PDA

View Full Version : PowerPC


Pages : [1] 2 3

Morpheus
2004-07-18, 03:37
In the begining there was the chip known as GPUL aka PPC970. Next the GPUL 10S aka Altair aka PPC970FX. Next? An****s aka PPC9****!

Anyone care to speculate? :smokey:

Morpheus

hmurchison
2004-07-18, 04:21
Next up is the GRUL POWER5 derivative. SMT will be added amongst a few other new features.

Morpheus
2004-07-18, 04:42
Next up is the GRUL POWER5 derivative. SMT will be added amongst a few other new features.

Incorrect. Try again... Read this old NMR report (http://www.macedition.com/nmr/nmr_20020914.php) for some clues... But for sure, something else will be delievered around the end of the year to Cupertino...

Has the name Antares been mentioned somewhere?

Morpheus ;)

staph
2004-07-18, 07:47
With respect, I think this might be one of those situations which is designed for that great Australian phrase: "put up or shut up".

What are you trying to tell us?

Morpheus
2004-07-18, 08:25
With respect, I think this might be one of those situations which is designed for that great Australian phrase: "put up or shut up".

What are you trying to tell us?

Fair enough:

Antares is a dual core (two processors on single die) follow-on to Altair, using IBM's CMOS SOI10K technology with SOI (Silicon On Insulator) and copper bus with 10 layers of metal. Each core has a 1M L2 cache. L3 cache is not supported.

How is this for a start?

Morpheus :cool:

staph
2004-07-18, 09:02
Fair enough:

Antares is a dual core (two processors on single die) follow-on to Altair, using IBM's CMOS SOI10K technology with SOI (Silicon On Insulator) and copper bus with 10 layers of metal. Each core has a 1M L2 cache. L3 cache is not supported.

How is this for a start?

Morpheus :cool:

Thank you. That's much better.

Any thoughts on what core it's using (modified 970? stripped down P5 w/ Altivec, perhaps with SMT?)

Morpheus
2004-07-18, 09:17
It will be marketed as PowerPC 970MP. Take your own conclusions... :)

Morpheus

hmurchison
2004-07-18, 12:40
Incorrect. Try again... Read this old NMR report (http://www.macedition.com/nmr/nmr_20020914.php) for some clues... But for sure, something else will be delievered around the end of the year to Cupertino...

Has the name Antares been mentioned somewhere?

Morpheus ;)

Antares is the name of an music company

I think I'd trust an IBM pdf mentioning the GR-UL over a NMR report from 2002. Do you have any "real" information or have you not woken up from the matrix yet
:smokey:

The POWER5 is the GR the derivative GR-UL (ultra light or whatever)

please come with something more substantial than 2 yr old links or don't come at all.

Morpheus
2004-07-18, 18:27
Antares is the name of an music company

I think I'd trust an IBM pdf mentioning the GR-UL over a NMR report from 2002. Do you have any "real" information or have you not woken up from the matrix yet
:smokey:

The POWER5 is the GR the derivative GR-UL (ultra light or whatever)

please come with something more substantial than 2 yr old links or don't come at all.

Eh? It just happens that that 2 yo link is the first reference to GPUL out there. Anyway hmurchison, why would you give any credit to my words? There may be some around here these days (like Nick-de-Plum from ThinkSecret), that could put a word or two for me.

You want "real" info? You will have it. Does this board supports attachments? how?

Morpheus

hmurchison
2004-07-18, 19:20
I don't doubt your info but http://www.fz-juelich.de/zam/files/docs/vortraege/ibm-nutzung/hh_pwr4_arch.2up.pdf this link which is now dead used to show that the POWER5 is known as the "GR" and below it had "GR-UL" referenced to a 97x chip. Wouldn't this naming scheme seem to corrolate with how they handled the POWER4 and hence GP-UL? Can we expect Dual Core during 1H 2005? What about ondie memory controllers do you think that will make it in the POWER5 derivative? I'm just razzin' you for more info but I'm always keen on hearing any speculation. It's fun to me.

Morpheus
2004-07-18, 19:58
I don't doubt your info but http://www.fz-juelich.de/zam/files/docs/vortraege/ibm-nutzung/hh_pwr4_arch.2up.pdf this link which is now dead used to show that the POWER5 is known as the "GR" and below it had "GR-UL" referenced to a 97x chip. Wouldn't this naming scheme seem to corrolate with how they handled the POWER4 and hence GP-UL? Can we expect Dual Core during 1H 2005? What about ondie memory controllers do you think that will make it in the POWER5 derivative? I'm just razzin' you for more info but I'm always keen on hearing any speculation. It's fun to me.

hmurchison, GR is a different story...

Antares is Power4 based. There was a time when a dual-core GPUL was the only plan (as mentioned on the NMR linked above), although that changed for various reasons...

As for planned dates, Protos to Apple Aug/2004 and Oct/2004, Prod Jan/2005... as for Apple's plan, that I do not know...

Is it possible to post attachments here?

BTW, the link you refer to is live at http://www.fz-juelich.de/zam/files/docs/vortraege/ibm-nutzung/ibm-nutzung-0401/hh_pwr4_arch.2up.pdf
:)

Morpheus

Morpheus
2004-07-19, 01:47
Some more details:

"The IBM PowerPC 970MP RISC Microprocessor
is a dual-core, 64-bit implementation of the IBM
PowerPC ® family of reduced instruction set computer
(RISC) microprocessors that are based on the PowerPC
Architecture. This dual microprocessor, also called the
PowerPC 970MP, includes a Vector/SIMD facility which
supports high-bandwidth data processing and compute-intensive
operations. The 970MP is also designed to
support multiple system organizations, including desktop
and low-end server applications, up through 4-way SMP
configurations.
Note: The IBM PowerPC 970MP incorporates two complete
microprocessors on a single chip, along with some
common logic to connect these microprocessors to a system."

Morpheus

hmurchison
2004-07-19, 02:03
Morpheus- Interesting. Does this mean we see Quads next MWSF? Hmmm how would Apple position a Dual DC Powermac within the lineup?

$3499 Quad 2.5Ghz
$2799 Dual 2.8Ghz
$2399 Dual 2.5Ghz
$1999 Dual 2.0Ghz

All PCI Express of course.

FFL
2004-07-19, 02:25
Is it possible to post attachments here?
Morpheus

Well, I can, but I think that's just because I'm a Moderator.

Morpheus, check you PMs.

Morpheus
2004-07-19, 08:49
Well, I can, but I think that's just because I'm a Moderator.

Morpheus, check you PMs.

FFL,

Read and actioned.

Many thanks

Morpheus

FFL
2004-07-19, 09:07
Morpheus look pretty genuine to me....

very interesting stuff, thanks Morpheus

FFL

Barto
2004-07-19, 09:30
/Title (BLUELOGO.EPS)
/Creator (Adobe Illustrator\(TM\) 3.0)
/Metadata 734 0 R
/Author (CAPT. JAMES T. KIRK)

That's lines 1062-1065 of the PDF.

I'm just sayin'

sCreeD
2004-07-19, 09:34
"Clock Dithering (New feature for 970FX DD3.0, enhanced in 970MP)"

Hmm...

Screed

Barto
2004-07-19, 09:39
Of course someone at IBM's graphics department could have a sense of humor.
And line 5439 lists the creation date (2004-06-15) and CreatorTool (FrameMaker 7.1) and line 5441 lists the author as IBM Microelectronics.

Somebody willing to devote a lot of time and effort could have made this. So at this stage, as with almost all "leaked" images and documents, I can't tell (and nobody else can really - like with the 10.4 screenshots for example).

FFL
2004-07-19, 10:08
/Title (BLUELOGO.EPS)
/Creator (Adobe Illustrator\(TM\) 3.0)
/Metadata 734 0 R
/Author (CAPT. JAMES T. KIRK)

That's lines 1062-1065 of the PDF.

I'm just sayin'
That would seem to be referring to the IBM corporate logo, which is not something very easy to get in a EPS format. Although the Capt. Kirk thing seems kinda fun and free-spirited for an IBM employee (probably using a Mac in the graphics dept...)

I don't see that as a cred-killer. Glad to see skeptical investigation though!

FFL
2004-07-19, 10:10
Of course someone at IBM's graphics department could have a sense of humor.
And line 5439 lists the creation date (2004-06-15) and CreatorTool (FrameMaker 7.1) and line 5441 lists the author as IBM Microelectronics.

Somebody willing to devote a lot of time and effort could have made this. So at this stage, as with almost all "leaked" images and documents, I can't tell (and nobody else can really - like with the 10.4 screenshots for example).

For the record, Morpheus has shown me a bit more stuff, which strongly indicates he's for real.

So, what say we get on with discussing the material itself, rather than the source and the validity thereof? :smokey:

Henriok
2004-07-19, 13:06
I guess we won't se any integrated memory controller this time either. Good news anyway!

hmurchison
2004-07-19, 13:24
Good job Morpheus. I'm not worthy ;)

The pdf makes sense to me. Not because I'm an engineer but because It's very believable. If I was a IBM employee wanting to create a false document for fun..I think I would be a little bit more bold than creating a dual core datasheet. Hell I would have wrote some bogus cell processor info with made up stuff tossed in. This doc convinces me that Apple/IBM are definitely planning dual core...perhaps by next model revision.

Apple's nextgen Powermacs may all be dual core computers using only once socket. I'm sure this would allow them to make the mobo cheaper. However they could go crazy and go for dual sockets for dual cores. Option options.

Henriok
2004-07-19, 14:30
Will this 970MP be pin compatible with existing 970 processors? This information should be available in the PDF but the index doesn't reveal anything about it. It it is, chanses are pretty good that Apple quite easily could just put two of these on the existing mobos and pray to God that the liquid cooling will be sufficient.

Powerdoc
2004-07-19, 16:21
If this info is true, then we should not see hyperthreaded G5, but dual core G5.
This dual core G5 will be huge chip,even, bigger than the power4 (even if there is no L3 cache manager, there is two altivec unit : one for each core).

I day it's not impossible to see such a chip, but I am surprised that they do not create an hyperthreaded G5 who recquieres much less transistors (but who will be less performant either).

This G5 MP, will produce lot of heat. It may be possible to see one in a desktop with the 90 nm process, but we will never see them in an I mac or a powerbook, unless they will be fabbed on 65 nm process.

oldmacfan
2004-07-19, 16:38
Why am I only gettting 3 pages of the PDF?

Powerdoc
2004-07-19, 16:43
Why am I only gettting 3 pages of the PDF?

By showing us this PDF, the man broke his NDA, but as he is only showing the table of content, he do not broke it too much.
If he show the internal content, he will truly betray the companie from who he is working for. There is some secret infos here.

This is explanation one;

Explanation two, it's a fake, and the less he say, the less work it is, and the less mistakes he will do. but FFL think it's a real PDF file, so I will vote (and be happy to) for explanation one.

Morpheus
2004-07-19, 16:46
Many thanks to FFL to post the file. Do I now have your attention? :D

Die size? 13.23mm * 11.63mm Pin comp? yes

Morpheus
Guardian of Dreams

hmurchison
2004-07-19, 17:05
Powerdocs right. That's a big chips. Seems strange in the light that IBM has struggled with yields. Say we get dual core in the 1st half of 2005. Does that mean we get the POWER5 derivative in the second half?

curiousuburb
2004-07-20, 01:14
Antares is in the constellation Scorpio

As for planned dates, Protos to Apple Aug/2004 and Oct/2004, Prod Jan/2005... as for Apple's plan, that I do not know...

Can we get Carol to measure this CPU's Natal Clock cycles and give a horoscope?

"Scorpios are flirtatious in their communications with others and network well" = xGrid benefits?

iMeowbot
2004-07-20, 06:19
"Clock Dithering (New feature for 970FX DD3.0, enhanced in 970MP)"

Hmm...
That's part of PowerTune. They're using it to ramp between frequencies so that current surges are smoothed over.

Barto
2004-07-20, 07:50
Antares is in the constellation Scorpio



Can we get Carol to measure this CPU's Natal Clock cycles and give a horoscope?

"Scorpios are flirtatious in their communications with others and network well" = xGrid benefits?
I'm sitting back, waiting for Carol to chase you around with a switch now :D

jeannot
2004-07-20, 10:57
Very interesting, Morpheus.. :)
Can you tell us if this CPU is taped out ?

Morpheus
2004-07-21, 05:53
Very interesting, Morpheus.. :)
Can you tell us if this CPU is taped out ?

According to the
traditional definition (http://www.free-definition.com/Taped-out.html) , yes, it has :)

Morpheus

Snoopy
2004-07-21, 15:55
. . . This dual core G5 will be huge chip,even, bigger than the power4 (even if there is no L3 cache manager, there is two altivec unit : one for each core). . .

This G5 MP, will produce lot of heat. It may be possible to see one in a desktop with the 90 nm process, but we will never see them in an I mac or a powerbook, unless they will be fabbed on 65 nm process.

Maybe the water cooled 970FX is just a shakedown cruise for the upcoming 970MP. Apple may want early experience with water cooling in a less demanding implementation. It will be one less new technology for Apple when the G5 Quads are introduced.

Morpheus
2004-07-21, 16:30
Main features: (cont)

In order dispatch of up to five operations into distributed issue queue structure
Out of order issue of up to 10 operations into 10 execution pipelines
- Two load or store operations
- Two fixed-point register-register operations
- Two floating-point operations
- One branch operation
- One condition register operation
- One VMX permute operation
- One VMX ALU operation


(tbc)

Morpheus

Kurt
2004-07-21, 16:35
What are the yields going to be on this monster? If IBM is having a hard time with the 970fx how are they going to make a chip that is (or almost) twice as big? I would think that it would be definitely for a high end machine.

Morpheus, any idea what speeds they are planning?

hmurchison
2004-07-21, 16:50
What are the yields going to be on this monster? If IBM is having a hard time with the 970fx how are they going to make a chip that is (or almost) twice as big? I would think that it would be definitely for a high end machine.

Morpheus, any idea what speeds they are planning?

The problems IBM faced are likely due to 90nm transitioning and not size. If that is true then IBM and Apple is correct that 970fx CPU are constrained until Oct 2004 then it's reasonable to conclude that IBM fixed the issue prior to the end of summer and is working on ramping production. If indeed the 970MP is tape out this fall then mass production could commence in January/Feb. Anyone know how long the PPC takes from tapeout to mass prod?

Kurt
2004-07-21, 21:07
The problems IBM faced are likely due to 90nm transitioning and not size. If that is true then IBM and Apple is correct that 970fx CPU are constrained until Oct 2004 then it's reasonable to conclude that IBM fixed the issue prior to the end of summer and is working on ramping production. If indeed the 970MP is tape out this fall then mass production could commence in January/Feb. Anyone know how long the PPC takes from tapeout to mass prod?

I understand that a lot of the issues were due to a smaller process size but since every transistor has a certain probability of being faulty, the chances to make a good chip definitely increase as the number of transistors decrease. Therefore, a bigger (more transistors) chip is more likely to have a defect. If IBM is already having problem with the process, I would think that yields on dual core chips are going to be lower.

Morpheus
2004-07-22, 04:39
Morpheus, any idea what speeds they are planning?

Yes.

Morpheus

Morpheus
2004-07-22, 04:47
tpbd!

BTW, an interesting IBM Research report (http://domino.watson.ibm.com/library/cyberdig.nsf/papers/E6EBD3C859FB49F785256ED8006A3F4A/$File/rc23276.pdf) on the PM G5.

Morpheus

709
2004-07-22, 11:04
Yes. :D

DaveGee
2004-07-22, 11:11
I can vouch for my dear old brother Morpheus he has a direct line with the Oracle don't ya know. ;)

Dave

tomierna
2004-07-22, 13:41
I can also vouch for Morpheus. He's never given wrong info, to my knowledge.

frontline lamb
2004-07-22, 14:31
Another vouch for Morpheus. Just never seen him so forth coming in public however.

rickag
2004-07-22, 15:06
Morpheus

You mention CMOS SOI10KE, but nothing about strained silcon. Just curious, the current 970FX reportedly is using a strained silcon process, but as I understand it not SSDOI(strained silcon directly on insulator). Why your omission of strained silcon, is IBM leaving strained silcon tech behind, or is my understanding of this teminology/process badly misconstrued?

:confused:

DaveGee
2004-07-22, 15:59
Good to see you again FLL!

Dave

Brad
2004-07-22, 18:38
FWIW, I can concur that the above new members that "vouched" for Morpheus are unique based on their IPs.

hmurchison
2004-07-22, 18:59
The info sounds great but what are we to base these newfound numbers on. Should be be comparing each features with the spec sheet for the 970 processors? What we're missing here is the baseline.

Kurt
2004-07-22, 19:28
Yes.

Morpheus

What is that saying about all those out of work comedians? ;)

Please tell us what the speeds are going to be.

Also, what is the timeframe for a 970fx at 3GHz?

Thanks.

DaveGee
2004-07-22, 20:52
What is that saying about all those out of work comedians? ;)

Please tell us what the speeds are going to be.

Also, what is the timeframe for a 970fx at 3GHz?

Thanks.

Speeds?!?! You don't know much about CPU design (and neither do I) but I can say: What the speeds are 'on paper' and what they are when they come off the line usually are two totally different things.

Proof is in the pudding!

That being said I'd guestamate they are shooting for 2.5 to 3 GHz and anything else would be a bonus. After all, TWO cores BOTH running at 3GHz would be amazing and the fact is 2.5 would be amazing considering:

970: ~131 mm
970-fx: 96 mm (I think)
970-mp: ~154mm

2 cores each with their own 1 MB L2 cache!! Heck the initial gpul and even the fx only had 512k. I have a feeling the MP is gonna SCREAM even if it **only** clock to 2.5GHz.

I wish someone could invite 'programmer' over here I'd love to hear his input on these specs. I know I know - we shouldn't but...

Dave

NosferaDrew
2004-07-22, 21:03
I miss guys like Programmer when we have threads like these.
Interesting stuff nonetheless.

PhenixReborn
2004-07-22, 21:29
Morpheus

You mention CMOS SOI10KE, but nothing about strained silcon. Just curious, the current 970FX reportedly is using a strained silcon process, but as I understand it not SSDOI(strained silcon directly on insulator). Why your omission of strained silcon, is IBM leaving strained silcon tech behind, or is my understanding of this teminology/process badly misconstrued?

:confused:


Not to answer your question, but here's a quick summary of Strained Si:

By stretching the crystalline structure of the silicon, you can do two things.

The first is that you can match it's structure with other materials. This means you might have better luck with SSOI since the Insulator might not have the same crystalline orentation or cell spacing as the Silicon. By straining the silicon, you can come close to matching the insulator structure and thereby bond the two materials together. You have to bond them together to get a decent chip.

The second reason has to do with increasing the distance between the silicon atoms. When you increase the distance, you increase the probability that a mobile electron will move between two atoms without colliding with either of them. In macroworld terms, this means you have less resistance from the silicon which increases processor speed and reduces heat generation.

To quote from a Nanotechnology book large enough to bludgeon you with:

...MOSFETs with strained Si channels...exhibit superior performance. Leading chip manufacturers have decided to adopt strained silicon for high performance CMOS microporcessors

Does that follow with your understanding of Strained Silicon? If you want more, I'll write up the paragraph they give to strained Si. It's rather informative. :D

iMeowbot
2004-07-22, 22:05
2 cores each with their own 1 MB L2 cache!! Heck the initial gpul and even the fx only had 512k. I have a feeling the MP is gonna SCREAM even if it **only** clock to 2.5GHz.
Higher clocking might not be so far-fetched. It *is* two cores, and they're implementing PowerTune and independent caches, which hints at (but doesn't dictate) independent clocks and asynchronous glue. *If* they've gone partially async, propagation delays hold a lot less sway over what the clocked sections do. It also makes overall throughput much, much harder to guess at.

Eugene
2004-07-22, 23:36
I just hope Apple doesn't give it some silly label like "Shotgun" or "Double Barrel."

I also think the 970MP in no way implies quad G5s.

gsxrboy
2004-07-23, 02:39
Will powertune et all be able to throttle down individual cores?

Morpheus
2004-07-23, 04:58
Dave, Tom, Frontline! I'm starting to feel at home! Anyone else? Pitty there are no walls here... I guessed right, that some of us would be lurking around these boards...

Morpheus

Morpheus
2004-07-23, 05:04
Dave,

Make that 125, 66, 153

Morpheus

Morpheus
2004-07-23, 05:14
Will powertune et all be able to throttle down individual cores?

No

Morpheus

Morpheus
2004-07-23, 05:18
Small correction: Pin comp = false, of course...

Morpheus

LudwigVan
2004-07-23, 16:53
I see ThinkSecret (http://www.thinksecret.com/news/antares.html) is hot on the case now too.

hmurchison
2004-07-23, 17:11
I see ThinkSecret (http://www.thinksecret.com/news/antares.html) is hot on the case now too.


Methinks perhaps Morpheus and Thinksecret's "sources" are one in the same.

Bring on the Quads!!

Eugene
2004-07-23, 17:14
Bring on the Quads!!
A far premature specification, I think. 154mm^2 90nm core x 2 = you're going to need a bigger boat.

DMBand0026
2004-07-23, 17:15
I see ThinkSecret (http://www.thinksecret.com/news/antares.html) is hot on the case now too.

Haha. Most of that article made sense to me. You guys might as well be speaking Chinese, cause I don't understand 93% of this thread ;)

This is exciting though, I hope IBM can push the dual core stuff out the door before AMD and Intel.

DrGruv
2004-07-23, 17:39
What happened to the 980 Q3 of 2004?

I was expecting the 980 based on the power 5 in jan. 05.

PERHAPS, an IDEA...

NEW powerpc980 for the pro market?

New 970 chip for laptops and imac - throttled down to a lower power for powerbooks and imacs YET still having peformance from the dual core?

(just a thought) (no flames) What do you think? Plausible?

-mike

hmurchison
2004-07-23, 18:14
A far premature specification, I think. 154mm^2 90nm core x 2 = you're going to need a bigger boat.

Premature Specification indeed :)

However in light of Morpheus' posted pdf about heat dissipation it seems that the Dual 2.5 PM is over engineered going with liquid cooling. I believe the first G5s where about 120mm squared per proc. So the duals used 240mm of total space. Would it really be that hard to shoe horn in another 68mm total and keep them cool? Perhaps but with improved Powertune and SOI not to mention liquid cooling the idea doesn't seem as far fetched after first reflection.

onlooker.org
2004-07-23, 19:37
just saying hi! What the heck? Was this appleinsider.org once?

I like the new name better. No legacy BS.

Morpheus
2004-07-23, 19:55
No comment. :smokey:

May all your dreams be good dreams,

Morpheus
God of Dreams, Ruler of the Shadowlands

iMeowbot
2004-07-24, 09:26
Higher clocking might not be so far-fetched. It *is* two cores, and they're implementing PowerTune and independent caches, which hints at (but doesn't dictate) independent clocks and asynchronous glue. *If* they've gone partially async, propagation delays hold a lot less sway over what the clocked sections do. It also makes overall throughput much, much harder to guess at.
...and in the ThinkSecret version of the story, it does use the elastic bus from POWER4. This deals with memory clock skew by using handshaking, while traditional interfaces just clock everything down to the worst case delay. Not just better overall performance here, but the longer wire lengths it allows will simplify cooling problems and open up some possibilities for external interfaces. Not quite the same as async, but very similar benefits. Neat.

Moogs
2004-07-24, 16:51
I don't know about you guys but Dual-processor, Dual-core G5 would be the first computer Apple has released in a long time, that would actually make me want to sell an already kick-ass machine (Dual 2Gig G5) to get a newer model. That would likely be a quantum jump in performance, especially with the new stuff going into Tiger (AFAICT).

I aspire to the day when turning on my Mac will be as responsive as turning on my television. 5 second boot time, anyone?

CoreMac
2004-07-25, 10:05
I aspire to the day when turning on my Mac will be as responsive as turning on my television. 5 second boot time, anyone?

Sorry, but you'll need to sell that Dual Dual G5 and buy a Dual Quad 980 for that. :lol:

Spart
2004-07-26, 00:36
A question to those who might know:

Given a current dual 2.5 GHz G5 machine with one bus per processor vs. a dual core, single chip 2.5 GHz G5 with one bus per dual core processor, would the non-dual-core version be faster at some tasks because of the twin elastic busses? Besides differences caused by larger caches and whatnot...

hmurchison
2004-07-26, 04:02
A question to those who might know:

Given a current dual 2.5 GHz G5 machine with one bus per processor vs. a dual core, single chip 2.5 GHz G5 with one bus per dual core processor, would the non-dual-core version be faster at some tasks because of the twin elastic busses? Besides differences caused by larger caches and whatnot...

I don't think the Twin FSBs would make much of a difference. You still have Dual Channel PC3200 memory as the "real" bottleneck. 6.4GBps is nice but the 1250Mhz bus alone can handle 5GBps up/down. So the 1250 FSB only becomes a limitation when we hit more than 10GBps memory throughput which would be DDR2 667 in Dual Channel config (PC 5300)

Powerdoc
2004-07-26, 08:10
I don't think the Twin FSBs would make much of a difference. You still have Dual Channel PC3200 memory as the "real" bottleneck. 6.4GBps is nice but the 1250Mhz bus alone can handle 5GBps up/down. So the 1250 FSB only becomes a limitation when we hit more than 10GBps memory throughput which would be DDR2 667 in Dual Channel config (PC 5300)

Exactly the dual FSB make only sense if you have two separates chips : there is no penalty hit for sharing the same bus. A dual core chip has only one memory*bus, but the two core will be interconnect at full clock speed. Futhermore, the latency between the two cores, will be much smaller than the one of two separate chips.

It's nearly*sure that a 1,25 ghz bus, won't do much difference against a simple 1 ghz bus, with dual channel PC memory and it's 6,4 GB/sec. Apple did go for the 1,25 ghz bus with the 2,5 ghz bus, because there is a limited choice of ratio between the chip and the bus : 1/2, 1/3 (and smaller), and the use of 1/3 with a 2,5 GB/sec chip will induce a performant penalty. At 3 ghz the bus is clocked at 1 ghz, something sufficiant for exploiting at 100 % the memory bandwitch of the mobo.
I am ready to bet, that if Apple shipped a 3 ghz G5 on the current design (PPC 970 fx) , he will use the 1 ghz bus, and not the 1,5 ghz one.

staph
2004-07-26, 08:24
You'll all be glad to know that this just got us referenced at The Register (http://www.theregister.co.uk/2004/07/26/ibm_ppc970mp/)

Barto
2004-07-26, 09:00
Urgh. Getting referenced by TheReg for having inside information is like getting referenced by John Howard (or George Bush, or Tony Blair) for having Iraq intelligence.

alcimedes
2004-07-26, 09:54
yeah, but the BOFH makes up for any wrong doing.

:D

Powerdoc
2004-07-26, 10:59
Some people wonder why IBM won't produce directly an multithreaded bi core PPC 970 à la power 5.
I think that the main reason is the size. The core of the PPC 970 mp is certainly larger than the PPC 970 due to the deeper pipelining (and the more a chip is pipelined, the more transistors it recquieres). Adding multithreading will recquieres 25 % more transistors. At 90 nm the heat issue could become critical.
That's why multithreaded may be a feature that will appear when IBM switch for 65 nm process. :)

hmurchison
2004-07-26, 12:00
Some people wonder why IBM won't produce directly an multithreaded bi core PPC 970 à la power 5.
I think that the main reason is the size. The core of the PPC 970 mp is certainly larger than the PPC 970 due to the deeper pipelining (and the more a chip is pipelined, the more transistors it recquieres). Adding multithreading will recquieres 25 % more transistors. At 90 nm the heat issue could become critical.
That's why multithreaded may be a feature that will appear when IBM switch for 65 nm process. :)

Powerdoc that's what I'm thinking. I also think Apple will keep the ondie memory controller functions as well for the mythical 980. You have to figure, just doing rough estimates, that the inclusion of SMT, OMC and dual-cores in a POWER5 derivative would be 170mm+ die. Moving to 65nm should drop that back to 80-100mm yet we get all the "goodies".

ATS
2004-07-27, 11:55
eWeek has an article (http://www.eweek.com/article2/0,1759,1627893,00.asp) about the new dual-core PPC, which it says will be "available for testing purposes later this summer."

jl51
2004-07-29, 06:25
Why am I only gettting 3 pages of the PDF?
Where and how to get the pdf :confused:

gsxrboy
2004-08-18, 07:13
Almost a whole month and no new gossip?

Morpheus
2004-08-27, 02:50
Almost a whole month and no new gossip?
970GX

Morpheus

DrGruv
2004-08-27, 03:09
970GX

Morpheus


Boy's got the goods :)

come on --- spill it!!!!

What is the 970GX?

-mike

Henriok
2004-08-27, 04:36
970GXLet me guess... "nothing" more that 1 MB L2 cache? Can we expect speed bumps and som other goodies too? I hear that we can expect 970GX in about 6 months time, do you concur?

jl51
2004-08-27, 06:28
970GX

:) Welcome back Morpheus !

Is the 970GX the son of the 970FX as the 750GX is the son of the 750FX ?

A low power 970GX would be a good chip for embeded computing and laptops.

DrGruv
2004-08-27, 15:20
Well?

c'mon man drop the other shoe...

:lol:

kscherer
2004-08-27, 15:50
Fair enough:

Antares is a dual core (two processors on single die) follow-on to Altair, using IBM's CMOS SOI10K technology with SOI (Silicon On Insulator) and copper bus with 10 layers of metal. Each core has a 1M L2 cache. L3 cache is not supported.

How is this for a start?

Morpheus :cool:

Sounds to me like someone has been reading all of the news postings. Care to quote from any other articles? :no:

AirSluf
2004-08-27, 18:11
XXXXX

Henriok
2004-08-27, 18:43
Sounds to me like someone has been reading all of the news postings. Care to quote from any other articles? :no:Please check your dates. I can guarantee you that this thread was the first. Before all the articles.

FFL
2004-08-27, 19:12
Sounds to me like someone has been reading all of the news postings. Care to quote from any other articles? :no:
I can "confirm" that this thread was the first public mention of the 970MP. So could you, if you simply checked the date of the post you quoted, and compared it to the articles you refer to.
DOH!
"better to keep quiet, and appear a fool, than to speak up and remove all doubt"

Morpheus
2004-08-27, 20:03
Let me guess... "nothing" more that 1 MB L2 cache? Can we expect speed bumps and som other goodies too? I hear that we can expect 970GX in about 6 months time, do you concur?

and where did you hear that? ;-)

Morpheus

Henriok
2004-08-27, 20:31
and where did you hear that? ;-)You are not the only one with sources :)

windowsblowsass
2004-08-27, 22:04
macrumors s where he heard that

DaveGee
2004-08-27, 22:38
Sounds to me like someone has been reading all of the news postings. Care to quote from any other articles? :no:

As others have already started... Grand old Morpheus was the father of ALL the other forum posts and articles.... If it wasn't for him you'd still not know about the dual core stuff.

Not a great way to start off your posting life eh? :no: :err: :no:

Maybe you should kick back and read more before you make yourself look foolish again... Unless you have room for that other foot? :lol:

Dave

curiousuburb
2004-08-27, 22:54
Hey, maybe he's got a dual-core mouth. :p

futuretheory
2004-08-28, 01:05
Hey, maybe he's got a dual-core mouth. :p

:lol:

trismegistus
2004-08-28, 01:21
You are not the only one with sources :)

Did you get the info at AmiGBG 2004 (http://web.kicker.nu/amigbg/nyheter/nyheter.asp?sub=visanyhet&nid=30)?

Henriok
2004-08-28, 07:41
Did you get the info at AmiGBG 2004 (http://web.kicker.nu/amigbg/nyheter/nyheter.asp?sub=visanyhet&nid=30)?No, but i'm sure not to miss next year's exhibition if they talk about such things there. AmiGBG is in my home town after all.

Eugene
2004-08-28, 12:26
http://www.hp.com/united-states/designjet/index.html

OMG, 5 GHz G5!!!11one

Morpheus
2004-08-29, 04:49
Well?

c'mon man drop the other shoe...

:lol:

Well,... There is AntaresMP and there is Antares....

Morpheus

Morpheus
2004-08-29, 04:58
As others have already started... Grand old Morpheus was the father of ALL the other forum posts and articles.... If it wasn't for him you'd still not know about the dual core stuff.
Dave

Brother Dave... If they only new about everything else... mmm... I never posted at the "other place that shall not be mentioned here" - gosh... this reminds me of "the one that shall not be named" :-) - but I do remember you dropping a few hints about monikers never before heard...

Morpheus

DaveGee
2004-08-29, 08:18
Brother Dave... If they only new about everything else... mmm... I never posted at the "other place that shall not be mentioned here" - gosh... this reminds me of "the one that shall not be named" :-) - but I do remember you dropping a few hints about monikers never before heard... Morpheus

That oughta cause some heads to spin... Not quite as bad as Otto (Oh!) but close. :) :lol: as far as info-accessibility your hand has mine beat hands down, in fact your up cards can beat my whole hand. :D

Dave

DaveGee
2004-08-29, 08:26
Well,... There is AntaresMP and there is Antares....

What would IBM do with all of those MPs where one out of the two cores fails... (after all IIRC, IBM perfected such a process with it's POWER line of CPUs didn't it?) So, my guess would be:

AntaresMP - Dual Core
AntaresSP - Single Core

And if that's the case... the SP and MP would be available on/around the same time. ETA - 1Q 2005

Dave

DrGruv
2004-08-29, 14:41
Well,... There is AntaresMP and there is Antares....

Morpheus

Antares 970GX? - You do enjoy the game - :smokey: ok, deal me in...

Ok, lettme guess - (ante up please) a newly designed low power version of the 970 with many special features just for the new powerbook g5?

What's your bet? Want any cards? or do you wanna show your hand? :cool:

--------
did that other shoe drop? :cool:

Morpheus
2004-08-30, 06:52
What's your bet? Want any cards? or do you wanna show your hand? :cool:

--------
did that other shoe drop? :cool:


My hand shows: eCLipz

:smokey:
Morpheus

DaveGee
2004-08-30, 07:05
My hand shows: eCLipz

:smokey:
Morpheus

Interesting... here is something I dug up on eCLipz.

Article: http://www.midrangeserver.com/tfh/tfh110303-story01.html

[midway down the page you'll read]

----
That brings us to Power6 and a related project known as ECLipz. If you want to make IBM stop talking, start talking about ECLipz. A few months ago, I got my hands on some internal roadmaps from IBM about ECLipz, which, according to this documentation, is a Power6-generation project that was started in late 2001. The project name suggests a few things. First, "ipz" obviously refers to iSeries, pSeries, and zSeries, or so one might surmise. "ECL" is a little tougher to crack, but the "E" could be for eServer, the "C" could be for something like Consolidated, and the "L" could be for Linux. I showed this document to some IBM sources, and soon thereafter I got a denial (not attributable to anyone, mind you) that what this document clearly said--that the ECLipz processor, as Power6 was known, would be used across all non-Intel IBM servers--was not true, and that heads rolled because this roadmap said so.


I'm not so sure I believe this. But IBM's top brass (including Bill Zeitler, who heads Systems Group) has been quoted in the trade press recently as saying that IBM will maintain distinct platforms for mainframes and other IBM Power-based machines. The roadmap mentioning ECLipz is from sometime in 2003 (date unknown), so it is not that outdated. But for all I know, IBM started this project to consolidate the zSeries line into the Power line and it did not work. This would not be the first time IBM tried this. The Fort Knox and Future Systems projects that tried to consolidate IBM's mainframe and midrange lines decades ago failed, too. This is hard stuff, which is why vendors don't do it. So ECLipz could be alive or dead. Only IBM knows.
----

Hmmm... now this was written back in November 2003 and here we are at the end of summer 2004 and ECLipz pops up again... I'd have to say that ECLipz isn't dead yet. :) but what does it hold for us?

Dave

jl51
2004-08-30, 07:30
My hand shows: eCLipz
Morpheus
Is the 970GX the one which will cross the 3 Ghz boundary :confused:

windowsblowsass
2004-08-30, 07:43
he just fucking with our minds now, the sick bastard he gets off n our pain. have you no shame?!

AirSluf
2004-08-30, 11:49
XXXXX

DrGruv
2004-08-30, 12:50
My hand shows: eCLipz

:smokey:
Morpheus


Okay.... You bid an eClipz

I'll see your bid, I raise it.

'L' stand for laptop or low power? :smokey:

DaveGee
2004-08-30, 12:53
Dude, you have to do better than a veiled reference already out in the open from the Reg on 7 August 2003 (http://www.theregister.co.uk/2003/08/07/ibms_power5_to_hit_3ghz/). Just a snippet here:
[snip]
Yeah, I know. I'm a dikhed skeptic, but someone's gotta be. :p

AirSluf,

No need to be a 'dikhed skeptic' (using your words) when it comes to Morph... If he's still bringing up eCLipz there's a good reason for him to do so.

Most important, it means said project is very much alive, well and moving forward. The references I've found all seem to point to 2006/2007 for an ETA - Maybe, and I'm just guessing here... Has that ETA has since moved closer to 2005 perhaps even early 2005? This would be welcome news for the Power family but I'm still wondering how it connects together with the PPC family.

e = eServer
C = ??? (Converge)
L = Linux
ipz = i p z types of IBM servers.

Are lines that are drawn between the 'Power' and 'PowerPC' going to 'thin' or even disappear all together? That would be HUGE news but not possible since (to my layman's eyes) Power = multicore, BIG MAMA CACHE and expensive iron in implementation (as compared to desktop hardware anyway) where PPC has usually equaled single-core (about to change) BABY CACHE and Altivec.

Brother M... You know I love-ya but could ya help me out here? Am I getting warm (with the notion that PPC and Power are slowly becoming one)

Dave

stoo
2004-08-30, 16:05
Am I getting warm (with the notion that PPC and Power are slowly becoming one)

I'm sure I've heard this suggestion before. Probably in one of the PowerPC 970 articles/forum posts on ArsTechnica.

Henriok
2004-08-30, 16:20
I'm sure I've heard this suggestion before. Probably in one of the PowerPC 970 articles/forum posts on ArsTechnica.As far as IBM is concerned, POWER is PowerPC and vice versa. The destinction between the POWER and PowerPC ISAs seems to be purely academic since a while back. Everyting from POWER5 to PowerPC 405 through 970FX is labled "Power architecture" and I think that the future Cell will be too.

Mentioning eCLipz, Morpheus might have some new info on POWER6. I fail to see the immediate connection between POWER6 and Antares though.

Morpheus
2004-08-30, 18:04
:grumble:

ok, I guess it is time for some ground rules... some know me, most don't... maybe this will help:

1. I'm not regular, here or anywhere. May be here today, may only return next month... questions may be ununswered for a long time or even not answered at all if I decide to ignore them.

2. Don't like the style, move over to other threads, or if you wanna bitch, go ahead, please yourself (see 1.)

3. Just because I hint at something, does not mean that full details will follow. Or they may take a long time to come. That's the nature of the beast. Otherwise I would have to kill you all. Don't like it? See 2.

4. I like to play games. Mind games, not bluffing games. Those that don't, see 2.

5. There is no 5.

Morpheus

Gargoyle
2004-08-30, 18:09
http://www.hp.com/united-states/designjet/index.html

OMG, 5 GHz G5!!!11one

For some reason it always tickles me when someone does the !!!111one thing. :lol:

AirSluf
2004-08-30, 20:07
XXXXX

gsxrboy
2004-08-30, 23:12
By golly didnt Moki cause a whirlwind of activity when he mentioned gpul way back then :)

What ever happened to the man?

DrGruv
2004-08-30, 23:38
Morpheus

( just casual conversation during the game :smokey: )

with all the 970mp 970gx, 970fx, 970 et. al, what became of the powerpc 980 due q3 of 2004 based on the power 5?

My guess, is the 980 got pushed back to next summer with the 970mp coming in early '05. (for me please)

I am buying one computer that needs to last many years...

Go for a quad 970mp at (guesstimate $3499) or wait for the 980? :confused:
Would the 980 be significantly better than the 970mp? :confused:

The power5 has over 270 million transistors and the powerpc980 perhaps in the 90 million (or more range)

The 980mp (dual core) or quad would definitely kick some serious numbers.. :)

Jump in at the 970mp or wait for the 980? :\

thanx for the advice in advance...

-mike

(I'll take two cards) :cool:

http://atlanta.creativeloafing.com/2003-03-05/happs-1.jpg

Barto
2004-08-30, 23:39
Moki is the president of a software company which has a bundled software deal with Apple (Deimos Rising on the iMacs, eMacs and iBooks). I'm sure Moki wants to keep on Apple's good side. Notice how Moki never revealed what the GPUL was, making it meaningful only to those already "in the know."

ApplePI turned out to be real. (although Apple changed the acronym to API)

DorsalM was a proven hoaxer. (damn)

I miss Programmer and Eskimo. :(

Barto

Nick dePlume
2004-08-30, 23:42
The accuracy of Morpheus' information is not in question. Enough said.

-Nick

Playing games is fine. But games don't generate respect, confirmation does. I'm not calling anyone out, but I have seen the result of hi-jacked handles and hoaxers much more often than the real deal, even when admins supposedly verified IP's. Some like Moki were right on. Stuff happens. I now await history to write itself.

drewprops
2004-08-31, 00:42
When Nick speaks I listen.
And yes, this is like pre-blackout AI. Kind of honeybee-esque if you ask me. Pick up the phone Morpheus, get out of here before Agent Schiller shows up.

No joke. This dustup went bigtime quick.

gsxrboy
2004-08-31, 00:47
Moki is the president.... I'm sure Moki wants to keep on Apple's good side. Notice how Moki never revealed what the GPUL was....

Yeah I am up with who he is, just wondering where he got to, he has been AWOL on all the sites for a while, but like you say he is prob endeavouring to be a good boy :)

Jim S.
2004-09-13, 18:53
It's been two weeks so hopefully Morpheus has calmed down and will come back to enlighten us again. Some of us appreciate Morpheus and hopefully those who attack Morpheus will stay away this time.

More on eCLipz, please?

AirSluf
2004-09-13, 22:50
XXXXX

CoreMac
2004-09-13, 23:15
Please, Mr. Morpheus, may we have another serving of eCLipz? ;)

Morpheus
2004-09-14, 06:38
Please, Mr. Morpheus, may we have another serving of eCLipz? ;)

Power Everywhere, 6-6-6

Morpheus

Morpheus
2004-09-14, 06:46
mmm...

STI is commmmming...

Morpheus
:smokey:

Eugene
2004-09-14, 06:47
I'm coming! I'm coming!

jeannot
2004-09-14, 07:31
STI ? (http://www-1.ibm.com/businesscenter/venturedevelopment/us/en/featurearticle/gcl_xmlid/8649/nav_id/emerging)

trismegistus
2004-09-14, 08:41
IBM had already announced CELL based workstaion (http://www-306.ibm.com/chips/news/2004/0511_cell.html). Sony Corporation (Sony), Sony Computer Entertainment Inc. (SCEI) and IBM today announced plans to develop a digital content creation environment, the first computing application planned for the Cell* processor. The companies expect to build the first prototype Cell-based workstations in 4Q 2004.

DrGruv
2004-09-14, 09:15
Here's what I dug up...

http://www.midrangeserver.com/tfh/tfh111703-story01.html

rickag
2004-09-14, 09:56
Power Everywhere, 6-6-6

Morpheus

Much too cryptic for me to even venture a wild a## guess.
:confused:

Henriok
2004-09-14, 10:08
Much too cryptic for me to even venture a wild a## guess.
:confused:A raving wild guess. Power6 = 6 cores at 6 Ghz. One devil of a chip! 666 million transistors.

DrGruv
2004-09-14, 10:23
LINKED POWERPC'S?

Powerpc's linked to others to form any size super computer desired?

6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6-6

IBM used its own Self-Timed Interface (STI) backplane and Remote I/O (RIO) I/O bus technologies to attach disk and tape subsystems to the server. The STI backplane was developed for IBM's S/390 mainframes and adapted for use in the big SMP versions of the AS/400.

In the future--which could be any time over the next few years--IBM has a whole new set of backplane and I/O adapter technologies coming out, probably starting with the Power5 generation and rolling into the Power6 generation. IBM has made no secret that it likes the InfiniBand switched mesh architecture, not only because it has high bandwidth (10 Gbit/sec and 30 Gbit/sec on the roadmap) and low latency, but also because it allows memory-to-memory interconnections between processor nodes in a cluster or SMP. (In effect, InfiniBand turns a cluster into an SMP, with the latencies as low as they are.) InfiniBand looks like a decent replacement for RIO-2, but it is unclear if this is what IBM will do with the iSeries. http://www.midrangeserver.com/tfh/tfh111703-story01.html

6-6-6 powerpc everywhere :smokey:

AirSluf
2004-09-14, 16:03
......

CoreMac
2004-09-14, 16:08
Power Everywhere (http://www-1.ibm.com/technology/power/about.html)

CoreMac
2004-09-14, 16:13
Search Power Everywhere and 666 on Google and this pops up. :eek: The Fourth Reich (http://www.totse.com/en/conspiracy/institutional_analysis/thefourthreich172250.html)

DrGruv
2004-09-16, 13:10
Eclipz

L = This? New Linux Servers Based on Power5 http://www.eweek.com/article2/0,1759,1645044,00.asp

Henriok
2004-09-17, 03:10
Eclipz

L = This? New Linux Servers Based on Power5 http://www.eweek.com/article2/0,1759,1645044,00.aspSeems logical enought if you believe that Eclipz is an acronym. All I'm waiting for is an IBM PC-class computer based on PowerPC. 970 preferrably but a G3 will do since my vision includes rather thin clients. This Cell based prototype workstation that will come in Q4'04 is very interessting,. Someting tells me that it will be priced against the visualization machines from SGI, Sun and HP. Nothing cheap that will reach a larger crowd that is.

windowsblowsass
2004-09-17, 06:13
Search Power Everywhere and 666 on Google and this pops up. :eek: The Fourth Reich (http://www.totse.com/en/conspiracy/institutional_analysis/thefourthreich172250.html)

ITS SETTLED THEN MORPHEOUS IS A FASCIST :eek: :eek: :lol:

Jim S.
2004-09-18, 14:39
Thanks, Morpheus. Since you have not winked, we may not have unravelled the mystery. Could this new technology open the doors on a new class of devices, perhaps including next generation iPods? Could this lead us into a wonderful new world of set-top boxes and tablet devices that network with each other?

Could you share a few more words of enlightenment, please?

FallenFromTheTree
2004-09-19, 09:56
I was just checking out the IBM workstation hardware at their site and did not see any mention of dual cores.

I DID see that their top of the line workstations are in the $5,800 and up price ranges.
I was easily into $9000 for a dual CPU alone before things started getting interesting in the Z Pro series

I DO hope we will get some solid news to update our speculation on the Apple/IBM dual core PPC machines soon.
Lets also hope that Apple and IBM manage to beat Intel and AMD to
a showroom near you.

Antares
Quadra
970MP
970GX
980
Power 6

A rose by any other name will smell as sweet! :-)


http://www-132.ibm.com/webapp/wcs/stores/servlet/CategoryDisplay?lang=en_US&smrfnbr=2072488&catalogId=-840&categoryId=2436210&cgmenbr=1&langId=-1&cgrfnbr=2436210&cntrfnbr=1&storeId=1&cntry=840

These Power 4 servers make a loaded G5 dually sound surprisingly affordable.

http://www-1.ibm.com/servers/eserver/pseries/hardware/workstations/

Caught this off of AMD's home page as well.

http://www.amd.com/us-en/0,,3715_11787,00.html?redir=CPDC03

It would be very nice indeed if Apple provides a user with an ability to upgrade a single core to a dual core system,
at least within the 90nm ( liquid cooled) tower series.




The race is on!

CoreMac
2004-09-20, 09:11
Cell is coming! "It will change the world." http://www.gamesindustry.biz/content_page.php?aid=4447

Henriok
2004-09-20, 09:32
Cell is coming! "It will change the world." http://www.gamesindustry.biz/content_page.php?aid=4447Great news eveyone! My source tell me that the PS3 is even harder to code for than PS2 which was a nightmware from what I've heard. This massive parallellism thing is probably the way to go for the future but it will require some radical change in how people code. It will probably require some new fancy operating system too to handle all the threads or do some cool auto shredding of single threaded tasks. Everything about Cell seems brand spanking new. Hard to tell how well it will be recieved. Quite a gamble.

FallenFromTheTree
2004-09-20, 09:37
Well it looks like we'll have lots to talk about this spring of 2005.

I just hope Apple chooses wisely and continues to offer the best of the best.

CoreMac
2004-09-20, 10:25
Great news eveyone! My source tell me that the PS3 is even harder to code for than PS2 which was a nightmware from what I've heard. This massive parallellism thing is probably the way to go for the future but it will require some radical change in how people code. It will probably require some new fancy operating system too to handle all the threads or do some cool auto shredding of single threaded tasks. Everything about Cell seems brand spanking new. Hard to tell how well it will be recieved. Quite a gamble.

Gee, I wonder who would possibly be up to building such a radical OS? ;)

Tuttle
2004-09-21, 07:25
Great news eveyone! My source tell me that the PS3 is even harder to code for than PS2 which was a nightmware from what I've heard.

I also hear Macs cost twice as much as x86 machines, it must be true!

:no:

But what do I know? I just make PS2 games for a living...

Henriok
2004-09-21, 08:39
But what do I know? I just make PS2 games for a living...Yeah what _do_ you know? My source is making games for GC, Xbox and PS2. I trust him but he has his own oppiniums of course. If you think differently then by all means, spill your beans. If you do make PS2 games for a living you probably know a lot more then the average Joe, me or 98% of all AppleNova users.

FallenFromTheTree
2004-09-21, 09:22
I also hear Macs cost twice as much as x86 machines, it must be true!

:no:

But what do I know? I just make PS2 games for a living...


I've looked at a few workstation quality machines in the PeeCee lines
and found that the prices for a serious workstation are much higher
when you start installing dual processors and the best GPU cards.



The custom built machine I priced at Alienware with dual 2.4 GHz AMD 64's
was nearly $6000.

The price wasn't any cheaper custom building a dual processor IBM workstation.

You may be able to build a machine like this 30%
cheaper if you do it yourself,
but the general public usually wants a finished product with a factory warranty, especially when you're spending that kind of money.

If Apple can offer a dual core Quadra Tower for anything less than $4000 including the best available 256 0r 512MB GPU, I'm SOLD!

nowayout11
2004-09-21, 10:12
But it doesn't take 2 x86 processors to compete with the performance of the dual G5s the majority of the time, so dual x86 CPU configs are irrelevant.

FallenFromTheTree
2004-09-21, 10:27
I'm aware that there is no fair comparison between APPLES & LEMONS,
but the best hardware in either platform is still way up there.

Henriok
2004-09-23, 10:58
Please.. stay on topic! There's no point of starting yet anaother debate over what's cheaper, PCs or Macs. Since we are all Mac users they obviously offer us more value than PCs. Let's agree on that, and continue with the discussion at hand.

Jim S.
2004-09-23, 19:12
So right, Henriok! Here's hoping that Morpheus rejoins the conversation. Perhaps another hint at the meaning of 6-6-6?

FallenFromTheTree
2004-09-23, 23:15
Please.. stay on topic! There's no point of starting yet anaother debate over what's cheaper, PCs or Macs. Since we are all Mac users they obviously offer us more value than PCs. Let's agree on that, and continue with the discussion at hand.

Totally fine with that.

Let's see here 6 6 6

G6
3 GHz X 2 dual core
$6000 :err:

FallenFromTheTree
2004-09-24, 09:37
Or just maybe

June 6
Power 6
Dual core 3 + 3 =6

;)

rickag
2004-09-24, 12:45
going back to eCLipz

e = e servers
CL = Control Language
ipz = iSeries, pSeries, and zSeries

Come on Morpheus, need more clues, much too cryptic.

CoreMac
2004-09-24, 18:07
CL= CelL :D

Morpheus
2004-09-24, 18:15
Close guesses on 666... but not exactly...

one still missing: Power6, 6Ghz, ?

Morpheus

Morpheus
2004-09-24, 18:26
going back to eCLipz

e = e servers
CL = Control Language
ipz = iSeries, pSeries, and zSeries

Come on Morpheus, need more clues, much too cryptic.

Not really, you got it right... except that CL really means nothin... just to make it easy to pronounce it like eclipse...

and don't mix things up... let me make it clear:

eCLipz = 666

eCLipz != STI

STI = Cell

STI != Antares


Morpheus

windowsblowsass
2004-09-24, 20:44
Close guesses on 666... but not exactly...

one still missing: Power6, 6Ghz, ?

Morpheus
could it be 2006

curiousuburb
2004-09-24, 21:13
*waves hand*

Hexagonal. ;)

Jim S.
2004-09-25, 08:19
eCLipz (Power 6) is pretty far off in the distance. Antares (PowerPC 970MP) is coming soon. STI (Sony, Toshiba and IBM) cell processors are also coming soon. If I understand correctly, IBM has modified the PowerPC 970 to become the PowerPC 970MP which is the first cell processor and will be used in things like game consoles. Since Antares = STI, Apple is along for the ride which is a good thing from a cost and features viewpoint.

Since STI will have built-in communications and be a scalable design, one can hope that future iPods would use them and be able to stream to Airport Express and perhaps even link through the network to the music store.

As for that last 6, Power6 will be 65nm but 2006 seems more likely.

CoreMac
2004-09-25, 08:29
6 cores! :D

Morpheus
2004-09-25, 08:57
eCLipz (Power 6) is pretty far off in the distance. Antares (PowerPC 970MP) is coming soon. STI (Sony, Toshiba and IBM) cell processors are also coming soon. If I understand correctly, IBM has modified the PowerPC 970 to become the PowerPC 970MP which is the first cell processor and will be used in things like game consoles. Since Antares = STI, Apple is along for the ride which is a good thing from a cost and features viewpoint.

Since STI will have built-in communications and be a scalable design, one can hope that future iPods would use them and be able to stream to Airport Express and perhaps even link through the network to the music store.

As for that last 6, Power6 will be 65nm but 2006 seems more likely.

Not listening Jim...

STI != Antares
Antares != Cell

Cell on Apple hardware? Don't think so, but wouldn't have a clue... mmm.. Maybe, if Steve uses that blue phone quick...

As for the 6, you are correct, eCLipz chips - noticed the plural? more on that later - will be 65nm, but I was referring to 2006 for the last 6.

Morpheus

Jim S.
2004-09-25, 09:07
Thanks, Morpheus. I need some help with symbology since I'm not sure what "space-!" means in your equations. Anyone familiar with this?

Snoopy
2004-09-25, 09:19
Thanks, Morpheus. I need some help with symbology since I'm not sure what "space-!" means in your equations. Anyone familiar with this?

the symbol != means not equal. ! goes with the = sign, not the space.

Jim S.
2004-09-25, 09:30
Thanks, Snoopy. We should stay on topic but perhaps we need another thread to discuss what Apple might do with STI cell processors. I wonder about exclusivity in the Sony-Toshiba-IBM partnership to develop these processors that could keep Apple in the cold. I'm sure that Sony wants to create cool toys that Apple can't match.

CoreMac
2004-09-25, 09:31
It's not "space-!". It's "!=". I think it means does not equal.

edit:
Oops! Too late.

curiousuburb
2004-09-25, 10:46
It's not "space-!". It's "!=". I think it means does not equal.

edit:
Oops! Too late.
There is a character for "not equals" (Option =), but it doesn't render properly in some systems.
!= is an alternative way to represent the same thing.

AirSluf
2004-09-25, 14:24
......

DrGruv
2004-09-25, 15:00
Can Anyone Summarize What's Known To This Point? :\

trismegistus
2004-09-26, 00:56
Reliable info of POWER6 features had been posted to ArsTechnica Forum (http://episteme.arstechnica.com/eve/ubb.x?a=tpc&s=50009562&f=8300945231&m=9080959175&r=841005225631#841005225631) POWER6 features

Some features of POWER6 were revealed at NERSC user group meeting (http://www.nersc.gov/about/NUG/meeting_info/Jun04/). According to page 28 of this presentation (http://www.nersc.gov/about/NUG/meeting_info/Jun04/kramer.ppt), the specifications of POWER6 will be;
peak performance: 20 Gflops,
memory bandwidth: 3.75 bytes/flop,
hardware support for ViVA(Virtual Vector Architecture)-2.
If POWER6 will have two FPU and will perform 4 flops per cycle, the frequency will be 5GHz and the memory bandwidth will be 75 GB/s. Previously only a few details (http://www.cs.utexas.edu/users/cart/arch/fall03/KallaSlides.pdf) were known about POWER6.
The architecture will be based on POWER4/5,
It will be fabricated with 65nm process,
It will have "ultra high frequency cores".
Now, we have known the meaning of "ultra high frequency core". IBM will move to another "netburst" architecture?

It is interesting that the figure of ViVA-2 (p.32) in the presentation is same to Hitachi SR8000 (p.7 in this presentation (http://www.atip.org/HPC/HPCTW1/Proceedings/Hitachi.PDF)). The processor of SR8000 (http://www.hotchips.org/archive/hc11/hc11pres_pdf/hc99.s1.2.Kurihara.pdf) is custom PowerPC designed by Hitachi, and it has 160 FPR and "pseudo vector processing".

FallenFromTheTree
2004-09-26, 10:12
I'm sure that I've missed some important detail somewhere in these post, but the "CL' in eCLipz may simply mean

CLIENT ? No way to be sure.

OR that these eCLipz CPU blocks may be of some modular design
allowing for easy user upgrades.

Which would be very cool!

Sometimes I begin to think that "ignorance is bliss"
when considering to buy an Apple product.

While we know that things will ALWAYS be better in the future,
the question remains as to WHEN that technology has reached a
reasonably solid plateau where your investment will stand the test of time.

For the advanced home/hobbie user like me,
spending $2000-$6000 on a computer system has literally become nerve wracking when you look less than a year into the future.

You may be ready to upgrade your system NOW, but doing so
with any haste will have you kicking yourself for letting your
money burn your pockets (also like me).

It all comes down to needs vs. wants I guess.

Anything I get NOW would be way better than what I have,
but before I spend upwards of $3400 on a new Tower, I want to be certain
that I'm not making a rushed & foolish purchase.

Will the current single core motherboards and CPU connectivity
support a plug and play dual core conversion?
That would be really nice to know!

If Apple does not provide for major CPU upgrade compatability
within either the 90nm or future 65 nm arena everything you buy NOW
will have serious limits to expansion.

I honestly wish that Apple would at least publish some idea of where they are headed just for peace of mind.

????????




:err: :err: :err: :err:

silvergun
2004-09-26, 18:49
power everywhere 6 6 6?

I reckon it's either 6 ghz Triple Core or Power 6 Derived Triple Core. Linked with the rumoured triple core Xbox 2 chip?
Triple core PPC 97(6)?
Morpheus?

dglow
2004-09-26, 21:05
I'm reading that the intended meaning for '6 6 6' is Power 6, running at 6 Ghz, arriving in 2006. Is that right, Morpheus?

Morpheus
2004-09-27, 04:30
I'm reading that the intended meaning for '6 6 6' is Power 6, running at 6 Ghz, arriving in 2006. Is that right, Morpheus?

Correct, although I think it will probably be 6 6 7...

eCLipz is a lot more than Power6. It is the program that will take "Power Everywhere", from the embedded market, up to mainframe computing... core projects go by names like P6, Z6, P6L and P6UL... but that is not all.

Kick-starting those Dreams,

Morpheus

Henriok
2004-09-27, 04:46
P6, Z6, P6L and P6UL... but that is not all.
P6 = Power6 in pSeries
Z6 = Power6 in zSeries
P6L = L as in Linux or Light, in OpenPower perhaps?
P6UL = Power6 Ultra Light (analogous to what we have previously seen in GP-UL (970) and GR-UL (not yet revealed)). Possibly for Mac ans blade use use down the line. Very nice.

Yes.. my dreams are kick-started and blowing on all cylinders :)
Thanks Morpheus.

Question: Will this be four flavours of the same processor or four different processors of a common architechture?

CoreMac
2004-09-27, 16:34
Correct, although I think it will probably be 6 6 7...


The revolution always seems to be 2 (no, make that 3) years away. ;)

DaveGee
2004-09-28, 04:45
The revolution always seems to be 2 (no, make that 3) years away. ;)

Heh... One of the downsides (especially if your into instant gratification) to following the preaching of Brother Morpheus... Much of what he speaks of (while the truth) is usually six months to a year (or more) out. Also when it comes to Apple specific plans, the crystal ball he uses tends to be a little less clear. Not unreasonable given the timeframes we're talking about.

The satisfaction for you is in 6 to 12 months time when '667' rumors begin to make the rounds by the rest of the rumorologists you'll be able to post knowing little 'wink' messages in the forums (saying to yourself - well here it comes, finally!) as well as posting 'shaking-head' messages who think it's a joke. :) - Something that I had the pleasure of doing in the year prior to the G5 being released.

Dave

CoreMac
2004-09-28, 08:58
The satisfaction for you is in 6 to 12 months time when '667' rumors begin to make the rounds by the rest of the rumorologists you'll be able to post knowing little 'wink' messages in the forums (saying to yourself - well here it comes, finally!) as well as posting 'shaking-head' messages who think it's a joke. :) - Something that I had the pleasure of doing in the year prior to the G5 being released.

Dave

6 7 8. Maybe I already know something. ;)

:no:

DaveGee
2004-09-28, 14:15
6 7 8. Maybe I already know something. ;)

:no:

I doubt it very much. :lol: :\ :grumble: :no:

CoreMac
2004-09-28, 16:36
Just yankin' your chain, Dave. Still I can't help but muse about our lot in life. When these chips finally ship, Morpheus will be teasing us with the next generation. As I said, the revolution is always 2 years away. Will the cycle never end??!! Of course, I hope not. I'm a true addict.

Morpheus
2004-09-28, 16:49
Heh... One of the downsides (especially if your into instant gratification) to following the preaching of Brother Morpheus... Much of what he speaks of (while the truth) is usually six months to a year (or more) out. Also when it comes to Apple specific plans, the crystal ball he uses tends to be a little less clear. Not unreasonable given the timeframes we're talking about.

The satisfaction for you is in 6 to 12 months time when '667' rumors begin to make the rounds by the rest of the rumorologists you'll be able to post knowing little 'wink' messages in the forums (saying to yourself - well here it comes, finally!) as well as posting 'shaking-head' messages who think it's a joke. :) - Something that I had the pleasure of doing in the year prior to the G5 being released.

Dave

Brother Dave... Yes, you are correct like always, my insights into the Mothership's plans are nothing without our beloved Diablo Numero Uno... Where is he, anyway?

If I remember correctly, you were the one that alerted the Moneastry that someone by the name of Moki had aluded to the name GPUL in a place that I believe is not polite to mention here, meaning someone else outside the Sacred Walls had some knowledge of it. You dropped some "winks" into Moki's threads at the time.. It took about what? About six months to one year before what was common knowledge of the chosen ones was knowkn to the greather world? Same with eCLipz. Did you savage any artifacts from the ruins of the Great Place?

If only there were some safe rooms around here...

Enough of memories...

Morpheus

CoreMac
2004-09-28, 18:48
Memories! Misty, water colored memories of the insiders we once were!

Yep. I was there, too. ;)

Morpheus
2004-09-28, 19:43
Memories! Misty, water colored memories of the insiders we once were!

Yep. I was there, too. ;)

CoreMac: Who r u? Give me a hint, Brother.

Morpheus

CoreMac
2004-09-28, 22:26
I was (rips off mask)... moki! No, I was a just minor comic character in the play. No one would remember poor little me. Moki was a major player and Dave appeared to be a rising star. The again, perhaps I just hadn't read NMR that week and he had. ;) Towards the end of the play, an unfriendly ghost appeared. I don't remember you there, and I now get the feeling that you and Dave are talking about a different play and theater entirely. Somewhere secret? Sounds like it might have been more interesting. :D Sorry for the confusion.

Bucolic Old Sir Henry
2004-09-29, 05:00
That's enough of the group hug, lads, tell me more about the 970GX. A single core version of Antares. Enhanced Powertune, lower power? Scale to higher clock speeds? When?

Pip pip!

DaveGee
2004-09-29, 06:58
I now get the feeling that you and Dave are talking about a different play and theater entirely. Somewhere secret? Sounds like it might have been more interesting. :D Sorry for the confusion.

I'm afraid you're right.. The play house we once had was quite off the beaten path and we never had public performances.

Brother M. D is still around.... you should be able catch him via chat.

Dave

Brad
2004-09-29, 07:14
:) If you guys are talking about the "moki" that posted on AI, that was just Andrew "el presidente" Welch of Ambrosia Software. He never had any real inside information. At best, he may have received some good tips that he played off, but that's all.

DaveGee
2004-09-29, 07:22
:) If you guys are talking about the "moki" that posted on AI, that was just Andrew "el presidente" Welch of Ambrosia Software. He never had any real inside information. At best, he may have received some good tips that he played off, but that's all.

Not true, he was the first (to the best of my knowledge) to call out GPUL in any public forum. I remember this quite well... The reason I remember it was due to the fact that I had known about it some time before Moki's first post and was happy to hear that word of the CPU was beginning to rise to the surface.

CoreMac
2004-09-29, 07:32
I'm afraid you're right.. The play house we once had was quite off the beaten path and we never had public performances.

Dave

Was there any nudity involved in this play? :lol: :devil:

Brad
2004-09-29, 07:40
All I'll say is that we've sent correspondence back and forth and that he himself was never the recipient of any big information from Apple, IBM, et al. What he got was through another party. Is it inconceivable that someone like yourself with inside info just decided one day to slip it out to him? Anywho, it's all old old business that is long under the table. I'll leave you guys to your devices...

skid_boy_99
2004-09-29, 10:03
From http://www.freescale.com/webapp/sps/site/overview.jsp?code=DRPPCDUALCORE:

MPC7448 Processor
The MPC7448 represents another performance leap in Freescale's discrete processor line. It provides more than 1.5GHz processing power, increased L2 cache (1MB), and a higher-speed bus than its predecessors. It is also pin-for-pin and software compatible with the MPC74xx processors - an easy drop-in to those existing designs. But the MPC7448 keeps the power low, able to consume less than 10 Watts running at 1.4 GHz.


Rock 'n' roll?

DaveGee
2004-09-29, 16:34
All I'll say is that we've sent correspondence back and forth and that he himself was never the recipient of any big information from Apple, IBM, et al. What he got was through another party. Is it inconceivable that someone like yourself with inside info just decided one day to slip it out to him? Anywho, it's all old old business that is long under the table. I'll leave you guys to your devices...

Oh okay, I misunderstood your definition of 'insider info'. To me insider info is non-public info that came to me from someone that I know I can trust and not just info that was sent to me directly from a specific source and not some 'crazed lunatic rumor-fabricator'. :D

Dave

silvergun
2004-09-29, 17:15
Look around 1/4 of the way down this article

http://www.theinquirer.net/?article=18759

On the other hand, the dual-core IBM POWER5 was off to a good start performance-wise a few months ago, however the MCM-based systems at midrange and high-end which should show the platform's full potential will only arrive by yearend. At the same time, much of the POWER5 core architecture is expected to come over into the G5 family, supposedly in the dual-core PPC 970MP chip, then move to the 90 nm process in the POWER5+, which should be available around the same time as Montecito - late next year.[U]

curiousuburb
2004-09-29, 17:25
From http://www.freescale.com/webapp/sps/site/overview.jsp?code=DRPPCDUALCORE:



Rock 'n' roll?
Freescale reports also note that the 7448 isn't expected to even sample until mid '05

oldmacfan
2004-09-29, 20:59
Freescale reports also note that the 7448 isn't expected to even sample until mid '05
Which report is that in, I can't seem to find that sampling information on the 7448.

Is it just me or is there a reason Apple should use the 7448?
When I look at the three chips, I see the MPC8641D and MPC8641 as the chips to go to. What benefit, other than a quick speed fix (if any increase in speed is available), does the 7448 have?

It's 90nm?
It uses less power?
So, big honkin deal. If Apple thinks it is going to woo people with a mediocre new chip, then they are truly just treading water.

Henriok
2004-09-30, 02:55
Which report is that in, I can't seem to find that sampling information on the 7448.
Almost everywhare. The regular news sources cites this time frame, but nothing is stated in Freescale's press release.

Is it just me or is there a reason Apple should use the 7448?It's pin compatible to previous G4s and it wil be available at least half a year before 8641 and 8641D. It will probably be a bit cheaper to produce and implement too.

It's 90nm?
Yes. All the new procesors is 90 nm, SOI.

It uses less power?
Signifcantly. 7448 will use 10 W @ 1.4 GHz compared to ~20W for 7447.
The dual core 8641D will use 15-25W typical at 1.5 GHz. Not much.
If Apple thinks it is going to woo people with a mediocre new chip, then they are truly just treading water.For someone with som many questions, you have a lot of oppinions.

oldmacfan
2004-09-30, 12:44
Almost everywhare. The regular news sources cites this time frame, but nothing is stated in Freescale's press release.

It's pin compatible to previous G4s and it wil be available at least half a year before 8641 and 8641D. It will probably be a bit cheaper to produce and implement too.


Yes. All the new procesors is 90 nm, SOI.


Signifcantly. 7448 will use 10 W @ 1.4 GHz compared to ~20W for 7447.
The dual core 8641D will use 15-25W typical at 1.5 GHz. Not much.
For someone with som many questions, you have a lot of oppinions.


You missed the point of my post. I wasn't asking if the 7448 was 90nm or used less power, that was stated in the article. What I was asking for was a compelling reason Apple should use this chip.

What I did want to see was something that verified the 7448 not sampling till 2005.

Please read alittle better next time.

curiousuburb
2004-09-30, 12:53
You missed the point of my post. I wasn't asking if the 7448 was 90nm or used less power, that was stated in the article. What I was asking for was a compelling reason Apple should use this chip.

What I did want to see was something that verified the 7448 not sampling till 2005.

Please read alittle better next time.
The "Freescale chip details (http://forums.applenova.com/showthread.php?t=2013)" thread cites an eetimes article which cites the mid '05 sample date.

Also posted in this thread (http://forums.applenova.com/showthread.php?t=1966)

Save your stones about "read better next time". House looks pretty glassy.

oldmacfan
2004-09-30, 16:16
Save your stones about "read better next time". House looks pretty glassy.

If someone is going to completely misread something I write, I will use my right of free speach to comment about it. :)

If I comment on something someone writes and completely misread it, then I expect them to do the same. :)

This is not an in person conversation, sometimes things are lost in the conversion (thought - text) some times people need to reread things so they will better understand what the author is trying to convey.

Henriok
2004-10-01, 02:53
You missed the point of my post. I wasn't asking if the 7448 was 90nm or used less power, that was stated in the article. What I was asking for was a compelling reason Apple should use this chip.I don't think I missed miss your point it was to elaborate around why Apple should use 7448 over 8641/8641D.

7448 is pin compatible with existing G4s so Apple won't need to redesign the chip sets in eMac, iBook or PowerBook. The 8641/8641D is not pin compatible and has a lot of stuff on die so Apple would have to do a completely new chipset to use it. That takes time and money.

I guess that the 7448 will be cheaper to buy since it probably is smaller than 8641 (it's certainly smaller than 8641D) and it's a proven design. It will be cheaper to implement since Apple can use existing chipsets and drivers. But the SoC designs of 8641/D will be cheaper in the long run since Apple won't have to worry about the companion chips like PCIe bridges, Memory and Ethernet controllers.

The 7448 is less power hungry than the 8641/8641D solution so a machine would need less cooling and have longer battery life. That might be an illusion though since we're not taking the larger and more complex chipset that the 7448 need into account.

The 7448 will be out at least half a year before the other two chips, and I think that's a good reason why to use it at least as an interim solution.

All these answers was in my original entry but I guess you missed that.
If you wanted a comparision between 7448 and 970FX I would have answeded it pretty much the same way.

I cincerely hope that Apple will put the 8641/8641D processors to good use. Even if they are a year (or more) away they would rock i portables or entry level desktops. Even if IBM gets the G5 straight and produces a low power version I still believe that the 8641D will be the processor I'd like to see in a potable Mac.

Commodus
2004-10-04, 21:31
The "whither 8641?" question is a good one, because it would be really hard to say what Apple's long-term PowerBook plans are. If the G5 doesn't show up until 2006, the 8641 is a possibility. But if not, it would almost seem silly of Apple to use it. You wouldn't want to have a mainboard which is only useful for one revision, unless it was passed off to the iBook.

I imagine that Apple will have a PowerBook G5 sometime between July and September of 2005, even if shipping isn't right away.

Morpheus
2004-10-08, 07:14
I imagine that Apple will have a PowerBook G5 sometime between July and September of 2005, even if shipping isn't right away.

I certainly hope by January... How does low-power 1.6 and 1.8Ghz (maybe even a 2Ghz) sound like?

Morpheus

709
2004-10-08, 07:18
:eek:

Seriously?

That would certainly explain the extra PowerBook #s in 10.3.5....

Wow. If this happens I will be *VERY* pleased, as I'm planning on getting the next Rev, G4 or G5.

Morpheus
2004-10-08, 08:20
Sure. It is in Apple's hands though...

... and shouldn't be long till we see Altair-based 2.7Ghz PMs...

Morpheus

onlyafterdark
2004-10-08, 14:58
:drool:

Snoopy
2004-10-08, 17:35
... and shouldn't be long till we see Altair-based 2.7Ghz PMs...

Morpheus

Is that like before the G5 PowerBooks?

DMBand0026
2004-10-08, 17:46
Probably not. Not that I know much about what's going on in this thread, but judging based on typical update cycles we'll see the PBs before the next PM revision.

oldmacfan
2004-10-09, 10:52
Here is a little tidbit to ponder in this discussion.

http://news.zdnet.com/2100-9584_22-5402114.html

AirSluf
2004-10-09, 11:19
......

hobold
2004-10-19, 04:18
You might be interested to learn that Apple has recently re-hired Keith Diefendorff, a processor architect. According to PowerPC lore, he was the driving force behind AltiVec and the "Max" project, later known as the MPC7400 processor, and marketed by Apple as "G4" with "Velocity Engine".

For more fun, try to get a Freescale rep to tell you anything about possible AltiVec support in the e700 core. ;)

I have even fewer hard facts about IBM's future plans. But I am lead to believing that both simultaneous multithreading and on-chip memory controllers are things that will eventually migrate from Power5 to 'lighter' chip variants that Apple could use. As far as I can tell, though, it could be quite a while before these things surface (maybe two years?).

Admittedly I didn't offer much information here. Just telling you that you should not focus so much on IBM that you lose track of Apple and Freescale. :)

AirSluf
2004-10-19, 17:10
......

Snoopy
2004-10-19, 18:35
I've heard that the Power series chips are made to higher standards, for greater reliability in those high priced servers. Also, the package may be larger to handle power dissipation. I'd be surprised if there was no lite version of the Power 5. IBM uses the lite versions in their Blade Center systems too.

AirSluf
2004-10-19, 20:12
......

Snoopy
2004-10-20, 12:37
Here are comments about the Power5 that suggest it likely will not have a vector unit, which would make it unsuitable as a workstation class CPU. Other comments suggest additional reasons a separate lite version would be needed for Apple. One thing seems certain from this article, any Power5 lite would be dual core only, no single core version.

http://arstechnica.com/articles/paedia/cpu/POWER5.ars

AirSluf
2004-10-20, 16:21
......

Snoopy
2004-10-20, 22:01
. . . His list of difficulties a Power5-lite would face seem VERY serious compared to the differences between a Power4 and 970. . .


I didn't see such a list in the discussion of a Power5 lite. The first four paragraphs in this section do not discuss any difficulties, and end saying: "it's pretty clear to me that if IBM is working on a workstation-class POWER5 derivative, then it'll look a lot like today's POWER5, but with smaller caches, no on-die switch, no on-die L3 interface, etc.."

The next paragraph appears to address a Power5 lite, but the issues with SMT applies to the Power5 as well. Quoting the paragraph: "It would be tricky to make such a chip work well in SMT mode, though, because of POWER5's aforementioned reliance on AIX for what amounts to processor-level execution resource allocation. But the fact that IBM is supporting Linux on POWER5 suggests that they can and will make public enough of the documentation for the interface to the dynamic resource logic for OS developers to tune their process scheduling code accordingly."

In any case, this does not appear to be an insurmountable problem, and would have to be solve even if Apple used the Power5. The only thing close to being a problem with a Power5 lite is in the last paragraph of this section. Quoting this paragraph: "The other factor to consider is that a POWER5 "lite" wouldn't have the same cache hierarchy as POWER5, and as I mentioned above this memory hierarchy accounts for a large part of POWER5's performance. So the POWER5 derivative's performance – especially in SMT mode – would be significantly constrained by the redesigned cache hierarchy and reduced memory bandwidth."

As I read this paragraph, a Power5 derivative's performance would be constrained compared with the Power 5, just as we would expect. Neither does the 970's performance match the Power4's. A point is made that SMT performance will suffer especially. So here we have the strongest hint of a real problem with a Power5 lite. Yet, this statement is true only if SMT in a lite version of the chip is left unchanged from the Power5, and if there is a problem I would expect IBM to tinker with the design to solve it.

. . . But it seems that Power5-lite cores as Hannibal would envision them would seem castrated compared to the full version, something not true in the Power4/970 core relationship.


The way IBM and Apple envision the Power5 lite cores may be different from the way Hannibal envisions them.

Morpheus
2004-10-21, 06:41
Some here may be interested in following the Power/PowerPC Architecture Forum (http://www-106.ibm.com/developerworks/forums/dw_forum.jsp?forum=385&cat=5)... You never know what you can find there...

Those same ones may also be interested in reading the Power Architecture Community Newsletter (http://www-1.ibm.com/technology/power/newsletter/october2004/newsletter.html)

Morpheus

Rhumgod
2004-10-21, 09:13
Morpheus, are you Mr Macphisto over at AI?

Snoopy
2004-10-21, 10:38
Thanks for the links Morpheus. I had already read some of the newsletter, but went back to see what I missed. I didn't find anything, but then I'm not too sharp much of the time. I had not seen the forum before, and noticed a little discussion on Power5 SMT. It did not cover the OS connection mentioned in the Arstechnica article, however.

If there is something in particular we should be looking at, you need to be a little more direct, at least with me. I can miss things that are right in front of my face.

micmoo
2004-10-21, 19:46
Does anyone have a guess (or better) on what chip/technologies we will see in the next gen G5 we will see this spring? I am planning a large (in my world) purchase of all new photoshop workstations in April, and am chomping at the bit for news on the machines that will/could be shipping at that time. Thanks in advance for any feedback.

micmoo

AirSluf
2004-10-21, 22:48
......

UnixMac
2004-10-22, 00:03
The $100,000 question I have is... if Apple goes to a dual core CPU... will they then have dual dual cores?? or will they simply have one "chip" per machine operating as a dual processor?

Effectively, a dual core vs two separate single core.... any difference?

DMBand0026
2004-10-22, 00:06
Big difference. A lot of the problems associated with multi processor machines is the lag in shuttling data between the two processors. A dual core design essentially eliminates this. However, I do think we'll eventually see a dual dual core machine, but not right off the bat.

hobold
2004-10-22, 04:17
I expect that dual core chips will show up in the high end first, so there will be quad processor machines right from the start. I have no solid evidence, though.

This sort of quad machine is relatively easy to build when you already have mainboards and chipsets for (single core) dual CPU systems. The dual core chips are not necessarily a drop in replacement, but there is not much of a technical challenge except inside the chips themselves.

Dual core chips will make dual machines a lot cheaper as well, but I can't see Apple offer a dual iMac; that would eat too much into tower sales. Unless, of course, the towers were quads. :)

DMBand0026
2004-10-22, 07:05
My question is, does the current architecture of the logic boards in the Power Macs have the bandwidth to be able to swing what is essentially 4 processors? Won't there be bottlenecks?

UnixMac
2004-10-22, 08:44
My question is, does the current architecture of the logic boards in the Power Macs have the bandwidth to be able to swing what is essentially 4 processors? Won't there be bottlenecks?

that makes a lot of sense. I see Quad (dual dual) way.. down the road...

hobold
2004-10-22, 11:14
Memory bandwidth is a hard limit, no doubt about it. But not every task is bottlenecked by it. Remember the times when two 1.4GHz G4 processors were fed from a single 1.333 GB/sec front side bus? Now we are at 2.5GHz G5 processors sharing 6.4 GB/sec RAM bandwidth, that's 4.8 times the bandwidth per processor. Going to four CPU cores would still be a reasonably balanced system, especially considering that the CPUs can communicate with each other much faster than they can communicate with RAM.

DMBand0026
2004-10-22, 11:17
Good to know. I wasn't sure exactly as I know very little about processors and things of that nature. I was just throwing an idea out there. Thanks for the clarification.

oldmacfan
2004-10-22, 12:09
Is the band width 6.4 per CPU or total?

hobold
2004-10-22, 14:53
6.4GB/s is the theoretical peak throughput of dual channel DDR400 SDRAM (also known as "PC-3200"). The front side bus of the G5 is actually faster than that for the 2.0GHz and 2.5GHz models. The G5 is slightly unusual in that FSB speed scales up with clock speed.

Furthermore, the processors in a dual G5 do not share a front side bus. Instead, each CPU has a dedicated communication path to the north bridge/memory controller.

So, in a strict sense, both is true:

1. Each CPU has at least 6.4GB/sec bandwith to the rest of the computer (i.e. RAM, I/O devices, the partner CPU).

2. There is at most a total of 6.4GB/sec main memory bandwidth that both CPUs have to share.

The 'excess' bandwidth of the CPUs is still useful, e.g. when one CPU is pushing data to the graphics card with AGP 'fast writes', while the other CPU is streaming data to/from main memory.

Morpheus
2004-10-23, 00:37
Morpheus, are you Mr Macphisto over at AI?

Morpheus is Morpheus... and I don't do AI.

Morpheus

DMBand0026
2004-10-23, 11:01
Morpheus is Morpheus... and I don't do AI.

Morpheus

This guy gets better everyday ;)

Rhumgod
2004-10-23, 20:13
Morpheus is Morpheus... and I don't do AI.

Just checking. ;) Just that what you are saying kinda mimics what Mr MacPhisto over at AI said too.

wizard69
2004-10-24, 13:06
Morpheus; glad to see you are staying away from that artficial intelligence. ;)

Since you do seem to be wired in a bit to what is happening at IBM have you heard anything about the 970MP having three cores as opposed to two? This come up due to rumors of MS using 6 cores on their new xBox contained in two chips. I can't see the wisdom in IBM building both two and three core variants of the the 970.

As to feasability I believe that it woud be completely doable considering how small the 970 core is. With AMD and Intel going dual core with their big processors, it might still be possible to end up with a much smaller die with three 970 based processors. If rumors are correct Intels dual offering could be huge by todays standards (huge as in die size).

So the question is how solid is the information on the 970MP being dual core?

Dave


Morpheus is Morpheus... and I don't do AI.

Morpheus

saumur85
2004-10-25, 00:58
Based on what I've read in this forum, I am very interested in the direction that both Apple and IBM will take with the new processors.

Morpheus,
First, I wanted to ask a relatively simple question, from the information that other members and yourself have looked into, Would you say that IBM is doing a remarkably better job with microprocessors than Moto did in the 90's?

Second, the 6-6-6 deal sounds remarkably like the scalable architecture of the new h.264 codec that Apple is supporting. Would that be a reasonable assumption?

Morpheus
2004-10-25, 04:45
Morpheus;

...

So the question is how solid is the information on the 970MP being dual core?

Dave

How solid is a diamond?

Morpheus

Morpheus
2004-10-25, 04:49
Based on what I've read in this forum, I am very interested in the direction that both Apple and IBM will take with the new processors.

Morpheus,
First, I wanted to ask a relatively simple question, from the information that other members and yourself have looked into, Would you say that IBM is doing a remarkably better job with microprocessors than Moto did in the 90's?

Second, the 6-6-6 deal sounds remarkably like the scalable architecture of the new h.264 codec that Apple is supporting. Would that be a reasonable assumption?

1. Even you can guess my answer.

2. Que?

Morpheus

saumur85
2004-10-25, 09:30
1. Even you can guess my answer.

2. Que?

Morpheus

On the second point, I meant that H.264 can be scaled from 3G Cell phones to HD audio and video. Similar to the Power6 being used in everything from consumer desktops to mainframes.

wizard69
2004-10-25, 18:43
Up to a couple of weeks ago I would have said nothing is better. Then I found out that there are guys making materials harder thab diamond. Rather neat to think about what the world will look like in 10 ore 20 years, when everything around you is made out of materials that did not exists 20 years ago.

Dave


How solid is a diamond?

Morpheus

hobold
2004-10-26, 03:28
The idea of a dual core G5 is really the best frame to fit all existing puzzle pieces into. (I can't say "I can confirm it", because I have less credibility than Morpheus, but IMHO anything other than a dual core with memory controller would be a surprise now.)

And who makes better processors clearly depends on the definition of "better". For example the 970FX has a worse power consumption than the 7447A, even at the same low clock speed. So you will have to qualify your question a bit more.

BTW, I am beginning to believe that some sort of "AltiVec 2" project is underway. I don't know what it will be and when we'll see the results, but lately there are many puzzle pieces pointing this way.

AirSluf
2004-10-26, 11:38
......

micmoo
2004-10-26, 11:40
As far as the next gen PowerMac Towers go, (January I'd gues) Any educated guesses on what we may see in those boxes? 970FX at 2.8 probably?

AirSluf
2004-10-26, 20:50
......

Henriok
2004-10-29, 02:55
The PowerPage as recieved word that Apple will be releasing a quad processor Xserve in January-March (http://www.powerpage.org/cgi-bin/WebObjects/powerpage.woa/wa/story?newsID=12539). They will use two dual core G5s according to them.

I don't believe this report for a second for many reasons. Apple always ups the PowerMac before Xserve and the 970MP isn't ready by far, to name two.

Sounds like MOSR is their source...

micmoo
2004-10-29, 10:58
Is the 970MP that far from ready? Am I in deep delusion in the hopes of upgrading all our Macs in my printshop by May '05 with new quad processor Macs or at least single chip dual core Mp's???

Xaqtly
2004-10-29, 16:48
I don't believe this report for a second for many reasons. Apple always ups the PowerMac before Xserve and the 970MP isn't ready by far, to name two.

A lot of people thought Apple would never, ever put a G5 in the iMac before the Powerbook. I tend to think Apple will put whatever chip into whatever box as long as it's ready.

wizard69
2004-10-30, 03:28
I don't believe this report for a second for many reasons. Apple always ups the PowerMac before Xserve and the 970MP isn't ready by far, to name two.

Why so negative? First off the Xserve would be an ideal platform to ramp production of these chips on. So I don't see Xserve being an issue.
AS far as the 970MP why not in January. Lets face it we have heard peep from IBM so they must be up to something. It is well known that the industry is going dual core, IBM and Apple would love to be there first with production machines.


Sounds like MOSR is their source...