<strong>Moto have optimized their 130 nm fab and the next Powermac revision will bear the fruits of this labor.</strong><hr></blockquote>
Didn't Motorola say that they'd have switched to the .13µ process only with the G5? I remember reading about it a while ago, but can't remember where... Can somebody correct me on this?
[quote]<strong>Clock speeds are impressive. Without deepening the G4?s pipeline, Moto has attained clock speeds other CPU manufacturers only dream of. We?ve seen boxes ranging from 1.0 1.6 GHz, however, most likely is a 1.4 GHz high end.</strong><hr></blockquote>
The newest Athlon 2200+ is "only" a 1800MHz part: everybody is well aware of the P4 dirty tricks to achieve such monster clock-rates... In comparison the PIII is a much better design, and currently tops out at 1400MHz. So I think that these speed figures are impressive indeed, given tha G4 IPC efficiency.
RapidIO: what?!?!? Are we already there??? Wow... This part could very well be true, but in my mind RapidIO belongs to the realm of the "maybe-in-a-distant-future-or-probably-never" technologies, like widespread use of H-powered cars... I'd love to see it happen though...
I don't know, real insider info or just a very good educated guess?
<strong>1.4 GHz is definitely in the realm of possibility for a .13 micron G4.
This may be fuzzy logic since I don't design chips but going on the fact that Motorola had an orderable part number for a 1.1 GHz MPC7455 on one of its webpages, and that .18 microns is ~1.385x.13...
1.385x1.1 = >1.5 GHz. 1.4 GHz doesn't seem that hard, though I guess AMD went to .13 and only went up to 1.8 GHz from 1.733.</strong><hr></blockquote>
You forget that chips shrink in two dimensions:
1.385^2 x 1.1 = 2,1
Of course, this makes it seem like a shrinking dye directly translates in more Mhz, which isn't true.
i'd like to respond to Ensoniq. Basically, you said AI members should have, or atleast admitted that your have, a blind faith in Apple. I don't. I know that SJ is great businessman (he is a selfmade billionaire afterall). He obviously knows that Apple has to get their shyt together. But, lacking any credible evidence of this, for me it is hard to believe they are. I don't mean to say that I think they are sitting on their asses and throwing pencils in the ceiling, but the shyt they do put out isn't always great. Their new shyt sometimes isn't top-of-line compared to similar wintels. If they are waiting to put all their shyt into the NeXT generation machine (aka g5), then I'll be happy and probably overexcited to see it. In the meantime tha shyt ain't there.
And as far as rumors go, it seems that too many people put up shyt on these boards that they completely makeup. I can start a thread about new g5s with whatever specs, but it doesn't make it true. The boards here lack credibility. We have the right to be skeptical. I don't like people jerking me around.
i do think there is a g5. i don't think it'll be ready for a while tho (sf, ny nxt yr, ..ad infinitum..?). I'm not an insider, tho, just a pessimist. [prove me wrong apple]
OK, keeping in mind that I know nothing about the technicalities of making a computer run, here's what I see.
IPC is interesting, but P4 is still the performance king.*** Read on, see what I mean.
Recent tests of DP Athlon 1.6 vs DP 1Ghz Powermacs in 3-d and video effects, put the athlon box at about 60% faster. Which, outside of altivec, puts the IPC of both 'systems' about the same. BUT, the Athlon was running a much faster bus and disk system. So these tests don't really compare chips so much as they compare systems. You can't fault the Athlon box makers for including superior componentry at a similar price to the PM.
What this tells me, though, is that the G4 must be seriously starved since it's working with less memory and disk performance, yet in Alivec capable tasks (FCP, DVD authoring, it manages to keep up pretty well, on some things, not, by any means, all things).
Look at the recent Xserver benchmarks. Even with the psuedo-DDR implementation, the fast disk system and, at least, the maximum throughput available to the CPU's unencumbered by the demands of the PCI, AGP, or ATA buses, allows the lowly G4 1Ghz to shine quite brightly against it's competition. It suggests that with a little system improvement a G4 can still get the job done.
Now think QE (which I must remind everyone I predicted way back in the days when Kormac was spewing 'Raycer' ). Even developments on the windows side suggest that the GPU will play a much greater role in system performance and number crunching ability in the next couple of years. In light of this, the Xserver 'hack' may be of great importance to ALL of Apples systems, the majority of which will have to live with G4's that may or may not get buses much faster than MPX 166 (mentioned by Mot somewhere). That may seem like only a 33% improvement to the CPU throughput, but if Apple had a system I/O controller that ran system memory independtly (which they do, in the Xserver) the picture is actually quite good! As long as the best (and most programmable) video cards are available (and working with QE) over the fastest connection possible to system RAM -- be that AGP 4x, 8x, 16x, 3GIO, hyper-transport, RapidIO, whatever, see industry standard and copy accordingly -- well then it doesn't really matter how fast the CPU bus is, does it?
None will interfere with the bandwidth available to the others. Video Bus, CPU bus, Disk Bus, PCI Bus -- all will have AMPLE bandwidth, especially the disks and GPU (which will be the real key).
While I believe Powermacs will move to a processor that doesn't require such a 'hack', this is a great thing for the rest of Mac-dome. All the computers need a system I/O-northbridge/southbridge etc etc... putting as much of that down on one chip as possible should make it ultimately cheaper to do, and simultaneously ease the pressure during PPC development lulls -- at least the rest of the system can move forward/minimize the CPU shortfall.
G4/5 with 'real' DDR/rapidIO on chip for the PM.
plus,
133/166MPX bridged to an Xserver type memory bus for everything else.
Equals,
A pretty good line-up.
Sorry for the long rant.
_________________
***PMS. P4 is NOT an inferior design. It may give up some IPC, but it throws a whole whack of extra C's at the problem. Coupled with good memory throughput that makes a fast P4 the fastest system money can buy (out of athlon, PPC, and P4) in virtually any timed test/benchmark (altivec excluded). You don't feel how efficiently your CPU solves a problem, you feel only how quickly or slowly it does so. Not liking something because it takes too many clocks to accomplish something is retarded. Seconds should be your only measure, and in that respect the P4 does better than anybody else.
<strong>P4 is NOT an inferior design. It may give up some IPC, but it throws a whole whack of extra C's at the problem. Coupled with good memory throughput that makes a fast P4 the fastest system money can buy (out of athlon, PPC, and P4) in virtually any timed test/benchmark (altivec excluded). You don't feel how efficiently your CPU solves a problem, you feel only how quickly or slowly it does so. Not liking something because it takes too many clocks to accomplish something is retarded. Seconds should be your only measure, and in that respect the P4 does better than anybody else.</strong><hr></blockquote>
I agree, and no, thanks, I don't think I'm retarded.
My point simply is that P4s @ 2.26 and 2.53 GHz coupled with RDRAM 1066 are as yet unavailable. Intel is the Evil Satan of Paper Launches.
Furthermore, P4 Xeons aren't selling anywhere near as good as the PIII Xeons, which are still in production and running at 1.4 GHz (Tualatin core IIRC). I'd really love to see one of those babies benchmarked on a good DDR2700 or dual-channel Rambus 800 system. Saying that the 2.53 P4 + RDRAM 1066 (which BTW is officially unsupported by Intel) is the fastest system around is like saying that the G5 1.8 (or whatever) is just as fast or faster: I still gotta see them...
Not only do they lack credibility, but I have never seen so many people who are totally incapable of learning from past mistakes / precedent. Baffling.
<strong>Moto have optimized their 130 nm fab and the next Powermac revision will bear the fruits of this labor. While migrating the G4 to a new fab, Moto has taken the time to incorporate a number of changes, the most significant of which is the bus. RapidIO is the new face of the G4, and of course with it comes full DDR support. What Apple finally will use is unknown to me, but test mules have employed both PC2100 and
--Lymphoid out</strong><hr></blockquote>
Lymphoid, I'm amazed - I've been thinking along these same lines for the last coupla days - that maybe MOT has been 'migrating' the G4 to BookE compliance, and in the process aalowing the G4 to pick up a lot of "G5" tech, such as you mention.
...Which would go a long way toward explaining moki's cryptic "these aren't the G5s you're looking for" - since this chip would not, in fact, BE a G5, but a hybrid of the two tech 'layers'.
It matters little what Apple or Stev Jobs wants vis a vis processors because they dont control the fab.
The reason why it isw safe to say "There is no g5 (in the near future) is ecause there is no evidence to support anything new coming out of Moto's fab any time soon.
Processor development is not a secretive process like sexy plastic shapes are.....anyone (and believe me, I'm not one of them) can scope out the processor forums, the development white papers and all the other engineering gobbledegook and get a fairly good idea of what's coming.
Before any of the Pentium cores came out, everyone knew what was coming. The same with the AMD series (K, Athlon, Duro, etc.). We also knew about the next iteration of IBM's POWER series. Also true with the G-whatever chips from Motorola, including all the variants of the G4 to date.
So compare that situation with today's: We know all about the next generation of x86 chips coming from the big two, we know all about the new chips coming in Sun's stuff, in SGI's stuff, and even IBM releases info about it's nextgen PowerPC, not to mention the G-frakin-3 upgrades.
But from Motorola? NAry a peep about a G5. This lack of hard (or even semi-hard) information makes many people think that yhr G5 is being fabbed in super-secrecy, and that Motorola is keeping it under wraps at Jobs' orders.
Which is crap, from Motorola's POV. The reason why chip companies are as forthcoming as possible about their engineering developments as possible is to keep their customers apprised of what's coming down the pipe, so they can adjust their business plans accordingly. Not to mention the corporate pretige involved in publishing the stuff about the next New Badass stuff.
I know people aren't going to take my counterpoint seriously because I'm taken as a mean ol' crank around here, but the simple fact of the matter is that Moto's semiconductor biz has been in trouble for a few years running, they have suffered enormous brain drain. Furthermore, they decidedon a difficult-to-scale processor design (the low-pipeling G) to be their flagship design. Finally, the bottom has just DROPPED out of the embedded proc biz (servers, routers, etc.), giving the bean counters even further pause in approving new designs to sell in a market that shows littel signs of rcovery lately.
Strangely, none of this bothers me, because of the fact that small improvemnts to the current G4 line-up can go a long way. Implementing a DDR-capable FSB version of the G4 is the first one, and all this hoohah about RIO and HT sem to set the table for significant performance improvements in the near future. Even with the problems at Moto, this is not a radical change to the core. Ading some SOI sauce to the mix, and other fab techs will keep the G4 in the game for a while.
Personally, it seems likely to me that G4 is here to stay, and Apple won't be done with it for some time.
And the last thing I'll say is this: there will NEVER be a Pentium-killer like many around here constantly whine for......Intel specializes in bringing the desktop and laptop power, and they face a real competition from AMD. Motorola is a glorified DSP maker some of whose topend designs can be applied to the desktop arena. It's no contest.
Nw, if we were talking IBM, maybe that would be different. ut even IBM has problems, the least of which is that they don't seem to really care about the desktop PPC market. But they do still make the insane POWER chips, and there is the Cell project, so who knows?
Flame away, kids, and tell me I don't knw my elbow from my assho1e......I'll readily admit it. But I do remember my history, and history shows us that a proc company that goes silent in the engineering arena, goes under soon after. Rightnow is NOT a good time to believe Motorola has anything up their sleeves except their arms.
You make a reasonable point ting5, but I think that you are also making generalisations about what may or may not be in the pipeline. The fact that both Motorola and IBM have published information related to their respective bookE compliant chips (the 8540 from Mot and 440? from IBM) leads me to the conclusion that these things certainly are in the pipeline (and I know that the 8540 only exists on paper). By your own logic about what we can expect based upon microprocessor presentations and white papers etc, I think it is pretty much guaranteed that a future (don't know when!) PowerMac will be based on a offering such as this...
<strong>Before any of the Pentium cores came out, everyone knew what was coming. The same with the AMD series (K, Athlon, Duro, etc.). We also knew about the next iteration of IBM's POWER series. Also true with the G-whatever chips from Motorola, including all the variants of the G4 to date.
So compare that situation with today's: We know all about the next generation of x86 chips coming from the big two, we know all about the new chips coming in Sun's stuff, in SGI's stuff, and even IBM releases info about it's nextgen PowerPC, not to mention the G-frakin-3 upgrades.
But from Motorola? NAry a peep about a G5. This lack of hard (or even semi-hard) information makes many people think that yhr G5 is being fabbed in super-secrecy, and that Motorola is keeping it under wraps at Jobs' orders.
Strangely, none of this bothers me, because of the fact that small improvemnts to the current G4 line-up can go a long way. Implementing a DDR-capable FSB version of the G4 is the first one, and all this hoohah about RIO and HT sem to set the table for significant performance improvements in the near future. Even with the problems at Moto, this is not a radical change to the core. Ading some SOI sauce to the mix, and other fab techs will keep the G4 in the game for a while.</strong><hr></blockquote>
A comment, and then a question (what a surprise, huh?).
First, I do agree with part of your post, in that I find it strange that ther is no noise coming from Moto about upcoming CPU's. It is a very competitive industry, and the manu's need to be very vocal to gain more press and spotlight their products. That said, we need to keep one thing in mind....Moto has no competition in the desktop market. Therefore, they don't need to spotlight anything except their embedded products (which they seem to do quite well). All the other companies you mentioned are competing in HEAVILY saturated markets.
Now, the question.....
IF Moto has really gone the "design your own" route, what would we hear? The new CPU (4+, 4.5, 5, whatever) would actually be an Apple designed chip, with their own wish-list in place. This would mean that there is nothing for Moto to announce.
Is this a reasonable, alternative explanation?
The other possibility that occurs to me (quite a bit less likely, I would think....but we can always hope), is that Moto has decide that the best route to go is some, as yet undisclosed, new proccess. In this case it might be wise to keep it very quiet until it's ready for release, to get a real headstart on the market. Moto seems to have quite a few new proccesses that they've been researching, so this has could be the case as well.
I shouldn't attempt to post before morning coffee.....jeezus, I haven't made that many spelling error since the first grade.....Some of the words are unrecognizable.
Thanks for adding another grown-up voice to the mix...tho I doubt you'll be thanked by many.
You've described my 14-year-old pretty accurately; she'll get in her head that something's not going to work out & get all grumpy and sullen and disagreeable WAY before anything actually happens. This much annoys both her mothers and me, and has occasionally caused her to screw herself out of something. That really pitches the fit to a new level!
In exactly the same way, so many people simply "decide" that Apple is screwed, that nothing's ever gonna happen again, it's all a waste, it's all for nothing, it's all MOT's / Apple's / Jobs' / Sculley or Amelio's or Gates' fault - and it's never ever EVERgonna change. And - just like teen-agers - mostly these folks take it personally if someone suggests another way of evaluating the situation, and especially when they don't agree with each other!
No wonder so much of the "conversation" here sounds like squally babies ready for a nap!
Anyway, thanks for your excellent post, and I hope that you and other seasoned lurkers will take a more active part in the proceedings around here: we could use the balance!
Of course, this makes it seem like a shrinking dye directly translates in more Mhz, which isn't true.</strong><hr></blockquote>
Um, .13 microns from .18 microns is a shrinkage in only one important dimension, right? Only the diameter of the transistors are changing and since these transistors are all on one plane, the only useful shrinkage should be somewhat proportional to my claim...?
Um, .13 microns from .18 microns is a shrinkage in only one important dimension, right? Only the diameter of the transistors are changing and since these transistors are all on one plane, the only useful shrinkage should be somewhat proportional to my claim...?</strong><hr></blockquote>
Well if the "wires" all ran in one direction that would be true, but they actually run in 2 (and a half - multiple layers), so a thinner wire makes for closer wires in all dimensions that you are using.
However I cannot understand TING5. if he's right then apple ARE screwed but I can't believe it. With all that money and power Apple MUST be able to do something - not just hope on Moto. Hell, if I was that rich I would walk into boardrooms and keep battering the crap out of CEos until I got what I wanted. Imagine being Steve, having to wait endlessly on others. if even a fraction of the anecdotes about him were true then this Guy would never just sit back and let things be decided by others out of his control. We need a pentium killing processor. Mac users know it, apple MUST know it and I bet it keeps Steve awake nights. Would he really bet his entire reputation and the future of apple is he thought Moto couldn't or wouldn't deliver? I wouldn't. hell, I'd burn Moto to the ground before I let those b*st*rds screw with my company.
If anyone believes that apple haven't considered their fate at the hands of an incompetent Moto then it THEY that believe that apple is doomed.
Now I have to go and take my medication.
BTW did you know that a 2nd user 8x scsi internal cd rom for a beige powermac retails for around 80 gb pounds here in the uk - from so called "bargain" mac sellers? man I wish I lived in the US.
Well if the "wires" all ran in one direction that would be true, but they actually run in 2 (and a half - multiple layers), so a thinner wire makes for closer wires in all dimensions that you are using.</strong><hr></blockquote>
Yes, that affects the overall area of the chip, which is proportional to the sqauer of the minimum feature size. The speed however is limited by one of two things, the speed of the individual transistors, or the delay between circuit elements. MOS transistor speed is (to a first approximation) linearly dependent on the gate length which is usually equal to the mimium feature size (halve the gate length, halve the switching time), signal delay is also dependent on a single linear dimension so tends to be linearly dependent on feature size.
Therefore, chip speed (very approxiamtely) is linearly dependent on the minimum feature size, not it's square.
I agree, and no, thanks, I don't think I'm retarded.
My point simply is that P4s @ 2.26 and 2.53 GHz coupled with RDRAM 1066 are as yet unavailable. Intel is the Evil Satan of Paper Launches.
Furthermore, P4 Xeons aren't selling anywhere near as good as the PIII Xeons, which are still in production and running at 1.4 GHz (Tualatin core IIRC). I'd really love to see one of those babies benchmarked on a good DDR2700 or dual-channel Rambus 800 system. Saying that the 2.53 P4 + RDRAM 1066 (which BTW is officially unsupported by Intel) is the fastest system around is like saying that the G5 1.8 (or whatever) is just as fast or faster: I still gotta see them...
ZoSo</strong><hr></blockquote>
I have a dual PIV Xenon 2.4 GHz with 3G RAM coming in in a week or so...ya want me to run some tests It's only running a 400 MHz front side bus with interleaved memory :eek: This should be a little bit faster than our dual PIII 933!
Comments
<strong>Moto have optimized their 130 nm fab and the next Powermac revision will bear the fruits of this labor.</strong><hr></blockquote>
Didn't Motorola say that they'd have switched to the .13µ process only with the G5? I remember reading about it a while ago, but can't remember where... Can somebody correct me on this?
[quote]<strong>Clock speeds are impressive. Without deepening the G4?s pipeline, Moto has attained clock speeds other CPU manufacturers only dream of. We?ve seen boxes ranging from 1.0 1.6 GHz, however, most likely is a 1.4 GHz high end.</strong><hr></blockquote>
The newest Athlon 2200+ is "only" a 1800MHz part: everybody is well aware of the P4 dirty tricks to achieve such monster clock-rates... In comparison the PIII is a much better design, and currently tops out at 1400MHz. So I think that these speed figures are impressive indeed, given tha G4 IPC efficiency.
RapidIO: what?!?!? Are we already there??? Wow... This part could very well be true, but in my mind RapidIO belongs to the realm of the "maybe-in-a-distant-future-or-probably-never" technologies, like widespread use of H-powered cars... I'd love to see it happen though...
I don't know, real insider info or just a very good educated guess?
ZoSo
<strong>1.4 GHz is definitely in the realm of possibility for a .13 micron G4.
This may be fuzzy logic since I don't design chips but going on the fact that Motorola had an orderable part number for a 1.1 GHz MPC7455 on one of its webpages, and that .18 microns is ~1.385x.13...
1.385x1.1 = >1.5 GHz. 1.4 GHz doesn't seem that hard, though I guess AMD went to .13 and only went up to 1.8 GHz from 1.733.</strong><hr></blockquote>
You forget that chips shrink in two dimensions:
1.385^2 x 1.1 = 2,1
Of course, this makes it seem like a shrinking dye directly translates in more Mhz, which isn't true.
And as far as rumors go, it seems that too many people put up shyt on these boards that they completely makeup. I can start a thread about new g5s with whatever specs, but it doesn't make it true. The boards here lack credibility. We have the right to be skeptical. I don't like people jerking me around.
i do think there is a g5. i don't think it'll be ready for a while tho (sf, ny nxt yr, ..ad infinitum..?). I'm not an insider, tho, just a pessimist. [prove me wrong apple]
IPC is interesting, but P4 is still the performance king.*** Read on, see what I mean.
Recent tests of DP Athlon 1.6 vs DP 1Ghz Powermacs in 3-d and video effects, put the athlon box at about 60% faster. Which, outside of altivec, puts the IPC of both 'systems' about the same. BUT, the Athlon was running a much faster bus and disk system. So these tests don't really compare chips so much as they compare systems. You can't fault the Athlon box makers for including superior componentry at a similar price to the PM.
What this tells me, though, is that the G4 must be seriously starved since it's working with less memory and disk performance, yet in Alivec capable tasks (FCP, DVD authoring, it manages to keep up pretty well, on some things, not, by any means, all things).
Look at the recent Xserver benchmarks. Even with the psuedo-DDR implementation, the fast disk system and, at least, the maximum throughput available to the CPU's unencumbered by the demands of the PCI, AGP, or ATA buses, allows the lowly G4 1Ghz to shine quite brightly against it's competition. It suggests that with a little system improvement a G4 can still get the job done.
Now think QE (which I must remind everyone I predicted way back in the days when Kormac was spewing 'Raycer'
None will interfere with the bandwidth available to the others. Video Bus, CPU bus, Disk Bus, PCI Bus -- all will have AMPLE bandwidth, especially the disks and GPU (which will be the real key).
While I believe Powermacs will move to a processor that doesn't require such a 'hack', this is a great thing for the rest of Mac-dome. All the computers need a system I/O-northbridge/southbridge etc etc... putting as much of that down on one chip as possible should make it ultimately cheaper to do, and simultaneously ease the pressure during PPC development lulls -- at least the rest of the system can move forward/minimize the CPU shortfall.
G4/5 with 'real' DDR/rapidIO on chip for the PM.
plus,
133/166MPX bridged to an Xserver type memory bus for everything else.
Equals,
A pretty good line-up.
Sorry for the long rant.
_________________
***PMS. P4 is NOT an inferior design. It may give up some IPC, but it throws a whole whack of extra C's at the problem. Coupled with good memory throughput that makes a fast P4 the fastest system money can buy (out of athlon, PPC, and P4) in virtually any timed test/benchmark (altivec excluded). You don't feel how efficiently your CPU solves a problem, you feel only how quickly or slowly it does so. Not liking something because it takes too many clocks to accomplish something is retarded. Seconds should be your only measure, and in that respect the P4 does better than anybody else.
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
Lymphoid: I am encouraged by your speculation.
All: What has happened to the idea of quad-processor machines?
Hope springs eternal,
Mandricard
AppleOutsider
Are you trying to give me a cranial haemorrhage?
[ 06-28-2002: Message edited by: Matsu ]</p>
<strong>P4 is NOT an inferior design. It may give up some IPC, but it throws a whole whack of extra C's at the problem. Coupled with good memory throughput that makes a fast P4 the fastest system money can buy (out of athlon, PPC, and P4) in virtually any timed test/benchmark (altivec excluded). You don't feel how efficiently your CPU solves a problem, you feel only how quickly or slowly it does so. Not liking something because it takes too many clocks to accomplish something is retarded. Seconds should be your only measure, and in that respect the P4 does better than anybody else.</strong><hr></blockquote>
I agree, and no, thanks, I don't think I'm retarded.
My point simply is that P4s @ 2.26 and 2.53 GHz coupled with RDRAM 1066 are as yet unavailable. Intel is the Evil Satan of Paper Launches.
Furthermore, P4 Xeons aren't selling anywhere near as good as the PIII Xeons, which are still in production and running at 1.4 GHz (Tualatin core IIRC). I'd really love to see one of those babies benchmarked on a good DDR2700 or dual-channel Rambus 800 system. Saying that the 2.53 P4 + RDRAM 1066 (which BTW is officially unsupported by Intel) is the fastest system around is like saying that the G5 1.8 (or whatever) is just as fast or faster: I still gotta see them...
ZoSo
Not only do they lack credibility, but I have never seen so many people who are totally incapable of learning from past mistakes / precedent. Baffling.
<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
<strong>Moto have optimized their 130 nm fab and the next Powermac revision will bear the fruits of this labor. While migrating the G4 to a new fab, Moto has taken the time to incorporate a number of changes, the most significant of which is the bus. RapidIO is the new face of the G4, and of course with it comes full DDR support. What Apple finally will use is unknown to me, but test mules have employed both PC2100 and
--Lymphoid out</strong><hr></blockquote>
Lymphoid, I'm amazed - I've been thinking along these same lines for the last coupla days - that maybe MOT has been 'migrating' the G4 to BookE compliance, and in the process aalowing the G4 to pick up a lot of "G5" tech, such as you mention.
...Which would go a long way toward explaining moki's cryptic "these aren't the G5s you're looking for" - since this chip would not, in fact, BE a G5, but a hybrid of the two tech 'layers'.
Call it G4.5....
It matters little what Apple or Stev Jobs wants vis a vis processors because they dont control the fab.
The reason why it isw safe to say "There is no g5 (in the near future) is ecause there is no evidence to support anything new coming out of Moto's fab any time soon.
Processor development is not a secretive process like sexy plastic shapes are.....anyone (and believe me, I'm not one of them) can scope out the processor forums, the development white papers and all the other engineering gobbledegook and get a fairly good idea of what's coming.
Before any of the Pentium cores came out, everyone knew what was coming. The same with the AMD series (K, Athlon, Duro, etc.). We also knew about the next iteration of IBM's POWER series. Also true with the G-whatever chips from Motorola, including all the variants of the G4 to date.
So compare that situation with today's: We know all about the next generation of x86 chips coming from the big two, we know all about the new chips coming in Sun's stuff, in SGI's stuff, and even IBM releases info about it's nextgen PowerPC, not to mention the G-frakin-3 upgrades.
But from Motorola? NAry a peep about a G5. This lack of hard (or even semi-hard) information makes many people think that yhr G5 is being fabbed in super-secrecy, and that Motorola is keeping it under wraps at Jobs' orders.
Which is crap, from Motorola's POV. The reason why chip companies are as forthcoming as possible about their engineering developments as possible is to keep their customers apprised of what's coming down the pipe, so they can adjust their business plans accordingly. Not to mention the corporate pretige involved in publishing the stuff about the next New Badass stuff.
I know people aren't going to take my counterpoint seriously because I'm taken as a mean ol' crank around here, but the simple fact of the matter is that Moto's semiconductor biz has been in trouble for a few years running, they have suffered enormous brain drain. Furthermore, they decidedon a difficult-to-scale processor design (the low-pipeling G) to be their flagship design. Finally, the bottom has just DROPPED out of the embedded proc biz (servers, routers, etc.), giving the bean counters even further pause in approving new designs to sell in a market that shows littel signs of rcovery lately.
Strangely, none of this bothers me, because of the fact that small improvemnts to the current G4 line-up can go a long way. Implementing a DDR-capable FSB version of the G4 is the first one, and all this hoohah about RIO and HT sem to set the table for significant performance improvements in the near future. Even with the problems at Moto, this is not a radical change to the core. Ading some SOI sauce to the mix, and other fab techs will keep the G4 in the game for a while.
Personally, it seems likely to me that G4 is here to stay, and Apple won't be done with it for some time.
And the last thing I'll say is this: there will NEVER be a Pentium-killer like many around here constantly whine for......Intel specializes in bringing the desktop and laptop power, and they face a real competition from AMD. Motorola is a glorified DSP maker some of whose topend designs can be applied to the desktop arena. It's no contest.
Nw, if we were talking IBM, maybe that would be different. ut even IBM has problems, the least of which is that they don't seem to really care about the desktop PPC market. But they do still make the insane POWER chips, and there is the Cell project, so who knows?
Flame away, kids, and tell me I don't knw my elbow from my assho1e......I'll readily admit it. But I do remember my history, and history shows us that a proc company that goes silent in the engineering arena, goes under soon after. Rightnow is NOT a good time to believe Motorola has anything up their sleeves except their arms.
ting5
Or then again a filtered down POWER4...
Or something else...
Maybe.
<strong>Before any of the Pentium cores came out, everyone knew what was coming. The same with the AMD series (K, Athlon, Duro, etc.). We also knew about the next iteration of IBM's POWER series. Also true with the G-whatever chips from Motorola, including all the variants of the G4 to date.
So compare that situation with today's: We know all about the next generation of x86 chips coming from the big two, we know all about the new chips coming in Sun's stuff, in SGI's stuff, and even IBM releases info about it's nextgen PowerPC, not to mention the G-frakin-3 upgrades.
But from Motorola? NAry a peep about a G5. This lack of hard (or even semi-hard) information makes many people think that yhr G5 is being fabbed in super-secrecy, and that Motorola is keeping it under wraps at Jobs' orders.
Strangely, none of this bothers me, because of the fact that small improvemnts to the current G4 line-up can go a long way. Implementing a DDR-capable FSB version of the G4 is the first one, and all this hoohah about RIO and HT sem to set the table for significant performance improvements in the near future. Even with the problems at Moto, this is not a radical change to the core. Ading some SOI sauce to the mix, and other fab techs will keep the G4 in the game for a while.</strong><hr></blockquote>
A comment, and then a question (what a surprise, huh?).
First, I do agree with part of your post, in that I find it strange that ther is no noise coming from Moto about upcoming CPU's. It is a very competitive industry, and the manu's need to be very vocal to gain more press and spotlight their products. That said, we need to keep one thing in mind....Moto has no competition in the desktop market. Therefore, they don't need to spotlight anything except their embedded products (which they seem to do quite well). All the other companies you mentioned are competing in HEAVILY saturated markets.
Now, the question.....
IF Moto has really gone the "design your own" route, what would we hear? The new CPU (4+, 4.5, 5, whatever) would actually be an Apple designed chip, with their own wish-list in place. This would mean that there is nothing for Moto to announce.
Is this a reasonable, alternative explanation?
The other possibility that occurs to me (quite a bit less likely, I would think....but we can always hope), is that Moto has decide that the best route to go is some, as yet undisclosed, new proccess. In this case it might be wise to keep it very quiet until it's ready for release, to get a real headstart on the market. Moto seems to have quite a few new proccesses that they've been researching, so this has could be the case as well.
I shouldn't attempt to post before morning coffee.....jeezus, I haven't made that many spelling error since the first grade.....Some of the words are unrecognizable.
Stev? ecause? isw? frakin? apprised? yhr? pretige? sem? topend? ut? knw?
sheesh.
ting5
Thanks for adding another grown-up voice to the mix...tho I doubt you'll be thanked by many.
You've described my 14-year-old pretty accurately; she'll get in her head that something's not going to work out & get all grumpy and sullen and disagreeable WAY before anything actually happens. This much annoys both her mothers and me, and has occasionally caused her to screw herself out of something. That really pitches the fit to a new level!
In exactly the same way, so many people simply "decide" that Apple is screwed, that nothing's ever gonna happen again, it's all a waste, it's all for nothing, it's all MOT's / Apple's / Jobs' / Sculley or Amelio's or Gates' fault - and it's never ever EVER gonna change. And - just like teen-agers - mostly these folks take it personally if someone suggests another way of evaluating the situation, and especially when they don't agree with each other!
No wonder so much of the "conversation" here sounds like squally babies ready for a nap!
Anyway, thanks for your excellent post, and I hope that you and other seasoned lurkers will take a more active part in the proceedings around here: we could use the balance!
<strong>
You forget that chips shrink in two dimensions:
1.385^2 x 1.1 = 2,1
Of course, this makes it seem like a shrinking dye directly translates in more Mhz, which isn't true.</strong><hr></blockquote>
Um, .13 microns from .18 microns is a shrinkage in only one important dimension, right? Only the diameter of the transistors are changing and since these transistors are all on one plane, the only useful shrinkage should be somewhat proportional to my claim...?
<strong>
Um, .13 microns from .18 microns is a shrinkage in only one important dimension, right? Only the diameter of the transistors are changing and since these transistors are all on one plane, the only useful shrinkage should be somewhat proportional to my claim...?</strong><hr></blockquote>
Well if the "wires" all ran in one direction that would be true, but they actually run in 2 (and a half - multiple layers), so a thinner wire makes for closer wires in all dimensions that you are using.
Now, from the sublime to the ridiculous:
However I cannot understand TING5. if he's right then apple ARE screwed but I can't believe it. With all that money and power Apple MUST be able to do something - not just hope on Moto. Hell, if I was that rich I would walk into boardrooms and keep battering the crap out of CEos until I got what I wanted. Imagine being Steve, having to wait endlessly on others. if even a fraction of the anecdotes about him were true then this Guy would never just sit back and let things be decided by others out of his control. We need a pentium killing processor. Mac users know it, apple MUST know it and I bet it keeps Steve awake nights. Would he really bet his entire reputation and the future of apple is he thought Moto couldn't or wouldn't deliver? I wouldn't. hell, I'd burn Moto to the ground before I let those b*st*rds screw with my company.
If anyone believes that apple haven't considered their fate at the hands of an incompetent Moto then it THEY that believe that apple is doomed.
Now I have to go and take my medication.
BTW did you know that a 2nd user 8x scsi internal cd rom for a beige powermac retails for around 80 gb pounds here in the uk - from so called "bargain" mac sellers? man I wish I lived in the US.
<strong>
Well if the "wires" all ran in one direction that would be true, but they actually run in 2 (and a half - multiple layers), so a thinner wire makes for closer wires in all dimensions that you are using.</strong><hr></blockquote>
Yes, that affects the overall area of the chip, which is proportional to the sqauer of the minimum feature size. The speed however is limited by one of two things, the speed of the individual transistors, or the delay between circuit elements. MOS transistor speed is (to a first approximation) linearly dependent on the gate length which is usually equal to the mimium feature size (halve the gate length, halve the switching time), signal delay is also dependent on a single linear dimension so tends to be linearly dependent on feature size.
Therefore, chip speed (very approxiamtely) is linearly dependent on the minimum feature size, not it's square.
Michael
<strong>
I agree, and no, thanks, I don't think I'm retarded.
My point simply is that P4s @ 2.26 and 2.53 GHz coupled with RDRAM 1066 are as yet unavailable. Intel is the Evil Satan of Paper Launches.
Furthermore, P4 Xeons aren't selling anywhere near as good as the PIII Xeons, which are still in production and running at 1.4 GHz (Tualatin core IIRC). I'd really love to see one of those babies benchmarked on a good DDR2700 or dual-channel Rambus 800 system. Saying that the 2.53 P4 + RDRAM 1066 (which BTW is officially unsupported by Intel) is the fastest system around is like saying that the G5 1.8 (or whatever) is just as fast or faster: I still gotta see them...
ZoSo</strong><hr></blockquote>
I have a dual PIV Xenon 2.4 GHz with 3G RAM coming in in a week or so...ya want me to run some tests