IBM at 90nm by Q4?

[MacRumors]

An Infoworld story (http://www.infoworld.com/article/03/09/24/HN90nanometer_1.html?networking) from last week notes that IBM is actively working on 90nm chips and actually plans on shipping these chips before long:

IBM is currently sampling 90 nanometer chips using SOI, and plans to ship those chips in the fourth quarter, said Scott Sykes, an IBM spokesman.

Current PowerPC 970s utilize a 130nm process. While no specific chips were referenced in the Infoworld article, a 90nm chip would bring higher clockspeeds and lower heat production.

Meanwhile, an unverified report (http://www.macrumors.com/pages/2003/09/20030929111636.shtml) that appeared on Page 2 (http://page2.macrumors.com) claims to have more details, but also claims a later ship date for the 90nm chips.

[sabbath999]

Good!

[Ambrose Chapel]

hopefully this will push PBs into G5-land way before the end of '04. Steve announced Desktop G5s at WWDC 03, maybe he'll announce the PowerBook G5 at WWDC 04?? Here's hoping

[idkew]

PowerBooks?




POWERBOOKS??

[Mr.Hey]

Hearing progress updates such as these warms my heart. Apple and IBM makes for a fatal combination for Apple's competitors. Go Apple! :D

[Lancetx]

Originally posted by Ambrose Chapel
hopefully this will push PBs into G5-land way before the end of '04. Steve announced Desktop G5s at WWDC 03, maybe he'll announce the PowerBook G5 at WWDC 04?? Here's hoping

Yep, here's hoping we see them not only in PowerBooks but also in iMacs too by next summer.

[Rocketman]

Originally posted by Macrumors
An Infoworld story (http://www.infoworld.com/article/03/09/24/HN90nanometer_1.html?networking) from last week notes that IBM is actively working on 90nm chips and actually plans on shipping these chips before long:


Looks like the year of the Powerbook is about to get a "leetle" bit longer :)
2003-2004.

Can you even imagine the astonishment (shock and awe) when a G5 Powerbook comes out and unix and wintel geeks start jumping ship to OSX? I cannot.

Rocketman

[etoiles]

yay, the all new powerbook rumors season has begun !
:p

[FlamDrag]

How nice it is to hear of progress! While I'd love for them to be in poweroboks, I'd be just as happy to see faster PowerMacs at a steady clip.

[dongmin]

Originally posted by Ambrose Chapel
hopefully this will push PBs into G5-land way before the end of '04. Steve announced Desktop G5s at WWDC 03, maybe he'll announce the PowerBook G5 at WWDC 04?? Here's hoping

Wouldn't get your hope up just yet. We still don't know what kind of heat the 90 nano process 970s give off.

This is according to the page 2 rumor:An anonymous/unverified report claims that Apple has already been seeded sample 90nm PowerPC 970s this month. Reportedly, they run from 1.8GHz to 3.2GHz and heat production is significant improved, with a 2GHz at only 34W, down from 57W. The 3.2GHz chip reportedly at 71W. 34W seems way high. At the same voltage, a 1.6 ghz 970 (90 nano) would give off 27W, too high for a laptop I presume. Of course, this is only a page 2 rumor so we don't want to read too much into it. And there are many other factors that come into play.

It struck me that a 2ghz 970 (130 nano) gives off 57W. Seems awfully high. Isn't that in the Pentium 4 range?

[isgoed]

So what will ibm make at 90nm? A Power5? I think it is a little bit too optimistic to already speak of production 970's.

And the article is about 90nm in general. It said that intel does not use silicon on insulator (SOI). That is probably causing the heat. And AMD is not making a 90nm before Q2 2004. Surely those AMD fans are waiting fore those high clocked athlon FX's

[WackyM]

I look forward to my next powerbook purchase in a year or two :D Right now I'm still waiting for my 15 1.25 to arrive... weeeeeeee!!!!!

[Samir 3.0]

I hope for an UberPower Book early next year...

We'll see

[manu chao]

I won't buy a G5 Powerbook until VirtualPC (if ever???) runs on a G5. And since dual G4s (e.g. @1.25Ghz) seem to be faster with MP-aware, non-G5-optimised programs than single G5s (e.g. @1.6Ghz), I am still hoping for something like a dual 1.33 Ghz 17'' Powerbook.

[Tim Flynn]

Xilinx ( a FPGA) vendor is all ready shipping FPGAs built upon 90nm design.
The chips are fabbed by IBM.
So a 90nm 970 in the near future sounds reasonable.
Normally for a CMOS design the power calculations (used to be ) are simple.
Basically power was linear with frequency and squared with voltage. But with the smaller geometeries, leakage current (power) has become a bigger factor. Early information for 90nm showed that the leakage current (power) was about half of the total.
So not all electrons are used to find aliens or fold a protein, a big part is used to heat the room :D

But, I'm sure IBM has some tricks up their sleeves ;)

[sharky2313]

very good news indeed. I like rumors of aggressive G5 development. This seriously shows us how pathetic Moto has been and developing and extending the G4 for the Mac.

[IJ Reilly]

The article cited also reports that Motorola expects to ship 90nm microprocessors in Q4, meaning we might be seeing faster/cooler G4s in the same time-frame. IMO, these would seem more likely to find their way into PowerBooks then the G5 at this point. Not to burst anyone's bubble, but it may be too soon to hope for G5 Mac-everythings any time soon. If IBM comes up with faster/cooler G5s, they're almost certainly destined for Apple's high-end products.

[bluedalmatian]

Personally I'd rather see a G5 Xserve before a Powerbook.

at present the g4 PBs are doing very well but i've a feeling that the Xserve is flagging a little

[Rincewind42]

Originally posted by dongmin
This is according to the page 2 rumor: 34W seems way high. At the same voltage, a 1.6 ghz 970 (90 nano) would give off 27W, too high for a laptop I presume. Of course, this is only a page 2 rumor so we don't want to read too much into it. And there are many other factors that come into play.

If you ran a 2Ghz part at 1.6ghz without reducing the core voltage, then yes you get about 27 watts. But dropping the core voltage also drops the maximum speed while reducing power usage a lot more. If you reduce the core voltage as well, then you could shed another 8 watts down to 19w. Probably still a little high compared to the 7457 though.

It struck me that a 2ghz 970 (130 nano) gives off 57W. Seems awfully high. Isn't that in the Pentium 4 range?

If you compare to a 2 Ghz P4, then I think you are seeing similar power usage (the primary factor in power usage is relative to the transistor count and size, and I wouldn't be surprised if the 970 & P4 had a similar transistor count). But of course, a 2Ghz 970 absolutely smokes a 2Ghz P4 so I don't think the comparison is particularly fair :D.

[Powerbook G5]

Man, it would suck if a G5 PowerBook were to come out after finally getting a new 15"! Ah well, I love this to death and don't need a G5 anyway. I wonder what they will look like, though. :)

[GUSTO]

So correct me if I am wrong;)

These new chip are ready THIS YEAR!? and will be shipped THIS YEAR!? so we could see these new chips in new powemacs??? aswel as powerbooks???

I am going to buy a dual G5 this week, but now with this new info should I hold off for these new G5 chips?????

And if these chips get into new powermacs when do you think they will see the light of day?

[IJ Reilly]

Another argument against G5 PBs any time soon: Apple has never jumped processor families in their portables without changing the form factor. The current 12 and 17 inch PBs form factors are less then one year old, and the 15 inch is brand new. By necessity and/or marketing, a G5 PB will have a new form factor -- and it's just too early in the product cycle for this to happen within the next 3-6 months.

My most optimistic guess for a G5 PB: next summer. Potentially, the G5 could creep into the consumer line (iMac) by the end of next year, but I'd call that very doubtful.

[Raman1970]

Okay.. can someone shed a little bit of light.... in not that savy on processor sizes..

So if I understand correctly.. a 130 nm processor is for desktop and 90 nm is for portables?

But then some mention that the 90 nm may find their way into powermac G5s.

Raman

[Rincewind42]

Originally posted by Raman1970
Okay.. can someone shed a little bit of light.... in not that savy on processor sizes..

So if I understand correctly.. a 130 nm processor is for desktop and 90 nm is for portables?

But then some mention that the 90 nm may find their way into powermac G5s.

Raman

Process size has no direct influence on what targets the chip is used for, insofar as a smaller process can certainly be used for the same tasks as a larger process. The advantages of a smaller process are less power, heat and a smaller package. Those advantages do tend to allow you to use chips in places they weren't usable in before.

[grahamtriggs]

It isn't clear whether the IBM spokesman meant Q4 actual or financial year... if it was the financial year, that would put it around the Q2 2004 / summer 3Ghz timeframe.

Although a nice speed bump for San Francisco would be nice... if the current 2Ghz is running around 50W, and the new 3.2Ghz will run at 70W, then there is probably some room for bumping the current generation of G5s...

[kris251069]

Originally posted by bluedalmatian
Personally I'd rather see a G5 Xserve before a Powerbook.

at present the g4 PBs are doing very well but i've a feeling that the Xserve is flagging a little

I agree with you

[Falleron]

And ready in time for the first revision of the Powermacs. Great, I am buying a new Powermac in january.

[csimmons]

Honestly, I think it would not be very wise to put a G5 in a iMac. I think the pro line and the consumer line should be kept separated; i.e., G5's in the Power Mac, Power Book and Xserve, G4's in the iMac, iBook and eMac. That way, there won't be too much of a canibalization in sales, and there's a very distinctive product line, like it used to be back in the day.

Just my opinion.

[stingerman]

The 7457 in the iMac with 2MB L3 cache or no L3 cache but 400MHz effective FSB is more than enough for the iMac along with a ATI 9600 GPU.

But it would be nice to get a headless desktop version of the PowerMac G5 for a typical corporate user who doesn't need all the features of the G5 but would prefer it over an iMac.

[ima_pseudonym]

Apple *has* switched processors without changing form factors. The original G3 powerbook was in the 3400 case.


Originally posted by IJ Reilly
Another argument against G5 PBs any time soon: Apple has never jumped processor families in their portables without changing the form factor. The current 12 and 17 inch PBs form factors are less then one year old, and the 15 inch is brand new. By necessity and/or marketing, a G5 PB will have a new form factor -- and it's just too early in the product cycle for this to happen within the next 3-6 months.

My most optimistic guess for a G5 PB: next summer. Potentially, the G5 could creep into the consumer line (iMac) by the end of next year, but I'd call that very doubtful.

[mmmbop]

Originally posted by grahamtriggs
Although a nice speed bump for San Francisco would be nice... if the current 2Ghz is running around 50W, and the new 3.2Ghz will run at 70W, then there is probably some room for bumping the current generation of G5s...

Or even expanding the range of dual processor machines, not just on the top model.

[rice_web]

And besides, the PowerBook enclosure right now is just about perfect.

Anyhow, I wouldn't be surprised to see another PowerBook revision this year, even so close to the last update. Frankly, I bet we'll see iBooks move to G4s in October, and PowerBooks move to G5s in November or December, leaving a PowerMac update for January.

Terribly optimistic, yes, but also entirely possible. Apple has seen their laptop marketshare skyrocket, and a move like this would certainly help that along.

[kenaustus]

I think Steve J is going to go for speed any where he can get it. He's squeezing as much as possible out of Panther and I can't see him sitting around on a faster chip for very long.

If that's the case then he will go for a speed bump in the PM as soon as it is available, probably around Feb 04. I would bet on 2.4 Ghz to 2.6 Ghz. If you look at the new G5 case it is hard to believe that they were not engineered to go a hell of a lot faster than 2 Ghz.

If the 90 nm chips allow for use in the iMac then I think that will also be announced, but at lower speeds than the PM. With the PM moving out the door at a good clip the next task will be to establish the same level of excitement in the iMac range, and that means a G5.

A G5 PB will also be brought into production as soon as it is ready, regardless of how long the G4 models have been on the market. By the end of the year the excitement of an Al case for the 15" will have dropped significantly and, let's face it, we'll all be wanting a G5 PB when it's announced.

The introduction of IBM into the mix with the G5 allows Apple to move away from a small speed bump every year to a situation where they can move up performance significantly every 6 months. The real challenge will be for the users to get used to Macs moving up in the world so fast that they will want to but a new one faster than ever before.

[singletrack]

Originally posted by csimmons
Honestly, I think it would not be very wise to put a G5 in a iMac. I think the pro line and the consumer line should be kept separated; i.e., G5's in the Power Mac, Power Book and Xserve, G4's in the iMac, iBook and eMac. That way, there won't be too much of a canibalization in sales, and there's a very distinctive product line, like it used to be back in the day.

Just my opinion.

A sale is a sale.

If Apple can increase market share by selling a kick-ass iMac and still make a profit doing it, then that's a good thing. You can argue that there's a higher profit margin on the G5 desktop but if they sell n-times more iMacs because they have a G5 processor and are as fast as consumer PCs then what do Apple care about how they got that profit.

I also don't know which day in the past you're talking about but there's always been G3 iMacs at the same time as G3 PowerMacs, and G4 iMacs at the same time as G4 PowerMacs. Crippling computers just to make them less powerful than the Pro line just cripples sales.

[stingerman]

It's not only Apple that is pushing for speed increases, but IBM's engineers are as well. Imagine the feeling of competition and pride that now exists among both the IBM and Apple engineers. For the first time in a long time, both firms can strike back at their competitors and win. For the first time in a long time, they can go to engineering conferences and trade shows and hold their heads a little higher than the other guys.

IBM invested a lot in their microprocessor plant and they see that their efforts will only increase production demand from Apple. IBM is doing everything they can to win more of Apple's business and help grow Apple's G5 appeal.

[FlamDrag]

It's also possible that we won't see a revision to the PM line, but an addition to the top end.

Keep the current 3 models and add a dual 2.5 (ish) model to the top end at a higher price while lowering (ever so slightly) the cost of the single 1.6 and 1.8.

[3.1416]

Originally posted by csimmons
Honestly, I think it would not be very wise to put a G5 in a iMac. I think the pro line and the consumer line should be kept separated; i.e., G5's in the Power Mac, Power Book and Xserve, G4's in the iMac, iBook and eMac.

Better idea: single G5s in the iMac (and PB if possible), and dual G5s in the towers. The G4 will not be remotely competitive a year from now unless Motorola pulls several rabbits out of a hat. Apple needs to move away from it as soon as possible.

[dho]

SWEEEEEEEEET!

[Mac User Canada]

WOW!!

I'm so glad that the new 23" G6 PowerPad with built-in RAID is shipping next week. http://forums.macrumors.com/images/icons/icon10.gif

I'm really curious to see just how they managed to get the holo emitters working in such a small package. http://forums.macrumors.com/images/icons/icon12.gif

But does anyone really believe that OS M fully supports temporal peripherals. http://forums.macrumors.com/images/icons/icon6.gif

Oh, wait, which multiverse is this?

MUC

[greenstork]

Originally posted by manu chao
I won't buy a G5 Powerbook until VirtualPC (if ever???) runs on a G5. And since dual G4s (e.g. @1.25Ghz) seem to be faster with MP-aware, non-G5-optimised programs than single G5s (e.g. @1.6Ghz), I am still hoping for something like a dual 1.33 Ghz 17'' Powerbook.

A PC emulator already runs on a G5. It's not Virtual PC but it's just as good. The emulator is called WinTel and is put out by an open source company called OpenOSX. It's G5 optimized to run all the Windows operating systems, check it out:

WinTel 1.0.1 (http://www.openosx.com/wintel/index.html)

[eric67]

Originally posted by dongmin

It struck me that a 2ghz 970 (130 nano) gives off 57W. Seems awfully high. Isn't that in the Pentium 4 range?

euh.... not really
current P4 are releasing around 75W and +, and the next generation to come (Prescott) at 2.8GHz is generating...103W!!!!!
so who is speaking about PPC970 being to hot????

[ShadowHunter]

Originally posted by eric67
euh.... not really
current P4 are releasing around 75W and +, and the next generation to come (Prescott) at 2.8GHz is generating...103W!!!!!
so who is speaking about PPC970 being to hot????

No joke!! My first impression was that the G5 was kinda warm, but take a glance at the competition and it is ice cold...

I think Sears is going to start carrying Pentium4 ovens - cooks food in half the time!

[greenstork]

I'll preface my response by saying that I think a computer only needs to be so fast unless you are doing video rendering or photoshop work all day. That said, if IBM and Apple want to truly compete for the fastest compter, they are facing some tough competition from Intel and others, and very soon.

Word on the street has Intel cranking out 5-7 Ghz chips with 2MB of L2 Cache and a 4000Mhz frontside bus speed (http://www.theinquirer.net/?article=11785) as soon as the end of next year. If Apple and IBM want to compete with this, they need to get cracking. I would argue however, that this is just too much speed and completely unnecessary.

It makes sense that a chip upgrade will take place by the end of 2003, beginning of 2004 since we can't expect the G5 to jump directly from 2.0 Ghz right to 3.0 Ghz by next summer. There should be something in between that time and if the 90nm process is in place and shipping by the end of 2003, it fits in perfectly with Apple's chips speed increase projections.

I think and hope consumers will look beyond this need for raw speed and value lower power, quieter computers. I say this only now that we have the G5 and would have never said this last year. I think above 2.0 or 3.0 Ghz PPC, we just don't need more speed, I'd rather have ultra-power-efficiency instead. Bring back the silent cube and make way for G5 Powerbooks. If you can fit a 2 GHz chip into a Powerbook, who really needs more than this, especially in light of the fact that we will soon be using 64-bit applications, providing an enormous speed boost that is completely independant of clock speed.

[eric67]

Originally posted by GUSTO
So correct me if I am wrong;)

These new chip are ready THIS YEAR!? and will be shipped THIS YEAR!? so we could see these new chips in new powemacs??? aswel as powerbooks???

I am going to buy a dual G5 this week, but now with this new info should I hold off for these new G5 chips?????

And if these chips get into new powermacs when do you think they will see the light of day?

this news is not new.
It was released few month ago by macbidouille.com (www.hardmac.com for the english version). they were also claiming 90nm by the end of this year; but in your computer not before Jan 2004; or June 2004. Actually this is what Steve said again during Apple Expo Paris...3GHz by June next year (but then Dual 3GHz, it's really going to smoke P4 Prescott)

[dongmin]

Originally posted by greenstork
A PC emulator already runs on a G5. It's not Virtual PC but it's just as good. The emulator is called WinTel and is put out by an open source company called OpenOSX. It's G5 optimized to run all the Windows operating systems, check it out:

WinTel 1.0.1 (http://www.openosx.com/wintel/index.html)

The 1 million dollar question: is it faster than VPC?

[eric67]

Originally posted by FlamDrag
It's also possible that we won't see a revision to the PM line, but an addition to the top end.

Keep the current 3 models and add a dual 2.5 (ish) model to the top end at a higher price while lowering (ever so slightly) the cost of the single 1.6 and 1.8.

no that's not the way to go. 1.8GHz single, then all models dual (1.8, 2.0 and 2.5) that is the way it's meant to be successfull

[greenstork]

Originally posted by eric67
this news is not new.
(but then Dual 3GHz, it's really going to smoke P4 Prescott)

The Prescott isn't going to be what Apple is up against by the middle to end of 2004. It's going to be the Tejas and Nehalem (http://www.theinquirer.net/?article=11785) and my guess is that they will either be on par or faster than a 3.0 Ghz G5.

[Analog Kid]

Originally posted by eric67
euh.... not really
current P4 are releasing around 75W and +, and the next generation to come (Prescott) at 2.8GHz is generating...103W!!!!!
so who is speaking about PPC970 being to hot????

The 103W is neither confirmed or denied by Intel, but they have said that they don't think 100W is such a bad number... (!?!)

Worth noting that Prescott is fabbed at 90nm. Smaller size doesn't always mean lower power anymore... If the die is filled with caches, your transistors leak like sieves, and the clock is ramped up to fight the MHz battle, then power is going to continue to climb...

[eric67]

Originally posted by mmmbop
Or even expanding the range of dual processor machines, not just on the top model.

exactly what I think...
having a fantastic OS taking full advantage of 2 processors and not having the machine for that.........
PLEASE APPLE makes them all DUAL G5

[stockscalper]

Exactly what is needed for the Powerbook! Sufficient power and very low heat output.

[Analog Kid]

Originally posted by Rincewind42
Process size has no direct influence on what targets the chip is used for, insofar as a smaller process can certainly be used for the same tasks as a larger process. The advantages of a smaller process are less power, heat and a smaller package. Those advantages do tend to allow you to use chips in places they weren't usable in before.

Smaller process also means lower per-chip costs because you can get more on the same wafer.

Alternately you can fill the die area with other logic or cache and keep the price roughly the same.

I'm guessing we'll see the later, though the extra logic/cache will mean higher leakage current.

Anyone know details on the Power5 power reduction techniques? If they're just turning off the clock it seems there will be less of a benefit as leakage goes up... Are they gating power to the various sections yet? Seems like this would also have limited benefit if you've got to cycle power at a GHz rate...

[Dont Hurt Me]

Originally posted by csimmons
Honestly, I think it would not be very wise to put a G5 in a iMac. I think the pro line and the consumer line should be kept separated; i.e., G5's in the Power Mac, Power Book and Xserve, G4's in the iMac, iBook and eMac. That way, there won't be too much of a canibalization in sales, and there's a very distinctive product line, like it used to be back in the day.

Just my opinion. i disagree, from the advances ibm is making(unlike moto) there is going to be some very fast g5's in the near future. why not just use single slower g5's in the consumers and duals at higher speeds in the pro's???why hold on to the g4 at all??? 90nm is going to be so sweet.

[eric67]

Originally posted by greenstork
The Prescott isn't going to be what Apple is up against by the middle to end of 2004. It's going to be the Tejas and Nehalem (http://www.theinquirer.net/?article=11785) and my guess is that they will either be on par or faster than a 3.0 Ghz G5.

well it is coming fron the inquirer, so take it really with forceps, it will happen (I mean Tejas) but hey the Prescott is not even out yet, currently planed in November, at starting clock speed around 2.8GHz.
so I do not believe that the tejas will be there before 1 year.
but I might be wrong... but before the tejas, Intel will have to solve a serious problem, high clock speed means heat release (a lot), and to keep up with IBM and AMD, they are preping now high cache processor, so which also means heat release increased...
I mean soon you will need a fridge to cool down a P4...

[dongmin]

Originally posted by greenstork
Word on the street has Intel cranking out 5-7 Ghz chips with 2MB of L2 Cache and a 4000Mhz frontside bus speed (http://www.theinquirer.net/?article=11785) as soon as the end of next year. If Apple and IBM want to compete with this, they need to get cracking. I would argue however, that this is just too much speed and completely unnecessary.

You gotta have a more solid source than a very vague Inquirer report to make your pint. For starters, it doesn't say at what clockspeed it will debut. I highly doubt Intel will be at 5 ghz six months from now (they're stuck at 3.2 ghz now, aren't they?). And I especially doubt Intel will be anywhere near 7 ghz by the end of next year. Silly.

[GUSTO]

Will this new technology make the chip run faster???

eg: a dual 2Ghz G5 (0.130nm) and a dual 2Ghz G5 (0.90nm), would the 0.90nm chip run faster, have more power?????

[fourthtunz]

Originally posted by eric67
no that's not the way to go. 1.8GHz single, then all models dual (1.8, 2.0 and 2.5) that is the way it's meant to be successfull

Yeah this is more like it, but how soon do we see dual 1.8's,2.0(and lower prices:D)
and when do we see the higher clockspeed? Nov and then Jan?
I'm waiting for my G5 until more dual comes out. I wonder if we'll ever have a G5 that will boot 9.2:D BOOO HISSSSSS
daniel

[greenstork]

Originally posted by dongmin
You gotta have a more solid source than a very vague Inquirer report to make your pint.

The folks at Slashdot (http://slashdot.org/article.pl?sid=03/09/28/1438254&mode=thread&tid=118&tid=126&tid=187) give this report some creedence. I know it seems a bit far-fetched but hardly silly IMO. Intel exploits the MHz myth/gap better than anyone.

[3.1416]

Originally posted by greenstork
A PC emulator already runs on a G5. It's not Virtual PC but it's just as good. The emulator is called WinTel

Um, no. All reports are that WinTel/Bochs is *far* too slow for any sort of useful work.

[Dont Hurt Me]

if you want a peek at the future look very closely to IBM. Apple see's this and so do i. did you know that ibm is working on technology that will make all ram obsolete?? even hard drives in the near future. there will be memory banks per say like on star trek and ibm is making it happen. what better to mate to a new 980 chip then a solid state harddrive/memory built into one device. its coming 2005/2006 anyways forget intel and windows.

[eric67]

Originally posted by greenstork
The folks at Slashdot (http://slashdot.org/article.pl?sid=03/09/28/1438254&mode=thread&tid=118&tid=126&tid=187) give this report some creedence. I know it seems a bit far-fetched but hardly silly IMO. Intel exploits the MHz myth/gap better than anyone.
during the recent Intel Development Meeting there were not such announcements from Intel side, so it looks to me more like excited guests who misunderstood what they have been told, and it end up in theinquirer hears...

[danielgrenell]

this is very exciting, but the question is, when will these chips make their way into powermacs?

[macmax]

Originally posted by manu chao
I won't buy a G5 Powerbook until VirtualPC (if ever???) runs on a G5. And since dual G4s (e.g. @1.25Ghz) seem to be faster with MP-aware, non-G5-optimised programs than single G5s (e.g. @1.6Ghz), I am still hoping for something like a dual 1.33 Ghz 17'' Powerbook.

are you the musician????

This is Cerati.

[seamuskrat]

Originally posted by greenstork
A PC emulator already runs on a G5. It's not Virtual PC but it's just as good. The emulator is called WinTel and is put out by an open source company called OpenOSX. It's G5 optimized to run all the Windows operating systems, check it out:

WinTel 1.0.1 (http://www.openosx.com/wintel/index.html)

Have you actually USED Wintel? Unless you are the developer or the folks selling it, I cannot possibly imagine how you can say Just As Good as VPC.
VPC was slow on a good day with a fast machine. With a dual G5, running Win2K and Office Wintel is a pokey slow dog. Its awful to the point of unusability. In fact, on Dual 1 and 1,25 G4 and any flavor of G5 which I use at work (university) Wintel has been bogged down with Win2K or XP severely.
I would urge potential users who need X86 emulation to be wary of the Wintel product and use it first before basing a purchase decision on its "just as good" G5 availbility.
Plus, on another notem, the openosx priduct line and support are quationable. The support in my case for purchased products was abyssmal. Where I come from, poor support and communication = poor company. But alas, the wheels of university purchasing turn slowly and they STILL manage to buy from them, thus Wintel. And its very slow and buggy. IME, in XP Pro is doies not even function at least 45% of the time. Now, the website suggestes NOT using 2k or XP. Well, in a university or corporate environment its basically NT or 2K or XP. 98 is so insecure on a corporate net its not safe, and for the record, it runs better, but still poorly. Not acceptible for even web design standards checking and such, IME/O.

[seamuskrat]

Originally posted by dongmin
The 1 million dollar question: is it faster than VPC?

ANSWER:
far from it.
On a dual G4 @ 1.0 with a gig of RAM, using a fully patched Win2K its far slower.

Now, using Win 98 its usable, and I have not done benchmarks, but VPC is quite peppy at win 98 as well. I would say its a toss up for speed leaning towards VPC. I have only tried Win2K and XP with a G5 using Wintel and that is dog slow.

[manu chao]

are you the musician????

not really

[Rincewind42]

Originally posted by GUSTO
Will this new technology make the chip run faster???

eg: a dual 2Ghz G5 (0.130nm) and a dual 2Ghz G5 (0.90nm), would the 0.90nm chip run faster, have more power?????

No, a 2Ghz G5 is a 2Ghz G5, regardless of the process that is was made with they will be the same speed. The power usage will be different, as the smaller process will use less power (assuming nothing else changes). What a process shrink does is allow for higher speeds of the processor. So while 2 Ghz might be the highest you can get a G5 at 130nm, you might be able to get a 2.5 or higher Ghz G5 on a 90 nm process.

[IJ Reilly]

Originally posted by ima_pseudonym
Apple *has* switched processors without changing form factors. The original G3 powerbook was in the 3400 case.

Ya got me there... nice call!

Still stand by my prediction, though.

[misr12]

Originally posted by Ambrose Chapel
hopefully this will push PBs into G5-land way before the end of '04. Steve announced Desktop G5s at WWDC 03, maybe he'll announce the PowerBook G5 at WWDC 04?? Here's hoping


The only thing I'm interested in from apple is something with a G5 in it. iMac is the ticket for me.....

[Abstract]

They won't have a G5 Powerbook until January 2005 (17 months or so from now) ;).

[Mineral]

Originally posted by Abstract
They won't have a G5 Powerbook until January 2006 (17 months or so from now) ;).


I would tend to agree.. There's simply no need for one yet.. The new G4's are plenty fast for a portable.

[Tim Flynn]

It would be interesting if Apple moved all lines to the G5.
Power Macs for performance -> fastest G5
iMac etc optimized for low cost and quiet - > slower G5, passive heatsink, low voltage, low power.
Perhaps G5 90nm at 1Ghz could get by with passive cooling.

[Phil Of Mac]

Originally posted by bluedalmatian
Personally I'd rather see a G5 Xserve before a Powerbook.

at present the g4 PBs are doing very well but i've a feeling that the Xserve is flagging a little

Xserve is important. But basically, what Apple product line COULDN'T use a G5?

The G5 is the new G3. It'll be across the line before the G6 comes out in January so July 2005. Mark my words.

Originally posted by csimmons
Honestly, I think it would not be very wise to put a G5 in a iMac. I think the pro line and the consumer line should be kept separated; i.e., G5's in the Power Mac, Power Book and Xserve, G4's in the iMac, iBook and eMac. That way, there won't be too much of a canibalization in sales, and there's a very distinctive product line, like it used to be back in the day.

Just my opinion.

Back in the day, the iMac had a G3, and so did the Power Mac, and so did the Powerbook. Back in the day, the iBook G3 and PowerBook G3 coexisted peacefully.

Originally posted by ima_pseudonym
Apple *has* switched processors without changing form factors. The original G3 powerbook was in the 3400 case.

That was 1997, Apple was desperate, and Steve Jobs had not yet asserted his full control. That PowerBook G3 was a hack, it was overpriced, it didn't sell, and it ultimately failed. It will not be repeated. Apple is on secure enough ground that they don't need to do that again.

Originally posted by greenstork
The folks at Slashdot (http://slashdot.org/article.pl?sid=03/09/28/1438254&mode=thread&tid=118&tid=126&tid=187) give this report some creedence. I know it seems a bit far-fetched but hardly silly IMO. Intel exploits the MHz myth/gap better than anyone.

The folks at Slashdot are mocking this even more than we are.

http://slashdot.org/comments.pl?sid=80306&cid=7077719

The article doesn't say the processor will have 64-bit extensions. The article doesn't say anything.
Some quotes:
_ "The Pentium V is likely..."
_ "The processor we believe..."
_ "The final design of this arrangement is not set in stone."
_ "...details have not been confirmed,..."
_ "... the source claimed..."
_ "The Pentium V could have..."
_ "...although this may be reserved for the next chip along, the Nehalem"

This isn't news, this is BS speculation.

Originally posted by Dont Hurt Me
if you want a peek at the future look very closely to IBM. Apple see's this and so do i. did you know that ibm is working on technology that will make all ram obsolete?? even hard drives in the near future. there will be memory banks per say like on star trek and ibm is making it happen. what better to mate to a new 980 chip then a solid state harddrive/memory built into one device. its coming 2005/2006 anyways forget intel and windows.

I would highly doubt that it will be released around 2005/2006. IBM's research is really, really long-term. It'll be awhile. But I anticipate it.

[Dont Hurt Me]

Originally posted by Tim Flynn
It would be interesting if Apple moved all lines to the G5.
Power Macs for performance -> fastest G5
iMac etc optimized for low cost and quiet - > slower G5, passive heatsink, low voltage, low power.
Perhaps G5 90nm at 1Ghz could get by with passive cooling. this is what we will see,but at higher clocks and if anyone didnt hear me the first time hard drives and ram will be obsolete in 3 years and those with deep pockests two.

[SelectBishopEgg]

Rest assured, Intel is not heading for the kind of machine hinted at earlier, at least not in the mentioned time frame. New 64-bit extensions (with a new instruction set which must be compatible with AMD's instruction set, or no 64-bit Winduz for Intel) suddenly on chip within 6 months? Erm...no. 64-bit extensions rather than a true 64-bit implementation? Complete waste of time. Stacked chips that give out as much heat as current Pentiums? It might just melt through your desk. 5-7 Ghz isn't entirely out of the question, Intel like the big numbers, but it will probably have a pipeline about 2000 instructions deep. You'll need hyperthreading emulating more than two cpus to make that ****er efficient.

All that said, IBM still need to get a move on, Opteron is already kicking G5 ass and is a major threat.

[Rocketman]

Originally posted by GUSTO
So correct me if I am wrong;)

These new chip are ready THIS YEAR!? and will be shipped THIS YEAR!? so we could see these new chips in new powemacs??? aswel as powerbooks???

I am going to buy a dual G5 this week, but now with this new info should I hold off for these new G5 chips?????

And if these chips get into new powermacs when do you think they will see the light of day?

Hold your horses, cowboy!

IBM is shiping INITIAL processors to people (Apple presumably) in LATE December 2003. Apple typically sells alot of chips at introduction so tends to hoard chips for a while before introduction of a product or conversely assemble entire products sans chips awaiting the arrivalof the coveted chip.

That means at the very luckiest Apple will announce a G5 Powerbook in January it might ship as early as late February (ie 5 MONTHS from now).

Back to the Batcave Batman, we have to fire up the Bat Computer and reclaculate the arrival of the evil Riddler using the hard to detect 90um 970 chips.

Rocketman

[dongmin]

Originally posted by Mineral
I would tend to agree.. There's simply no need for one yet.. The new G4's are plenty fast for a portable.

i think Abstract was trying to make a joke (although not funny). The G5 PowerBook is coming, before the end of year next year...ccording to Steve Jobs anyway.

[snofseth]

I hope there are G5's next summer for school next year i don't care for much before that because i don't think i would be let to buy a new one before that unless my 12'' breaks. Someone said the design is great i agree in most cases but the footpads are terrible they dont stay on and the battery isn't flush, those are the only things I dont like but I want a bigger wide screen, but not 15'' try like 13 that would be awesome but im not sure how the font would be if it would be two small, I really don't like hearing people saying things aren't possible, because they are and they probably start work on the newer product right after or even before they anounce its predecesor

[Rocketman]

You can bet Steve Jobs has a G5 Powerbook on his desk right now. It may have alot of unresolved manufacturability and functionality issues, but all preproduction computers do. They are complex interdependent devices.

What would be exceptionally cool is a "Powerbook cooling dock" that is NOT a dongle, but one of those flat plates that fit under your (12" min) Powerbook and COOL IT. Heck integrate a battery if you must. So what if you added a half inch or more. It is an OPTIONAL ACCESSORY not needed for superlightweight notebook needs. When you want to burn some power you are willing to be encumbered just a bit.

It would also make dual processor G5's immediately practical as the system would only run on half power, one processor, then when the dock is engaged, and full power dual or quad processors okay.

Rocketman

http://www.v-serv.com/-upload/avatar2.jpg

[tychay]

Originally posted by dongmin
This is according to the page 2 rumor: 34W seems way high. At the same voltage, a 1.6 ghz 970 (90 nano) would give off 27W, too high for a laptop I presume. Of course, this is only a page 2 rumor so we don't want to read too much into it. And there are many other factors that come into play.

It struck me that a 2ghz 970 (130 nano) gives off 57W. Seems awfully high. Isn't that in the Pentium 4 range?

There are two errors in this post.

1) You compute a linear relationship between clock speed and power requirements--the actual relationship is nonlinear. The current 970 G4 which drains 48W @ 2Ghz (not 57 as mentioned) also drains only 19W @ 1.2Ghz. This means a 1.6Ghz hypothetical 90nm PPC 980 would be under 20W and into the range allowed for a Powerbook.

2) No, 57W is not awfully high and it isn't in the P4 range. Yes, it is a lot larger than the thirty something watts that a G4 at top speed uses, but a 3.2Ghz P4 currently has drains over 100W at peak (Intels specs are a little funky so you have to add around 15 watts to their numbers). In other words, a 2x2G5 uses around the same power a a single 3.2Ghz P4! Maybe this is how you thought that it was "in the P4 range" as the P4 does not come in dual CPU configurations.

Having said this, I think the implication of the rumor is overly optimistic. IBM announced to the world the 970 @ 1.8Ghz a full year in front of the time they started to appear in shipping systems. Given that sort of lead time, I think it is more reasonable that may be some of the currently sampling 970s @ 90nm for production in 2Q or 3Q 2004 have leaked and are feeding the rumor mill.

Of course, if it comes sooner, I wouldn't object :) and neither should PC owners since Intel is sitting on the next generation Pentium Ms due to lack of competition.

Originally posted by IJ Reilly
Another argument against G5 PBs any time soon: Apple has never jumped processor families in their portables without changing the form factor.

While true in spirit, this isn't completely true. Apple introduced the G3 Powerbook (http://www.everymac.com/systems/apple/powerbook_g3/stats/powerbook_g3_1st.html) shortly after the G3 systems (http://www.everymac.com/systems/apple/powermac_g3/stats/powermac_g3_233_dt.html). Both had the same case as previous desktop (http://www.everymac.com/systems/apple/powermac/stats/powermac_7600_200.html) and notebook (http://www.everymac.com/systems/apple/powerbook/stats/mac_powerbook5300ce_117.html) counterparts. They were quickly replaced with much better (http://www.everymac.com/systems/apple/powermac_g3/stats/powermac_g3_300_bl.html) models (http://www.everymac.com/systems/apple/powerbook_g3/stats/powerbook_g3_233.html).

I doubt Apple will do this again.

Originally posted by greenstork
A PC emulator already runs on a G5. It's not Virtual PC but it's just as good. The emulator is called WinTel and is put out by an open source company called OpenOSX. It's G5 optimized to run all the Windows operating systems, check it out.

As long as people keep posting it, I'm going to keep adding this disclaimer. 1) It is not made by OpenOSX. All OpenOSX does is repackage what someone else makes for free (http://bochs.sourceforge.net/). Users would do better installing Bochs themselves than using that companies products giving their reputation. (http://fink.sourceforge.net/pr/openosx.php).

Also, it isn't G5 optimized. All they did was compile it with the latest version of gcc (possibly, but doubtful, they could have used IBM's XLC (http://www-3.ibm.com/software/awdtools/ccompilers/)). Anything else is a violation of LGPL. Since Bochs emulates even the endian in software and uses no hardware accelleration other than provided in the compiler, I'd bet is is slower than Virtual PC (I haven't confirmed as I already have a copy of VPC and don't own a G5).

Originally posted by greenstork
Word on the street has Intel cranking out 5-7 Ghz chips with 2MB of L2 Cache and a 4000Mhz frontside bus speed (http://www.theinquirer.net/?article=11785) as soon as the end of next year.

You can trust microInquirer as much as the paper its printed on. The rumor is that Intel has had serious issues getting anything coming out of Fab11x (their 90nm facility) that works like it does in their Oregon design fab. The reality is Intel has yet to release a single chip at 90nm (as noted by a previous poster, 90nm Xilinx FPGAs are coming out of Fishkill though FPGAs are simpler than processors and thus easier). I've said this for months and Intel has yet to prove me wrong, in fact one wonders why they had to come out with the "Extreme Edition" P4 @ 3.2Ghz to counter the Athlon64 launch if they could put out 90nm processors. Perhaps by December...

Take care,

terry

[brhmac]

Almost embarassed to ask, but since I don't know, I'm honest enough to ask.... :rolleyes:

Can someone explain to me how the heat of the processor chip is measured in watts?

I understand a lower number is better, but I don't understand the scale. X watts as this processor speed, Y watts at that.

So, for example, would a chip giving off 57W of heat be as hot as a 60W light bulb?

[Rincewind42]

Originally posted by tychay
1) You compute a linear relationship between clock speed and power requirements--the actual relationship is nonlinear. The current 970 G4 which drains 48W @ 2Ghz (not 57 as mentioned) also drains only 19W @ 1.2Ghz. This means a 1.6Ghz hypothetical 90nm PPC 980 would be under 20W and into the range allowed for a Powerbook.

Actually, there is a linear relationship between clock speed and power requirements. However, there is also a quadratic relationship between core voltage and power requirements. Specifically, the 1.2 Ghz part described in the original 970 presentation was a 1.1v part, while the 1.8 Ghz part was a 1.3v part. Assuming a similar speed up occurs independently of core voltage, by this rumor a 90nm 970 would be capable of running in the 1.9-2.1 Ghz range at 1.1v (while using approximately 75% of the power of it's 1.3v brother at the same clock speed). Relative to the 3.2 Ghz part on page 2, a 2Ghz 970 at 1.1v would use about 32 watts - similar to the rumored power usage of the 2Ghz part mentioned on the page 2 version of this rumor actually.

[Plutoniq]

The IBM PowerPC 750GX will probably also be reduced to 90nm along with the 970. One of the key selling points of the 750XX is it's low power consumption in the embedded market, so further reducing of the 750GX from 130nm to 90nm seems likely to happen. Unlike the G5, the 750xx line is not solely produced for Apple, meaning it moves ahead independent of Apples needs from a processor.

p9

[dongmin]

Originally posted by tychay
There are two errors in this post.

1) You compute a linear relationship between clock speed and power requirements--the actual relationship is nonlinear. The current 970 G4 which drains 48W @ 2Ghz (not 57 as mentioned) also drains only 19W @ 1.2Ghz. This means a 1.6Ghz hypothetical 90nm PPC 980 would be under 20W and into the range allowed for a Powerbook.

ok, I'm not an engineer but I'm pretty sure power dissipation does go up linearly as frequency increases. The 1.2 ghz 970 is of a lower core voltage than the 2.0 ghz 970. So, as I stated in my original post, under the same voltage, the 1.6 ghz 90 nano 970 would be around 27W.

Originally posted by tychay
2) No, 57W is not awfully high and it isn't in the P4 range. Yes, it is a lot larger than the thirty something watts that a G4 at top speed uses, but a 3.2Ghz P4 currently has drains over 100W at peak (Intels specs are a little funky so you have to add around 15 watts to their numbers). In other words, a 2x2G5 uses around the same power a a single 3.2Ghz P4! Maybe this is how you thought that it was "in the P4 range" as the P4 does not come in dual CPU configurations.


Actually my statement is accurate if you are comparing the chips at the same clock speed. According to this intel document (ftp://download.intel.com/design/Pentium4/datashts/29864310.pdf), Pentium 4s give off around 55W at 2 ghz. Again I'm not an engineer so I can't judge the validity of Intel's claims but just so you know that I'm not pulling this stuff out of my ass.

[Rincewind42]

Originally posted by brhmac
Can someone explain to me how the heat of the processor chip is measured in watts?

I understand a lower number is better, but I don't understand the scale. X watts as this processor speed, Y watts at that.

So, for example, would a chip giving off 57W of heat be as hot as a 60W light bulb?

Watts are simply a measure of energy transmitted per second. For electricity, that is amps times volts. For an entire processor, an estimate can be made by using the equation W = V * V * S * C, where W is watts, V is volts, S is clock speed, and C is capacitance. (physically, this is volts by volts by farads by hertz, which comes out as watts). Typically shrinking transistors reduces capacitance (smaller circuits hold less power), slower clock speeds reduce the power (the slower the chip goes the less power it uses). Additionally, the higher the voltage, the easier it is to raise the clock speed of the chip, hence you will often see chips with lower voltages having lower high end clock speeds.

Now, relating all this to the light bulb in your ceiling. Yes, 57 watts from your CPU is exactly the same as 57 watts from that light bulb. The difference is that the primary output from the light bulb is light (duh) and heat. A CPU however, converts far less energy into heat, as if it did it would not perform at all (it needs the electricity to do it's work). Thus a CPU consuming 57 watts will not be as hot as a light bulb doing the same, but then it also won't be a bright =).

[Analog Kid]

Originally posted by Rocketman
...
IBM is shiping INITIAL processors to people (Apple presumably) in LATE December 2003. Apple typically sells alot of chips at introduction so tends to hoard chips for a while before introduction of a product or conversely assemble entire products sans chips awaiting the arrivalof the coveted chip.

That means at the very luckiest Apple will announce a G5 Powerbook in January it might ship as early as late February (ie 5 MONTHS from now).
...


The article says "chips"... Not 970s, not even processors, just "chips".

There's no solid evidence that the 970 itself is going to 90nm, nor that it'll be one of the first chips off the new line...

Nothing about this article says anything about when G5 Powerbooks will ship. I'm guessing Power5 is first off the line...

The 970 just started shipping-- there's plenty that can be done with tuning the process on that before they start shrinking it.

Glad to hear IBM has a 90nm process running. Apple will get chips from it eventually. Life is good.

[Gyroscope]

Originally posted by SelectBishopEgg

All that said, IBM still need to get a move on, Opteron is already kicking G5 ass and is a major threat.


How? Where? Please backup your claim!

[dscottbuch]

Originally posted by Rincewind42
[snip]

Now, relating all this to the light bulb in your ceiling. Yes, 57 watts from your CPU is exactly the same as 57 watts from that light bulb. The difference is that the primary output from the light bulb is light (duh) and heat. A CPU however, converts far less energy into heat, as if it did it would not perform at all (it needs the electricity to do it's work). Thus a CPU consuming 57 watts will not be as hot as a light bulb doing the same, but then it also won't be a bright =).

Just a little clarification. The energy input in both examples is 57 watts. For output, in the case of the light bulb a significant portion (hopefully) exits a photons (light) and the rest as heat. In the case of a processor the outputs are heat and a very small (in terms of energy) change in the entropy of the system represented by the alterations in 1's and 0's in the information of the system. While I haven't done the calculation I'm pretty sure the entropy change is much LESS than the energy of the photons in the light bulb so I believe the processor gives off significantly more heat than the light bulb.

Cheers

[Analog Kid]

Originally posted by Dont Hurt Me
if you want a peek at the future look very closely to IBM. Apple see's this and so do i. did you know that ibm is working on technology that will make all ram obsolete?? even hard drives in the near future. there will be memory banks per say like on star trek and ibm is making it happen. what better to mate to a new 980 chip then a solid state harddrive/memory built into one device. its coming 2005/2006 anyways forget intel and windows.

I'm presuming you're talking holographic memories? 2010 on the inside... First as mass storage, and maybe eventually propogating into system memory.

Stuff like this comes and goes, and I wouldn't be too surprised if holographic memories went...

Originally posted by Dont Hurt Me
Lets build the Orbital Spaceplane that is fast,economical & safe, and use shuttle as a cargo mover not people mover. 70's technology needs to be replaced! and now Nasa is thinking of apollo like capsules from the 60's. They need some serious vision.

When the Shuttle was built, they thought it was fast, economical an safe. Turned out to be none of the three, but it was cool. Hard to think of anything faster, or more economical than a capsule. As unsafe as it looks, the physics are well understood, so it's safe too.

I like the Russian capsules-- set off a small explosion before touching down to cushion your impact! :)

Sorry, off topic...

[NeoMayhem]

Originally posted by manu chao
I won't buy a G5 Powerbook until VirtualPC (if ever???) runs on a G5. And since dual G4s (e.g. @1.25Ghz) seem to be faster with MP-aware, non-G5-optimised programs than single G5s (e.g. @1.6Ghz), I am still hoping for something like a dual 1.33 Ghz 17'' Powerbook.

Why the hell would you want virtual pc? The main reason to use a mac is to stay away from MS :P

[Phil Of Mac]

Originally posted by Analog Kid
When the Shuttle was built, they thought it was fast, economical an safe. Turned out to be none of the three, but it was cool. Hard to think of anything faster, or more economical than a capsule. As unsafe as it looks, the physics are well understood, so it's safe too.

The single purpose of the space shuttle was to be reusable. It is reusable. It was hoped that the reusability would make it more economic.

The question isn't what spacecraft we use to get into orbit, though. The question is what spacecraft we use to leave Earth entirely.

[cb911]

what... 90nm by the forth quarter... this year??:eek:

if they're available in quantity by the end of this year i think it's safe to say that Apple won't release G5 PowerBooks for at least another 6 months... maybe a little less. Apple usually stretches out their releases to get a bit more life out of them. and now they're seen they can get 8 months our of the 15" PB's... oh no!!:eek: :p :D

[Abstract]

Originally posted by dongmin
i think Abstract was trying to make a joke (although not funny). The G5 PowerBook is coming, before the end of year next year...ccording to Steve Jobs anyway.

Um....no, I was serious: January 2005. Yes, I posted 2006 at first, but I also said approximately 17 months from now. Actually, its more like 15 months. I didn't bother figuring it out.

Just because IBM has a 90nm 970 ready by Q4 of this year, doesn't mean that Apple will have developed and tested a laptop using this chip. And MWSF is the perfect place to debut it, isn't it?

So no, I wasn't joking, although I did wink at you ----> ;) See, there I go again!!!

[Plutoniq]

Originally posted by NeoMayhem
Why the hell would you want virtual pc? The main reason to use a mac is to stay away from MS :P

I go as far as using Virtual PC on my Powerbook solely for the use of Fruity Loops. It's one piece of music software that has no counterpart on Mac.
I'm sure many Mac users use Virtual PC for the same reason, because theres always "just that one program" that is irreplacable. It's only the OS that sux.

[Analog Kid]

Originally posted by Rocketman
You can bet Steve Jobs has a G5 Powerbook on his desk right now. It may have alot of unresolved manufacturability and functionality issues, but all preproduction computers do. They are complex interdependent devices.


Put a prototype, proof of concept machine on the CEOs desk?

What do you think a CEO does all day, play with toys?! :rolleyes:

Are there any reports of actually seeing a G5 Powerbook in Cupertino? 90% of the analysis can be done without actually building one... Can actually be done better by simulation.

[Analog Kid]

Originally posted by Phil Of Mac
The single purpose of the space shuttle was to be reusable. It is reusable. It was hoped that the reusability would make it more economic.

The question isn't what spacecraft we use to get into orbit, though. The question is what spacecraft we use to leave Earth entirely.

If I hadn't already gotten so far off topic, I'd ask why bother making your single purpose to be reusable just for the sake of being reusable?

No, the purposes were to build a transport which would allow more closely spaced missions at a lower cost-- the answer appeared to be reusability. Problem is that rather than just burning up a cheap engine in re-entry they had to go and find it and refurbish it. And then they had to comb over the entire hull with a microscope and ultra-sound. So it cost a fortune and took forever for each launch.

Trouble is by the time they realized this, they'd lost the ability to make the larger unmanned rockets. There's a last Saturn V sitting someplace waiting to be reverse engineered if need be, because NASA lost the blueprints...

But that would be way off topic in a thread about 90nm IBM chips, so I'll just appologize for the whim and try to get back on topic...

I'd bet we won't see 90nm chips on board any space craft before 2015, by the time they clear all the testing rigors and reliability requirements.

[Phil Of Mac]

Originally posted by Analog Kid
If I hadn't already gotten so far off topic, I'd ask why bother making your single purpose to be reusable just for the sake of being reusable?

Did you even bother reading what I wrote before commenting on it?

The single purpose of the space shuttle was to be reusable. It is reusable. It was hoped that the reusability would make it more economic.

Don't argue with someone who already agrees with you. It wastes time and annoys the person who already agrees with you.

[eric67]

Originally posted by Mineral
I would tend to agree.. There's simply no need for one yet.. The new G4's are plenty fast for a portable.

everything depends on Motorola, how long will they continue to provide G4 for Apple, I think since CEO departure from Moto, it is quite well admitted that there will be no futher development of the G4, so at the best we can expect 7457 finally taped out and produced...that's all.
so the rumors of IBM working on a altivec-enable G3 make lot of sens in that particular context

[oswizrd]

Originally posted by tychay
Also, it isn't G5 optimized. All they did was compile it with the latest version of gcc (possibly, but doubtful, they could have used IBM's XLC (http://www-3.ibm.com/software/awdtools/ccompilers/)). Anything else is a violation of LGPL. Since Bochs emulates even the endian in software and uses no hardware accelleration other than provided in the compiler, I'd bet is is slower than Virtual PC (I haven't confirmed as I already have a copy of VPC and don't own a G5).




I couldn't agree more.
I have used both, the original open source bochs v2.0 and VPC v6.
On my 1Ghz TiBook I tried to install Debian Linux under bochs but gave up after one day. It was unbearably slow.

On the other hand, installing it under VPC was almost as fast as on my old Pentium III 600Mhz. As a result, I have now several VPC instances running Debian, FreeBSD, PC DOS.

[Sol]

I doubt that we will see G5 PowerBooks before the end of this year. The latest revisions were announced only a few days ago.

Smaller 970s are inevitable. I imagine that Apple would wait for these before changing the innards of the PowerBooks. As for the G4 iBook, the current 12" PowerBook's motherboard can easilly be fitted in its form-factor so maybe that is what will happen. This is 2003 and G3s are so nineteen-nineties...

Apple can improve their PowerMacs now with dual G5s across the line. Heck, make the high-end PowerMac a quad-G5 if the technology permits it. I am sure IBM would love to see such a desktop put to shame anything with Intel Inside.

As for the XServe, get the G5s and Serial ATA drives in there all-ready Apple.

[Phil Of Mac]

Originally posted by Sol
Apple can improve their PowerMacs now with dual G5s across the line. Heck, make the high-end PowerMac a quad-G5 if the technology permits it. I am sure IBM would love to see such a desktop put to shame anything with Intel Inside.

The power to crush the other kids.™

[tychay]

Originally posted by dongmin
ok, I'm not an engineer but I'm pretty sure power dissipation does go up linearly as frequency increases. The 1.2 ghz 970 is of a lower core voltage than the 2.0 ghz 970. So, as I stated in my original post, under the same voltage, the 1.6 ghz 90 nano 970 would be around 27W.

I reread my post and realized I'm in error: I stated that the power requirements are "not linear" with frequency. You and others trying to give me a basic course in Physics are correct. Dynamic power requirements (capacitance charging) are linear with frequency and the square of core voltage. For some reason I was thinking of leakage current without considering that it would just shift the y-intercept and not in a manner favorable to my argument!

Your computation assumes that the 970 drains 0 Watts at 0 frequency and creates a generous estimation of the power requirements at the same core voltage for lower clock speeds. This is incorrect. There is a static power drain: a constant subthreshold current (leakage power) in a transistor. This can be quite significant. For instance, the 3.2Ghz P4 leakage is around 20 Watts(!) at any clock speed--enough to power a G4 or suck a laptop battery dry!

Intel solves this in the P4M by applying SpeedStep to power down or clock down when idle (or just generically when on battery power). However the real savings is much simpler. Core voltage needs to be kept high to allow fast switching and core voltage is quadratic in dynamic power and exponential in static power. This effect gets exemplified in the Pentium M/Banias by clocking even slower and getting a better return--the PM has a leakage of < 1W. Too bad the sequel 90nm Pentium M/Dothan leaks like a sieve (10W).

So yeah, you're mostly right but you are assuming that a 90nm 970 will need to be at the same voltage. That assumption is usually erroneous--it's really hard to say here since you are comparing a 1.6Ghz part to a 2.0Ghz one. Usually the core voltage would be lower in a notebook than a desktop and thus the power is significantly lower. In fact most low power chips implement Dynamic Voltage Frequency Scaling (the generic term for Intel "SpeedStep" or AMD "PowerNow") which scale the frequency and voltage for processor demand.

I think in the interest of completness there should be another power drain due to the finite voltage response of the transistor. I don't know if this is significant since I'm not a hardware guy.

Actually my statement is accurate if you are comparing the chips at the same clock speed. According to this intel document (ftp://download.intel.com/design/Pentium4/datashts/29864310.pdf), Pentium 4s give off around 55W at 2 ghz. Again I'm not an engineer so I can't judge the validity of Intel's claims but just so you know that I'm not pulling this stuff out of my ass.

So I should be comparing them at the same clock speed? Hmm, then the P4 must really suck because its performance per clock is so miserable. We should ditch them and dig up our old Pentium 3's

If I'm going to go through this absurdity, why don't I compare different generations of the same CPU to skew things back in my direction? For instance your middle of the line 55W @ 2Ghz P4/Northwood used to drain 75W when it was a 2Ghz P4 Williamette (Again this is due to a drop in core voltage from 1.75 to 1.5V made possible by moving from 180nm to 130nm: to give you a rough feeling how significant core voltage can be and what a die shrink enables).

The reason they have similar power requirements at the same clock is because they have similar transistor counts (and core voltage)--power being an absolute linear with transistor count (and thus exponential with time due to Moore's Law which is why we're in this mess: the days of the 6W max power G4 are gone forever).

I've not seen anyone consider a 2Ghz P4 in the same class as a 2Ghz G5. Nobody benchmarks this because we all know the answer. They compare the P4 at 3.2Ghz or, more appropriately the P4 Xeon at 3.06Ghz (Intel claims 82 watts for both).

I say "Intel claims" because all the above are Intel "thermal guidelines". In reality the P4 3.2Ghz peaks above 100W as I mentioned earlier. Compare peak to peak and average to average. A single 3.2Ghz P4 uses more power than two 2.0Ghz G5s!

[Analog Kid]

Originally posted by Phil Of Mac
Did you even bother reading what I wrote before commenting on it?

The single purpose of the space shuttle was to be reusable. It is reusable. It was hoped that the reusability would make it more economic.

Don't argue with someone who already agrees with you. It wastes time and annoys the person who already agrees with you.

Then don't agree with me so much... :p

Sorry, man. Bad day. I read your post to say that the goal wasn't economic but reusability with the hope that there might be an economic secondary effect.

Actually, looking at most of my posts today I've been rather antagonistic. Taking my bad day out on the rumor boards again, I guess...

[NicoMan]

Originally posted by etoiles
yay, the all new powerbook rumors season has begun !
:p
The break was relly short...

[NicoMan]

Originally posted by manu chao
I won't buy a G5 Powerbook until VirtualPC (if ever???) runs on a G5. And since dual G4s (e.g. @1.25Ghz) seem to be faster with MP-aware, non-G5-optimised programs than single G5s (e.g. @1.6Ghz), I am still hoping for something like a dual 1.33 Ghz 17'' Powerbook.
Don't hold your breath for that dual G4 PB. Besides we can assume (hope) that most of the apps will soon be able to take advantage of the G5.

[NicoMan]

Originally posted by Tim Flynn
Xilinx ( a FPGA) vendor is all ready shipping FPGAs built upon 90nm design.
The chips are fabbed by IBM.
So a 90nm 970 in the near future sounds reasonable.
Normally for a CMOS design the power calculations (used to be ) are simple.
Basically power was linear with frequency and squared with voltage. But with the smaller geometeries, leakage current (power) has become a bigger factor. Early information for 90nm showed that the leakage current (power) was about half of the total.
So not all electrons are used to find aliens or fold a protein, a big part is used to heat the room :D

But, I'm sure IBM has some tricks up their sleeves ;)
That's what the SOI is for.

[NicoMan]

Originally posted by IJ Reilly
Potentially, the G5 could creep into the consumer line (iMac) by the end of next year, but I'd call that very doubtful.
They might have to change the form-factor for the iMac too.

[AidenShaw]

Originally posted by tychay
(Intel claims 82 watts for both).

I say "Intel claims" because all the above are Intel "thermal guidelines". In reality the P4 3.2Ghz peaks above 100W as I mentioned earlier. Compare peak to peak and average to average. A single 3.2Ghz P4 uses more power than two 2.0Ghz G5s!

Don't be so sure that Intel is playing a game here.

The IBM papers for the PPC970 quote wattage use as "typical", which in my mind means that they are not using peak wattage.

It makes a lot of sense to quote numbers that actually reflect typical heavy use - and not to focus on a theoretical "worst case" instruction stream that manages to keep every part of the chip busy.

Intel's "thermal design" guidelines are merely an attempt to characterize the upper limit of power consumption that is likely to be seen in hard, sustained use. Since the P4 chips have internal temperature sensing and controls this is safe, even if you do hit the mythical worst case load (you can actually run a P4 without a heat sink, and it won't be damaged (http://www6.tomshardware.com/cpu/20010917/heatvideo-01.html)).

So, find the real peak numbers for both CPUs, or use the manufacturer's ratings for both. Don't inflate the Intel numbers by some "thumb in the air" factor just to align with your viewpoint.

[NicoMan]

Originally posted by stingerman
The 7457 in the iMac with 2MB L3 cache or no L3 cache but 400MHz effective FSB is more than enough for the iMac along with a ATI 9600 GPU.

But it would be nice to get a headless desktop version of the PowerMac G5 for a typical corporate user who doesn't need all the features of the G5 but would prefer it over an iMac.
Bring the Cube back!!

[MadMan]

Originally posted by Phil Of Mac
The power to crush the other kids.™

That's what my T-shirt says :)

:cool:

MM

[jouster]

--------------------------------------------------------------------------------
Originally posted by SelectBishopEgg

All that said, IBM still need to get a move on, Opteron is already kicking G5 ass and is a major threat.
--------------------------------------------------------------------------------

Well, maybe for people like us. But the Operating System is still the main decision driver for the average computer user. I would contend that OS X is generally well perceived and has a real chance to drvie increased Apple sales.

I am currently researching certain aspects of computer history for my senior year thesis, and one thing here may be valid:

The 'killer feature' that drove sales tends to jump about, at least in hardware (unfortunately, it has remained DOS/Win for soft ware for the majority of users).

First, it was having a floppy drive. Then RAM became the buzzword - in the early PC era, no one cared much about clock speeds (this can be seen by examining which features advertisers chose to promote in the early 80s). This lasted for a while, with two major blips: the moves to 16 bit then 32 bit, both of which were used to power sales. Only after the 32 bit era was well underway did clock speed start to become a major factor in Joe public's perception of what made a good computer (with Windows taken for granted unfortunately).

Well, we are at another junction: what will drive public perception in the next few years? Will the jump to 64 bits suffice? Perhaps not; it is hard to explain any benenfit of such a move to non geeks, especially as there is virtually no killer app yet taking advantage of it. Will it be clock speed? Perhaps not - even Intel seems to be moving away from promoting that aspect. I would tentatively suggest that the OS, in combination with integrated multimedia (remember that word?!) features like iTMS and so on will be the area to concentrate on. Most people now perceive computers as being able to fulfill their needs on a hardware level, and pay less and less attention to the processor. They simply want to be reassured that 'it will do the web' or 'it will play my games', abnd as long as they walk out of compUSA with such a system they could care less whether it has an Intel or AMD processor. The fact that Apple's processor is different is far less relevant than the fact that their OS is different.

All this is a long way of mildly disagreeing with the AMD related assertion quoted above, but we shall see......that said, it is always worth bearing in mind that most users are FAR less interested in the specs of their machine than the average macrumors poster. Even if the Opteron does perform better than the G5, will the majority care? Can processor choice ever become the most important factor in buying a computer, given the underwhelming reception of 64 bit ness?

[Rincewind42]

Originally posted by Analog Kid
Put a prototype, proof of concept machine on the CEOs desk?

What do you think a CEO does all day, play with toys?! :rolleyes:

Don't forget which CEO we're talking about here - the one that spent 20 minutes at MWSF describing the reasoning behind the shape of the (then) new iMacs :D .

[Rincewind42]

Originally posted by tychay
You and others trying to give me a basic course in Physics are correct.

(plus way more CPU information that I thought I would learn on these boards)


Now who's giving the physics lesson? :D

[NicoMan]

Originally posted by Rincewind42
...
Now, relating all this to the light bulb in your ceiling. Yes, 57 watts from your CPU is exactly the same as 57 watts from that light bulb. The difference is that the primary output from the light bulb is light (duh) and heat. A CPU however, converts far less energy into heat, as if it did it would not perform at all (it needs the electricity to do it's work). Thus a CPU consuming 57 watts will not be as hot as a light bulb doing the same, but then it also won't be a bright =).
Thank you for all this explanation. I am myself an engineer (in the French sense) but for some strange reason, I forgot it all (and believe me, I used to do this sort of ***** in my sleep:D). Anyway, for some other strange reason, I always associated the power numbers given in processor specs as heat dissipated as opposed to power consumed... silly me.

Now about how hot the chip is compared to a light bulb. Let's imagine that heat dissipation (in watts) was the same (highly unlikely) for the chip and the light bulb. That's not enough to say which one is hotter to the touch. Because you need to know how vast the surface of dissipation is (I don't know if it's clear). In our case, if the heat dissipation is the same the chip surface is smaller than the bulb's (I guess) so the chip would feel hotter. Does that make sense?

[Kurt]

Originally posted by Rincewind42

Now, relating all this to the light bulb in your ceiling. Yes, 57 watts from your CPU is exactly the same as 57 watts from that light bulb. The difference is that the primary output from the light bulb is light (duh) and heat. A CPU however, converts far less energy into heat, as if it did it would not perform at all (it needs the electricity to do it's work). Thus a CPU consuming 57 watts will not be as hot as a light bulb doing the same, but then it also won't be a bright =).

That does not make sense to me. What goes in must come out. If a processor is taking in 57 watts of power then that energy has to go somewhere. Some of it is used to drive the external circuits like the bridge chips but that is a small amount. The rest is converted to heat. IIRC from my physics courses all energy eventually becomes heat. While it is using the 57 watts to work (or operate), most of it eventually becomes heat in the chip that must be carried away. Hence the big heat sinks.

If you to back to the light bulb analogy, an incandenscent bulb is not very efficient. A 60 watt bulb converts a certain percentage of its input to light and the rest to heat. A fluorescent bulb converts a higher percentage to light and less to heat. A motor converts most of its energy to mechanical motion and a small percent to heat. I am not sure what 'work' a microprocessor does.

[daveL]

Originally posted by Kurt
I am not sure what 'work' a microprocessor does.
Uh, turning transistors on and off?

[Analog Kid]

Originally posted by Kurt
That does not make sense to me. What goes in must come out. If a processor is taking in 57 watts of power then that energy has to go somewhere. Some of it is used to drive the external circuits like the bridge chips but that is a small amount. The rest is converted to heat. IIRC from my physics courses all energy eventually becomes heat. While it is using the 57 watts to work (or operate), most of it eventually becomes heat in the chip that must be carried away. Hence the big heat sinks.

If you to back to the light bulb analogy, an incandenscent bulb is not very efficient. A 60 watt bulb converts a certain percentage of its input to light and the rest to heat. A fluorescent bulb converts a higher percentage to light and less to heat. A motor converts most of its energy to mechanical motion and a small percent to heat. I am not sure what 'work' a microprocessor does.

Power numbers are typically quoted with the outputs disconnected... So no energy is being transfered to external circuits.

The power required to drive the bus is above and beyond the power quoted for the chip.

There might be some heat generated in the ground plane because of currents between local differences in potential, but this isn't included in the chip power either-- whatever the local difference in potential to true ground, the power rail to the chip must be the specified voltage above the potential at the ground pins or nothing works.

Additional heat is dissipated in the power supply because of the inefficiency of converting 12V to 1.1V, and this is directly related to the current drawn by the chip, but it is not included in the quoted chip power because the external system is unknown.

The "work" done is charging the internal capacitors (transistor gates). This is the entropy (or reduction in entropy) that was described in an earlier post. Trouble is that all that organization gets changed with every clock.

What goes up, must come down-- every '1' eventually becomes a '0'. All those charged caps eventually are discharged. As this happens, the current in and out of the transistors is run through the internal wiring resistance and disipates as heat.

It's like lifting a weight repeatedly-- the fact that that the weight hasn't gone anywhere in the end doesn't mean your arm isn't tired. Your muscles aren't 100% efficient, and they don't have a means of reusing the energy released in bringing the weight back down-- so that is all wasted as heat.

In this case it's a very small weight but it's being lifted up and down billions of times a second (GHz).

Driving the external bus is the same process but with a much larger weight-- each copper trace has a capacitance much higher than an internal transistor. Current comes out of the pins to raise the bus signal to '1' and then flows back through the same pins to drain the signal to '0'.

[edit: the leakage power can be viewed as a resistor connected between power an ground. The current just flows through the silicon, dissipating heat as it goes with nothing useful coming of it...]

If current goes into a chip with the outputs disconnected, it comes out as heat. As stated above, all that heat is coming off a few square mm of monolithic sand... More heat than a 57W lightbulb over a much smaller surface area means a much higher temperature than that lightbulb.

The heat sinks try to give the heat more surface area to disipate over-- keeping the temperature down. The fans try to aid convection in transfering that heat to the air and then to the environment.

[Phil Of Mac]

Thanks for reminding me why I didn't become an EE major.

[Rocketman]

Originally posted by Kurt
That does not make sense to me. What goes in must come out. If a processor is taking in 57 watts of power then that energy has to go somewhere.

If you to back to the light bulb analogy, an incandenscent bulb is not very efficient. A 60 watt bulb converts a certain percentage of its input to light and the rest to heat.

I am no electrician. However:

1. A lightbulb generates heat in a very small space (under 1 cubic mm).
2. A lightbulb is optimized to generate light.
3. A lightbulb is not only 60 watts but also 110 volts at 60 hz
4. A processor generates heat in a very wide space with a heat dissipation system in direct contact such as the substrate, the heatsinks, the fans, the case, the table, the room. So the heat distribution is far higher. A lightbulb is in a vacuum ans its external surface is glass which is a near perfect heat transmission medium.
5. A processor is optimized to produce NO light, MINUMUM heat, and mazimum hz (Mhz or ghz to be exact)
6. A processor is not only 60 watts but also 3.3 volts or more recently as low as 1.1 volts at 1-4 Ghz

It is a power and energy DISTRIBUTION problem. The lightbulb is like a girl standing on you with all her weight with a single high heel, vs, a really wide piece of paper double her weight sitting on you. Most of it drapes onto the ground harmless. As for the girl, after you get out of the hospital she owes you a favor. FORCE her to buy you a quad-G6 Powerbook. No not the other thing, it doesn't last long enough :)

Rocketman

Simple terms, simple minds.

[Plutoniq]

I'm curious about heat generation of the PowerPC 750FX vs. the 750GX.

Both are produced on a 130nm process, however, because of the 750GX's 1mb L2 cache (double the 750fx), the die size will increase from 36.6mm squared to 51.9mm squared.

Out of these to situation, granting similiar clock speeds, would a smaller die sized 750FX generate less heat than the 750GX. Smaller die-size giving the 750FX lower power consumption, but wouldn't a larger die-size be better for optimal heat transfer to a heat sink.

http://www-3.ibm.com/chips/products/powerpc/newsletter/jun2003/newproductfocus.html

[jouster]

Originally posted by Rocketman
The lightbulb is like a girl standing on you with all her weight with a single high heel, vs, a really wide piece of paper double her weight sitting on you. Most of it drapes onto the ground harmless. As for the girl, after you get out of the hospital she owes you a favor. FORCE her to buy you a quad-G6 Powerbook. No not the other thing, it doesn't last long enough :)

Rocketman

Simple terms, simple minds.

Well, certainly a more entertaining analogy than the endless comparisons with cars.......:p

[Analog Kid]

Originally posted by Rocketman
I am no electrician. However:

1. A lightbulb generates heat in a very small space (under 1 cubic mm).
2. A lightbulb is optimized to generate light.
3. A lightbulb is not only 60 watts but also 110 volts at 60 hz
4. A processor generates heat in a very wide space with a heat dissipation system in direct contact such as the substrate, the heatsinks, the fans, the case, the table, the room. So the heat distribution is far higher. A lightbulb is in a vacuum ans its external surface is glass which is a near perfect heat transmission medium.
5. A processor is optimized to produce NO light, MINUMUM heat, and mazimum hz (Mhz or ghz to be exact)
6. A processor is not only 60 watts but also 3.3 volts or more recently as low as 1.1 volts at 1-4 Ghz

It is a power and energy DISTRIBUTION problem. The lightbulb is like a girl standing on you with all her weight with a single high heel, vs, a really wide piece of paper double her weight sitting on you. Most of it drapes onto the ground harmless.

Was this whole thing a parody, or were you serious?

[Kurt]

Originally posted by Analog Kid
Was this whole thing a parody, or were you serious?

My thoughts exactly.

[Kurt]

Originally posted by Analog Kid
Power numbers are typically quoted with the outputs disconnected... So no energy is being transfered to external circuits.

The power required to drive the bus is above and beyond the power quoted for the chip.

The "work" done is charging the internal capacitors (transistor gates). This is the entropy (or reduction in entropy) that was described in an earlier post. Trouble is that all that organization gets changed with every clock.

What goes up, must come down-- every '1' eventually becomes a '0'. All those charged caps eventually are discharged. As this happens, the current in and out of the transistors is run through the internal wiring resistance and disipates as heat.

It's like lifting a weight repeatedly-- the fact that that the weight hasn't gone anywhere in the end doesn't mean your arm isn't tired. Your muscles aren't 100% efficient, and they don't have a means of reusing the energy released in bringing the weight back down-- so that is all wasted as heat.


Driving the external bus is the same process but with a much larger weight-- each copper trace has a capacitance much higher than an internal transistor. Current comes out of the pins to raise the bus signal to '1' and then flows back through the same pins to drain the signal to '0'.


If current goes into a chip with the outputs disconnected, it comes out as heat. As stated above, all that heat is coming off a few square mm of monolithic sand... More heat than a 57W lightbulb over a much smaller surface area means a much higher temperature than that lightbulb.

The heat sinks try to give the heat more surface area to disipate over-- keeping the temperature down. The fans try to aid convection in transfering that heat to the air and then to the environment.

I think you meant to say pretty much what I did. If you think of work as energy transmitted in another form then heat, the processor does not do any. All 57 watts is eventually dissapated as heat.

[tychay]

Originally posted by AidenShaw
So, find the real peak numbers for both CPUs, or use the manufacturer's ratings for both. Don't inflate the Intel numbers by some "thumb in the air" factor just to align with your viewpoint.

I remember reading somewhere that Intel thermal design numbers were still about 10-15 watts below "typical". You are right however, there seem to be two numbers regarding a 2Ghz G5: 48 watts and 57 watts. I guessed the latter was peak and the former typical. The adjustment I need to make in my statement in both cases would be to say that a single 3.2Ghz P4 drains about as much power as two 2Ghz G5s.

Take care,

terry

[NicoMan]

Originally posted by Rocketman
I am no electrician. However:

1. A lightbulb generates heat in a very small space (under 1 cubic mm).
2. A lightbulb is optimized to generate light.
3. A lightbulb is not only 60 watts but also 110 volts at 60 hz
4. A processor generates heat in a very wide space with a heat dissipation system in direct contact such as the substrate, the heatsinks, the fans, the case, the table, the room. So the heat distribution is far higher. A lightbulb is in a vacuum ans its external surface is glass which is a near perfect heat transmission medium.
5. A processor is optimized to produce NO light, MINUMUM heat, and mazimum hz (Mhz or ghz to be exact)
6. A processor is not only 60 watts but also 3.3 volts or more recently as low as 1.1 volts at 1-4 Ghz

It is a power and energy DISTRIBUTION problem. The lightbulb is like a girl standing on you with all her weight with a single high heel, vs, a really wide piece of paper double her weight sitting on you. Most of it drapes onto the ground harmless. As for the girl, after you get out of the hospital she owes you a favor. FORCE her to buy you a quad-G6 Powerbook. No not the other thing, it doesn't last long enough :)

Rocketman

Simple terms, simple minds.
The wattage says it all in terms of power. Once you got that number, frequency and voltage are irrelevant (of course for our processors, the important stuff to start with is the voltage and frequency). I don't want to be pedant, but if you were to remove the heatsink on your microprocessor (don't do it, you risk killing your chip) the microprocessor would probably be a lot hotter (might melt?) than the bulb cause the area of contact with the outside is much larger on the bulb (of course it's gonna depend on what type of bulb we are talking about). Hence the need for heatsink, cause then there is direct contact between the chip and the heatsink (actually there is some sort of heat conducing paste/glue between them) which makes it easier for the chip to release its energy, and the heatsink provides the surface to release the heat to the outside world.

[Rincewind42]

Originally posted by NicoMan
The wattage says it all in terms of power. Once you got that number, frequency and voltage are irrelevant (of course for our processors, the important stuff to start with is the voltage and frequency). I don't want to be pedant, but if you were to remove the heatsink on your microprocessor (don't do it, you risk killing your chip) the microprocessor would probably be a lot hotter (might melt?) than the bulb cause the area of contact with the outside is much larger on the bulb (of course it's gonna depend on what type of bulb we are talking about). Hence the need for heatsink, cause then there is direct contact between the chip and the heatsink (actually there is some sort of heat conducing paste/glue between them) which makes it easier for the chip to release its energy, and the heatsink provides the surface to release the heat to the outside world.

A CPU is typically on the order 100 - 200 square millimeters in size. By comparison the filament in a light bulb is probably about 5 centimeters of wire (coiled) with a thickness of about .1 millimeters. The filament in a light bulb glows at around 2500 C - your typical microprocessor won't go over 50 C when properly cooled, and will blow way, way before getting as hot as the filament in a light bulb (silicon's melting point is "only" 1410 C). So no matter how you calculate it, the working parts of a light bulb are going to be hotter than the working parts of a CPU.

[Phil Of Mac]

I FIGURED IT OUT! I'LL MAKE MILLIONS!!!

You make a computer that attaches to a water input for watercooling. You set up a system where the water evaporates, and then you condense it again outside of the computer, thus having both a cooler computer and distilled water!

[manu chao]

Although I don't believe the numbers for the 7455 refer to the processor used in the Ti-books (there must have been a lower voltage version, and I remember having seen lower numbers) this PDF (http://e-www.motorola.com/files/32bit/doc/app_note/AN2436.pdf) shows 8 Watts for the 7457 (w/Altivec enabled). That is almost as low as a G3. Now you know why they reduced the capacity of the battery in the new Powerbooks.

[yamabushi]

Energy goes in as electricity, energy comes out as electricity and heat. A lot of this waste heat is the result of wasted clock cycles. Remove the requirement for everything to run at set frequencies and you get more efficient use of power.

Intel once experimented with a clockless Pentium II. It had about twice the performance and about half the heat dissapation of it's clock-bound version. The project was cancelled due to cost, reliability, and motherboard design issues. In this scenario the system bus and memory became the bottlenecks.