[Gul Banana]

Good news for laptop users - according to the Ars article, the lower-speed versions of the 970 can run as low as 19 watts, significantly lower than the G4 currently in the Powerbooks! In other words, Powerbooks and maybe even (eventually) iBooks with this chip: eminently practical.

[euphras]

quote:
--------------------------------------------------------------------------------
Quote:
"Originally posted by voodoo:
and one question:

the PPC 970 is a dual core processor.

what does that mean?
--------------------------------------------------------------------------------


That would mean that IBM had squeezed two processors in to one chip. That way, a dual processor machine would really only have one package inside of it, but that had two chips that are as independent as the two G4's in a PowerMac now. Going this route has advantages and disadvantages. Most notably, the fact that it consumes power roughly like two processors means that it wouldn't be practicable for laptops without shutting down one core, making it uneconomical. This among other reasons discourages dual independent core CPUs.

BlackGriffen"

The way i understand it, the 970 PPC will be a single core/die processor. Maybe with the opportunity to scale up to a dualcore/die proc, when it�s well accepted and sells like hotdogs

[Mark Tungston]

So let's talk real now. What we need to hear is Apple publicly announcing the ppc970 as it's next processor.

Will they name it a "G5"?

Will they wait till MWSF to announce the new chip?

[Graymalkin]

Originally posted by Gul Banana:
Good news for laptop users - according to the Ars article, the lower-speed versions of the 970 can run as low as 19 watts, significantly lower than the G4 currently in the Powerbooks! In other words, Powerbooks and maybe even (eventually) iBooks with this chip: eminently practical.

Or even better would be a quad processor XServe using the same amount of power as a single Pentium Xeon chip. As long as Apple didn't go entirely retarded and charge $10,000 for it a quad processor XServe would be a sweet server machine. Just having a quad processor system under 10k would be sweet but having a quad processor 970 based system would be even sweeter. A 1U rack could blow the crap out of 2 and 3U rack systems in everything from web serving to 3D rendering. Even a dual Athlon system with googlebytes of memory bandwidth would be seriously contested by a 4 way 970 system.

[Metzen]

Originally posted by Graymalkin:
Or even better would be a quad processor XServe using the same amount of power as a single Pentium Xeon chip. As long as Apple didn't go entirely retarded and charge $10,000 for it a quad processor XServe would be a sweet server machine. Just having a quad processor system under 10k would be sweet but having a quad processor 970 based system would be even sweeter. A 1U rack could blow the crap out of 2 and 3U rack systems in everything from web serving to 3D rendering. Even a dual Athlon system with googlebytes of memory bandwidth would be seriously contested by a 4 way 970 system.

More like a 4-way 7-Watt G4 7457 Processor would be better suited to a server... But that could just be me

You want battery life? You got battery life! Just wait till volume production picks up Q2 of 03

Kind of amazing though, that the move from .18u SOI to .13u SOI saves you greater than 2.6x on your power consumption.

[aaanorton]

Hey Scott,
I personally would like to see this thread re-stuck. It's handy having it right there, as this is and will continue to be an important Mac issue. Maybe leave it til IBM starts shipping them or at least MWSF.
Thanks.

[Skinlayers]

Reposted from /.

What I really want to know is how much this chip is going to cost. If its cheap for Apple to put 2 or 4 of these in a machine, then how much will it matter that an expensive P4 (P5) out performs it? Hmmm.... The current Wind-Tunnel G4s raised a few eyebrows when it first came out do to the new case design. These things were designed to disapate heat! A HUGE (7 lbs) heat sink w/ matching fan, a small case fan, 2 fans on the power supply, and a ton of ventalation in the back. WAY more cooling that those 2 little G4s require. I think Apple is trying to avoid the fiasco it had with the Sawtooth (1st gen) G4s where they just slapped a G4 onto a G3 mobo. This time around, I believe they're releasing a new mobo first and then put a new proc in it down the road. I've also read stuff in forums suggesting that the power supply for the Wind-Tunnel had way more juice than the system currently demands. Can anyone out there do the math on this? We know how much power the PPC 970 eats. Can we figure out how much heat the Wind-Tunnel case is designed to disapate? What about how much power the power supply is putting? With these numbers, can we figure out how many PPC 970 the Wind-Tunnel case could power and cool? I've been suffering with a 266MHz G3 iMac, and I refuse to upgrade until Apple comes out with a system that really is worth that premium they charge, and a G4 is not it.

[Scotttheking]

Originally posted by aaanorton:
Hey Scott,
I personally would like to see this thread re-stuck. It's handy having it right there, as this is and will continue to be an important Mac issue. Maybe leave it til IBM starts shipping them or at least MWSF.
Thanks.

It is going to be left unstickied, and once it becomes relevent again I'll make a 2nd one and link to this one, since this is a little long to read thru.
I'd like to only have 1 or 2 sticky threads, and I am definitely making a fan noise sticky thread.

[Mark Tungston]

Originally posted by Skinlayers:
I've been suffering with a 266MHz G3 iMac, and I refuse to upgrade until Apple comes out with a system that really is worth that premium they charge, and a G4 is not it.

that's masochistic!!!!

you should have bought a dual 533. it would have brought you into the g4 era and you'd be tided over until the 970 hits. which may not be for 2004 (if it ever does). my computer is important in entertainment and work for me...i'll upgrade on a regular cycle so i dont suffer.

it's like you're on a pentium 3 and skipping the p4s to wait for the p5.

how much experience have you had with g4 machines? although powermacs are expensive, the g4 is a fine performer. i'm on a laptop 500mhz g4 and it only shows it's age when i really open 5-7 apps

[CIA]

------------------------------------------------------------------------
Original Post:

One factor to consider would be just plain size. In order to make a proc, they need a big enough single wafer of silicon that is up to snuff. The bigger the wafer/crystal they need to grow, the more difficult and costly it is. Pretty much, figure that given the fact that defects in the silicon wafers are a certainty, the smaller the chip, the more usable area there is on a less than perfect wafer. Combine that with the fact that, in the market IBM is targeting this chip at, there is demand for single processor models that would outstrip the likely defect rate in a dual core chip (IBM improved profits on POWER4 production by selling chips with a single defective core as single core modules), and it all adds up to dual core chips simply not being worth it.

I unfortunately don't have numbers to back me up, just speculation based on the fact that growing large silicon wafers is hard.

BlackGriffen
------------------------------------------------------------------------

If I recall, the power 4 was rather over-engineered. Part of the reason they are so expensive is that they are designed to have a MUCH lower failure rate then a standard proc. Not that regular proc's have a high failure rate, but in the server market, it's a good selling point to show that your chip will be 99.999999% fail-safe. This rugged construction, (explained somewhat in that great Ars article) should make dual Core 970 chips much easier to create. Of course this is depending on how much of the power 4's design makes it into the 970. (I'm not a "chip" engineer though so I'm not really knowledgeable on this particular topic, just guesstimating). I believe that while possible, the economics of dual vs. single core chips will lead Apple to stick to multiple chips for dual or quad Macs, over dual or quad _core_ chips. Actually, outside of the server market, quad macs are not in Apple's best interest. remember, the bread and butter to Apple's bottom line is the iMacs and iBooks. While the power is in the powermacs, the money is in the cheap computers. Build a $4,000 quad 970 based machine, sell a few. Build a $1,200 single 970 based machine, sell a million. While style and power is important, the bottom line is what's important to Wall Street and Apple's shareholders. Place an order for 1 million 970 based chips, get a deal. Place an order for 600,000 970 single cores, 300,000 dual cores, and 100,000 quads (if the quads are even possible), most likely not as great a deal. This all depends of course on what these chips will cost in quantity, which no ones knows... Maybe we will get lucky and they will be a "Steal" in comparison to the current Moto G4's. Then we will finally get those 32 proc systems we all dream about. (how many seti units crunched in an hour?)

[JNG]

Originally posted by Mark Tungston:



that's masochistic!!!!

you should have bought a dual 533. it would have brought you into the g4 era and you'd be tided over until the 970 hits. which may not be for 2004 (if it ever does). my computer is important in entertainment and work for me...i'll upgrade on a regular cycle so i dont suffer.

it's like you're on a pentium 3 and skipping the p4s to wait for the p5.

how much experience have you had with g4 machines? although powermacs are expensive, the g4 is a fine performer. i'm on a laptop 500mhz g4 and it only shows it's age when i really open 5-7 apps

I agree with this. I too had the attitude of trying to wait and bridge the gap of the OS X infancy era until the Macs were built around a truly next-generation processor. Instead, when the eMac came out I saw my chance for an inexpensive smoothing of the way, so I traded my iMac (that I had gotten free from Apple) in for 40% of the new eMac's value through PowerMax, and boom, pretty good G4-based system for a pretty small (well, relatively) amount of money. I hope Apple does adopt IBMs offering and that it pans out well, so I can make that jump in 18 months or so and still get some trade-in on this eMac. Worth considering.

I guess some variant of the G4 will be lurking in Apple's product line for a good while, though, right? Unless Apple would want to split it, with the IBM PPC970 in desktops (or maybe only the PowerMac) and some flavor of Motorola 8xxx in portables. I think I might rather see clock-decreased or significantly stepped versions of the PPC970 in the (high-end) laptop, though; the tech people say it can be done.

[Mark Tungston]

Originally posted by JNG:


I agree with this. I too had the attitude of trying to wait and bridge the gap of the OS X infancy era until the Macs were built around a truly next-generation processor. Instead, when the eMac came out I saw my chance for an inexpensive smoothing of the way, so I traded my iMac (that I had gotten free from Apple) in for 40% of the new eMac's value through PowerMax, and boom, pretty good G4-based system for a pretty small (well, relatively) amount of money. I hope Apple does adopt IBMs offering and that it pans out well, so I can make that jump in 18 months or so and still get some trade-in on this eMac. Worth considering.

Yes, he is in no position to wait for the 970. We have no idea whats even going on with it and if it is coming at all. Not to mention if problems arise in manufactoring and etc etc etc

I am in a position to wait for maybe even the 2nd version of the ppc 970 in a powermac. i have a g4 500, 768 ram and it's chugging along quite nicely. you with the emac can wait too but this guy must be really suffering. He must be doing nothing but web, mail and word because a 233 g3 = waste of valuable working time.

To step up froma g3 sub 300mhz is not hard and you can find deals all over the internet to get a decent machine that can tide you over. i would say a the dual 533 is a excellent choice but there are many others.

anyway, the value in apple is that if you get a machine when you need it, then the price is justified with all the work you do on it until you buy the next machine. i think that maybe even now with the dual 867 for 1, 700 is a decent deal and in 3 -4 years you can upgrade to a 3rd - 4th generation PPC 970.

[clarkgoble]

Dang man. Get a dual 867 system. While not perfect and still a tad overpriced, it is a very nice system. It makes OSX a total joy to use. I bet by now you could even find some used ones on eBay for a reasonable price. (Me - I got mine the week they came out)

[Simon]

Originally posted by clarkgoble:
Dang man. Get a dual 867 system. While not perfect and still a tad overpriced, it is a very nice system.

I'll second on that. Even though I love OS X since PB came out, it was my new Dual 867 DDR that gave me back that OS 9 feeling of instant response to GUI action.

Of course the machine isn't cheap and even though it has "only" two G4s running at a sub-GHZ clock, it is really fast and reliable. The way I see it, a modern dual G4 PowerMac is an ideal purchase. It gives you enough time to wait for Apple to get the 970 or the 8500 or whatever it will be (G5?) in new Macs. Heck, it may even get you far enough for you to take a second generation model of the G5 when it arrives. By then the flaws (if any ) of the G5 could already be cured. Getting a dual G4 PowerMac is really a swell deal at the moment.

[rm199]

I would also like to get one, but id rather the older dual 1gig it looks nicer in my opinion and the bus speed/DDR ram is a bit of a con.

only thing holding me up is the games benchmarks i've seen on the web etc, i assume the 9700 will raise them somewhat but even another 40% leaves things quite far behind

RM

[Simon]

Originally posted by rm199:
I would also like to get one, but id rather the older dual 1gig it looks nicer in my opinion and the bus speed/DDR ram is a bit of a con.

I believe from a performance point of view there is no big difference between a QS dual and MDD dual at the same clock speed. Maybe the QS is slightly slower when doing intensive I/O or DMA but I suppose that's about it. I don't necessarily think the QS looks better but that's just my opinion. The QS is probably more quiet whereas the MDD has two optical bays and an ATA-100 channel. I also like the 2MB L3 cache on the QS.

only thing holding me up is the games benchmarks i've seen on the web etc, i assume the 9700 will raise them somewhat but even another 40% leaves things quite far behind

I guess for game specs you should really not get a Mac. A GF4Ti or R9700 is probably the best you can do and the 9700 isn't even out yet... For older games (like UT) I can assure you that the GF4MX (which is a real wimp compared to the other two mentioned GPUs) in a dual 867 works fine.

Have you thought of just getting a nice new PowerMac and a PS2 or GameCube?

[rm199]

hehe a new mac and a gamecube... hehe well i have a cube but there is something about 'pc' games that is just better... im running a powerbook and it deals with wolfenstein and mohaa fine but i think both of those and things like SOF2 will kill it with my new formac 2010.

who knows i'll just see what happens and hide a pc in the cupboard somewhere till the 970 makes an appearance.

RM

[Simon]

Originally posted by rm199:
who knows i'll just see what happens and hide a pc in the cupboard somewhere till the 970 makes an appearance.

Just don't forget to cut some vents in that cupboard and buy a couple of good fans.

[Leonard]

Originally posted by Mark Tungston:
So let's talk real now. What we need to hear is Apple publicly announcing the ppc970 as it's next processor.

Will they name it a "G5"?

Will they wait till MWSF to announce the new chip?

You're not a MacUser are you?

Apple never comments on future products. You'll only know that Apple is using the PPC970 when it appears in one of their PowerMacs.

[rm199]

personally I dont care what they call it... I just hope that between the CPU, bus, memory etc etc etc apple get the mac way in front of the pc. It cant be that hard the pc is a terrible botched together architecture

RM

[voodoo]

They will call the PPC 970 the "Tom Jones". It will give Apple such an incredible comeback after 15 years of obscurity. Word on the street.

[Mark Tungston]

Originally posted by Leonard:


You're not a MacUser are you?

Apple never comments on future products. You'll only know that Apple is using the PPC970 when it appears in one of their PowerMacs.

people in this forum have said the g4 was announced (for macs) a year before it came out at a processor trade show...

that's what i'm going by as i do not remember. i myself would think that a processor is too big a deal for Apple or IBM to keep quiet. From ibm's standpoint, they are bigger than apple and have their own sharholders to talk about. supplying a computer company with processors is a huge deal for ibm.

[Mark Tungston]

double

[Superchicken]

IBM has already anounced the chip. Same as Moto did, Apple will say hi we're using the PPC 790.. probably never they'll probably always call it the G5 or the F5 or something like that who knows, I'm sure the guys in marketing can come up with a fun name for it. But you'll find out for sure apple's using it a week before you can buy em in stores.

[Thain Esh Kelch]

Originally posted by Superchic[k]en:
they'll probably always call it the G5 or the F5 or something like that

F5??????

[voodoo]

Formula 5?

[Evangellydonut]

If I recall, the power 4 was rather over-engineered. Part of the reason they are so expensive is that they are designed to have a MUCH lower failure rate then a standard proc.Not that regular proc's have a high failure rate, but in the server market, it's a good selling point to show that your chip will be 99.999999% fail-safe. This rugged construction, (explained somewhat in that great Ars article) should make dual Core 970 chips much easier to create. Of course this is depending on how much of the power 4's design makes it into the 970. (I'm not a "chip" engineer though so I'm not really knowledgeable on this particular topic, just guesstimating). I believe that while possible, the economics of dual vs. single core chips will lead Apple to stick to multiple chips for dual or quad Macs, over dual or quad _core_ chips. Actually, outside of the server market, quad macs are not in Apple's best interest. remember, the bread and butter to Apple's bottom line is the iMacs and iBooks. While the power is in the powermacs, the money is in the cheap computers. Build a $4,000 quad 970 based machine, sell a few. Build a $1,200 single 970 based machine, sell a million. While style and power is important, the bottom line is what's important to Wall Street and Apple's shareholders. Place an order for 1 million 970 based chips, get a deal. Place an order for 600,000 970 single cores, 300,000 dual cores, and 100,000 quads (if the quads are even possible), most likely not as great a deal. This all depends of course on what these chips will cost in quantity, which no ones knows... Maybe we will get lucky and they will be a "Steal" in comparison to the current Moto G4's. Then we will finally get those 32 proc systems we all dream about. (how many seti units crunched in an hour?)

As you mentioned, the Power4 has much lower failure rate, due to the difference in the thickness of the oxide layer...which also makes the transistors slower. 970 uses "conventional" transistors thus will have failure rate equivalent to that of a desktop chip, but is able to run faster than Power4.

With that in mind, since the CPU is soldered onto the daughter card, failure rate is the same if you have a multi-core chip or a dual-CPU system (to the end consumer. Might be cheaper for Apple to fix it and use it in re-furb'd product though). I wouldn't be surprised if one variation of the 970 will have a dual core (contrary to IBM's current statement) once IBM can get the cost low enough or if they run into a speed scaling problem similar to Moto (failure rate really isn't a problem...when's the last time your none-overclocked CPU (mac or intel/amd) failed on you?). Multi-core CPU is overall faster than multiple CPUs on the same mobo, which is faster than Beowolf clusters over any type of connection...

[Metzen]

Originally posted by Evangellydonut:
With that in mind, since the CPU is soldered onto the daughter card, failure rate is the same if you have a multi-core chip or a dual-CPU system (to the end consumer. Might be cheaper for Apple to fix it and use it in re-furb'd product though). I wouldn't be surprised if one variation of the 970 will have a dual core (contrary to IBM's current statement) once IBM can get the cost low enough or if they run into a speed scaling problem similar to Moto (failure rate really isn't a problem...when's the last time your none-overclocked CPU (mac or intel/amd) failed on you?). Multi-core CPU is overall faster than multiple CPUs on the same mobo, which is faster than Beowolf clusters over any type of connection...

The 970 isn't designed for any kind of multi-core setup, or that's what I've read. It's only for multi-chip solutions (ala' G4 at present).

[Rickag]

The 970 appears that it will be a great chip, but Q3 2003 seems a long way off.

If you look at IBM's roadmap, sorry no link, the next generation PowerPC from IBM will be;

1+GHz
Multicore Superscalar
SMP Capable
Integrated SIMID Engine
RapidI/O
n-way Crossbar Core Connect
low-k Dielectric 0.13 - 0.10 �m process

Does any one have an opinion when this might appear. Isn't IBM currently manufacturing the G3 on a 0.13� process?? Any chance in #%** this might appear in Jan. 2003??

Hope springs eternal and all that you know.

[Evangellydonut]

Does any one have an opinion when this might appear. Isn't IBM currently manufacturing the G3 on a 0.13� process?? Any chance in #%** this might appear in Jan. 2003??

I don't think chip introduction goes by fiscal quarters...so nope, won't appear in Jan 2003. It's also a major upgrade for Apple to deal with, so I'll be happy if Apple can put a (or 2) 1.8GHz 970 in a system by Jan 2004...

[clarkgoble]

Just to go along with recent discussions here, Jon Stokes has an other interesting article over on Ars about Bandwidth in CPUs. Not so directly relevant to all our discussions, but it might clear up some issues regarding some misconceptions about some of Apple's recent choices and future moves.

[clarkgoble]

(Duplicate post due to forum hiccup)

[Rickag]

Originally posted by Evangellydonut:
I don't think chip introduction goes by fiscal quarters...so nope, won't appear in Jan 2003. It's also a major upgrade for Apple to deal with, so I'll be happy if Apple can put a (or 2) 1.8GHz 970 in a system by Jan 2004...

Evangellydonut, thanks for the input, but I don't think the processor I referred to is the 970 at all. The roadmap does not mention 64 bit processing, which I would think would be a big deal and be mentioned on the roadmap. I am guessing that the processor I refered to should appear before the 970.

IBM's PowerPCroadmap

[Eug]

Hmmmm... According to this description, the PPC970 is supposed to do RC5 at 18 million keys/sec on their 1.8 GHz chip. Pretty damn fast, but that's what one would expect with a Motorola 7455 at 1.8 GHz too.

Dunno what this means, but I thought it was something else you guys could bicker about.

EDIT: Apologies. I see this was already posted earlier in this thread.

[Metzen]

Originally posted by Eug:
Hmmmm... According to this description, the PPC970 is supposed to do RC5 at 18 million keys/sec on their 1.8 GHz chip. Pretty damn fast, but that's what one would expect with a Motorola 7455 at 1.8 GHz too.

Dunno what this means, but I thought it was something else you guys could bicker about.

It just shows that RC5 doesn't stream AltiVec code to the CPU, it sends it in chunks small enough to fit in the high-speed G4 Cache.

[Eug]

What's a 1 GHz G4 equivalent on average to in P4 terms these days anyway, ballpark of course? 1.5 GHz? (Excluding things like RC5 of course.)

[OreoCookie]

Originally posted by Eug:
What's a 1 GHz G4 equivalent on average to in P4 terms these days anyway, ballpark of course? 1.5 GHz? (Excluding things like RC5 of course.)

1 GHz PIII. At same clock speeds, a 1GHz PIII is faster than a P4, so I'd estimate about P4 1.2 GHz. (Based on SPECmarks)

[Eug]

Originally posted by OreoCookie:


1 GHz PIII. At same clock speeds, a 1GHz PIII is faster than a P4, so I'd estimate about P4 1.2 GHz. (Based on SPECmarks)

Where is everyone getting their G4 Spec benches? Was some of it based on this article? (I note the 2.95 GCC compiler was used by the way. Plus, being a general consumer, I'm more interested in Windows vs. OS X performance, not Linux vs. OS X performance.)

Just wondering. The reason I ask is because my totally non-scientific guesstimation of 1 GHz G4 performance based on non-spec benches in the past would have put it maybe into the PIII 1.2 GHz range on average, which in turn might have meant a P4 1.4-1.5 for the average desktop user. (Benches were putting the 1 GHz PIII at approx 1.2 to 1.4 GHz in P4 terms.)

But then again, people in this thread are suggesting that a current Athlon (which is faster than a PIII clock for clock) would be much faster than a G4 clock for clock on average. Perhaps the Jobs Reality Distortion Field has affected me after all?

I know it's hard to give an accurate estimate of processor speed with so many different applications being used, but I was just wondering what the general feeling was.

Anyways, part of the reason I ask lies in the fact that I just ordered a 1 GHz G4 TiBook. I'd be interested to know what best with which to compare it.

Power utilization-wise, a P4M 1.4 GHz should put out approximately 27 W (?) typically, vs. 21 W for a Motorola 7455. (These numbers are from memory from the published specs from the companies, so take them with a grain of salt.)

[Metzen]

Originally posted by Eug:
Just wondering. The reason I ask is because my totally non-scientific guesstimation of 1 GHz G4 performance based on non-spec benches in the past would have put it maybe into the PIII 1.2 GHz range on average, which in turn might have meant a P4 1.4-1.5 for the average desktop user. (Benches were putting the 1 GHz PIII at approx 1.2 to 1.4 GHz in P4 terms.)

Eeek! I base my estimates on a few factors, but when trying for total equality, I add the fudge.

Example:

Code:
G4 7455 - 7 Stage Pipeline
PIII    - 10 Stage Pipeline
P4      - 20 Stage Pipeline

Thus, all other things being equal (which they are not) a 700Mhz G4 = 1Ghz PIII = 2Ghz P4.

The G4 and PIII being architecturally similar (at least, more so than the P4) makes that comparison about roughly true. So we stick to the 7/10 ratio:

And we need roughly a 1.4Ghz PIII to equal a 1Ghz G4. It's not quite that, so adding the fudge puts it roughly around 1Ghz G4 = 1.3-1.4Ghz PIII.

The P4 mucks everything over, so we add a bigger fudge factor, but I'd say a 1.6Ghz-1.8Ghz P4 is roughly equal to a 1.0Ghz G4.

Originally posted by Eug:
But then again, people in this thread are suggesting that a current Athlon (which is faster than a PIII clock for clock) would be much faster than a G4 clock for clock on average. Perhaps the Jobs Reality Distortion Field has affected me after all?

Yeah... A Athlon has been compared to a PIII so many times I think it was like 900Mhz Athlon = 1.0Ghz PIII +/- the fudge factor.

It wasn't much to gawk at. Well... back then it was, I'm sure.

Originally posted by Eug:
I know it's hard to give an accurate estimate of processor speed with so many different applications being used, but I was just wondering what the general feeling was.

Anyways, part of the reason I ask lies in the fact that I just ordered a 1 GHz G4 TiBook. I'd be interested to know what best with which to compare it.

No P4M chip will come close to competing with the G4 while it's on the road. A 2.8Ghz P4M will quickly become a 1.4Ghz P4M when there's no wall outlet. I'm not 100% what the "speedStep" speed depreciation is, but it's pretty pathetic.

Originally posted by Eug:
Power utilization-wise, a P4M 1.4 GHz should put out approximately 27 W (?) typically, vs. 21 W for a Motorola 7455. (These numbers are from memory from the published specs from the companies, so take them with a grain of salt.)

Remember the fudge factor

[Eug]

Thanks for the info. I have to say I still wonder if 1.6-1.8 GHz is a bit optimistic though.

No P4M chip will come close to competing with the G4 while it's on the road. A 2.8Ghz P4M will quickly become a 1.4Ghz P4M when there's no wall outlet. I'm not 100% what the "speedStep" speed depreciation is, but it's pretty pathetic.

I should point out that there is no such thing as a 2.8 GHz P4M. The fastest current one is 2.2 GHz, although it runs at 1.2 GHz on the road. I find this strange, because the 1.4 GHz P4M also runs at 1.2 GHz on the road. The same is true for chips with a max speed of 1.5, 1.6, etc. I guess they figure that on the road 1.2 GHz is the sweet spot or something.

This is different from previous years, where the SpeedStepped speed increased when the max speed increased in the CPU line. Indeed, my PIII 600 laptop stepped down only to 500 MHz. The 1 GHz PIII stepped down to 733, and the 1.2 GHz PIII stepped down to 800 MHz.

Stepping on the 7455?

[OreoCookie]

[QUOTE]Originally posted by Eug:

Where is everyone getting their G4 Spec benches? Was some of it based on this article? (I note the 2.95 GCC compiler was used by the way. Plus, being a general consumer, I'm more interested in Windows vs. OS X performance, not Linux vs. OS X performance.)

Just wondering. The reason I ask is because my totally non-scientific guesstimation of 1 GHz G4 performance based on non-spec benches in the past would have put it maybe into the PIII 1.2 GHz range on average, which in turn might have meant a P4 1.4-1.5 for the average desktop user. (Benches were putting the 1 GHz PIII at approx 1.2 to 1.4 GHz in P4 terms.)

But then again, people in this thread are suggesting that a current Athlon (which is faster than a PIII clock for clock) would be much faster than a G4 clock for clock on average. Perhaps the Jobs Reality Distortion Field has affected me after all?

I know it's hard to give an accurate estimate of processor speed with so many different applications being used, but I was just wondering what the general feeling was.

Anyways, part of the reason I ask lies in the fact that I just ordered a 1 GHz G4 TiBook. I'd be interested to know what best with which to compare it.

Power utilization-wise, a P4M 1.4 GHz should put out approximately 27 W (?) typically, vs. 21 W for a Motorola 7455. (These numbers are from memory from the published specs from the companies, so take them with a grain of salt.) [/QUOTE

They have used the old compiler on purpose, because for other SPECmarks the new Intel Compiler for Linux was used that raises some SPECmarks significantly. Back then, there was no newer GNU compiler suite availble for OS X.

It is not easy to give a direct value in MHz. I am more productive on my iBook, but I am doing my simulations on AlphaStations and a Linux-PC cluster. Don't get me wrong, but I wouldn't even try that on my iBook.

It depends on the apps that you use. For web browsing, etc. your (very cool) 1 GHz PowerBook should do the trick for some time

No, seriously, there is no way to give a number, just a rought estimate. Mine is 1.2 GHz P4 (although there is no such thing).

[Metzen]

Originally posted by Eug:
Thanks for the info. I have to say I still wonder if 1.6-1.8 GHz is a bit optimistic though.

I should point out that there is no such thing as a 2.8 GHz P4M.

Sorry, I should have mentioned that as well, I was just using that as an example.

Originally posted by Eug:
The fastest current one is 2.2 GHz, although it runs at 1.2 GHz on the road. I find this strange, because the 1.4 GHz P4M also runs at 1.2 GHz on the road. The same is true for chips with a max speed of 1.5, 1.6, etc. I guess they figure that on the road 1.2 GHz is the sweet spot or something.

This is different from previous years, where the SpeedStepped speed increased when the max speed increased in the CPU line. Indeed, my PIII 600 laptop stepped down only to 500 MHz. The 1 GHz PIII stepped down to 733, and the 1.2 GHz PIII stepped down to 800 MHz.

Stepping on the 7455?

The "speed step" architecture on the G3, G4-style processors is called "processor-cycling". The difference is, instead of having a pre-set lower Mhz setting (ala Speed Step) the G3/G4 can "processor cycle" down to a lower Mhz when the processor isn't being pushed to use it's potential. This was all automatic till the PowerPC 750FX when IBM allowed it to be software controlled (this also allowed people to use software to OC on the fly their processors). All in all, if your playing Quake III on the new PowerBook with processor cycling enabled, you'll probably be running at a nice hot 1Ghz, where as if your watching a DVD you could be running at 500Mhz.

Anyways, that's the difference as I understand them.

[Eug]

This "processor cycling", is it similar to what the Radeon Mobility 9000 does?

Anyways, what are guesstimates of the 1.0 and 1.8 GHz PPC 970 speed in 32 bit mode in G4 7455 terms?