of Apple Legal]

where the article uset to be.

If that doesn't spell *CONFIRMED!* then I don't know what does...
I don't think that proves anything. It does suggest superficially that the PPC 970 will eventually be released, but the WWDC timing could be completely off.

Time will tell.

Anyways, here are my revised somewhat optimistic ballpark predictions for the PowerMac 970 launch:

Fast:
1.42 GHz PPC 970
256 MB DDR333 RAM
80 GB ATA100 hard drive
Radeon 9000 Pro 64 MB (on an AGP bus)
DVD/CD-RW combo drive

Faster:
Dual 1.6 GHz PPC 970
256 MB DDR400 RAM
100 GB ATA100 hard drive
Radeon 9000 Pro 64 MB (on an AGP bus)
DVD/CD-RW combo drive

Fastest:
Dual 1.8 GHz PPC 970
512 MB dual channel DDR400 RAM
120 GB ATA100 hard drive
Radeon 9700 Pro 128 MB (on an AGP bus)
4X SuperDrive

This is only assuming that IBM has good yields at 1.8 and plans on being somewhat aggressive with their speedbumps. Otherwise Apple might choose to give themselves some breathing room and for now have a top machine at 1.6 GHz. If the 1.6 is the top, however, then it may make more sense to have Panthwire loaded on this machine at release, to optimize speed.

But who knows...
     
 
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 05:42 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,