Originally posted by wizard69
It is not that we don't want to hear it it is just a matter of the information being wrong. Sure there have been problems with 90nm, and IBM's process could hardly becalled innovative but that does mean we won't see progress.
Are there physical limits to how fast a CPU can operate - most certainly. The problem is that we are far from those limits, the problems today are almost universally thermal and can be dealt with.
The problem is that the innovations on the horizon don't hold much promise. At best the pending process improvements look like they could deliver ~30% improvement in clock rate (based on old promises of more significant material changes). From 2.5 GHz that'll get the 970FX to ~3.2 GHz, and that will realize less than a 30% improvement in performance. And that is a wildly optimistic prediction -- sure they've got 3.4 GHz in the lab with super-duper cooling systems on sampled parts, but that's a far cry from what you could reasonably ship in hundreds of thousands of units in a consumer level product.
Intel doesn't seem to think they'll achieve even that much of an improvement, and AMD's first 90nm chips have a zero
clock rate improvement.
Maybe not so much a troll as somebody that maybe is being mis informed for various reasons.
Well if you can quote current industry publications that will inform us correctly, then please do.
Ah no I can't accept it because the premiss doesn't support what is happening in the industry. If I saw every manufacture give up on trying to produce smaller and faster trasistors I'd say we may be on to something. This isn't the case at all, corporations are still pursuing much faster devices.
Like who? If Intel (the leading clock rate advocate for the last 15 years) publicly abandons the pursuit of the all powerful clock rate bump, then who are you talking about? Freescale, who is still hanging around at a mere 1.5 GHz?
Besides, the pursuit of smaller transistors is not synonymous with the pursuit of higher clock rates. Smaller means you can fit more in the same space. More transistors means more cache, more memory, more cores. We're not suggesting that miniturization will stop (quite the opposite). Instead we're saying that the benefits of increasing frequency have reached or passed the point where they outweight the negatives.
One should not get side tracked by multicore processors either, this is just an out growth of having the hardware available to implement them. Multicore chips are not an excuse for poor single core performance, rather they are a avenue to increasing SYSTEM performance.
Time for a paradigm shift.
You may not be making such a decree but Nr9 certainly is trying to. This is what people are rejecting
Nr9 seems to "suffer" from communication issues. He seems to place lower value on clear and articulate conversation in these forums than I do -- but then I work with a lot of brilliant engineers who do much the same thing, so I'm not going to write him off purely on that evidence alone. I tend to be considered more credible, but please keep in mind that what I'm better at is communicating
technical issues (or perhaps I'm just a better typist) and this is not the same as having a better knowledge base. I'm not saying Nr9 is the lead hardware engineer at the IBM's processor design facility, or even that I'm convinced he works at IBM! However, what he is saying does resonate with what I
see in the industry and know about recent developments.
With the initial transistion to 90nm and the lack of process development this may be the case. The reality is that manufactures have no choice but to find a way around this issue.
Not everything has a way around. The way around may very well be approaches other than increasing clock frequency.
We have heard about such issues since the industry started. Heck I've been following this since Byte was a young magazine. At each point in the development of the industry somebody has dealt with the supposed limits of the time. There is nothing to keep that fm happening again.
Yeah, I'm old too. I remember everybody scoffing at Intel's wild predictions of 100 MHz. This time things feel different. This time the old tricks have been stretched so far that we've been seeing seriously diminishing returns for a few years. They are going to have to work a lot harder to continue making progress.... and they will. I'm not saying anybody is going to stop trying to make faster processors, but the basic they way they are going to go about it has to change for the first time. Can't just make it smaller and increase the frequency, decrease the power and reap the benefits anymore.
It is almost a given that IBM is working on lowering the power used by the 970 series. You are implying, along with Nr9 apparently, that they have given up on new technology that may be applied at 90nm.
No, I'm saying that the benefit to be gained by this new technology is limited. IBM will have lower power versions of the 970, but I don't think they're going to run at significantly higher clock rates. In fact at the top clock rates they may not be lower power at all.
Or maybe the approach that the team Freescale is involved with is leading them to more confidence. After all they have taken their time and do have the knowledge of the rest of the industiers public failings. It is not like there is only one 90nm process in this world and every one is being compelled to use it. There is still room in this world for novel approaches and innovation.
You said Freescale and confidence in the same sentence!!!
Freescale is at least a year from a production 90nm part by their own admission. And they are talking about taking a 1.5 GHz part to... 1.5+ GHz. Their later parts don't even exist yet. Sounds like IBM (and Intel) before they actually got the point of production. IBM and Intel have different processes at 90nm, but they suffer from the same problem. If I were a betting man would I put money on Freescale
coming up with a solution to something that IBM and Intel's best haven't? In the next year if a solution does come out of the ether, I'll be that it comes from IBM and/or Intel. But I don't expect it.
I do believe that many have been very fair with Nr9. He either has to offer more compelling evidence or modify his approach. To state absolutely that we have hit a wall and 2.5GHz is it, is just to much. IBM may have recently stumbled and fallen flat on their face but that doesn't mean they can't stand up again. Frankly they have to stand up again or be left in the position of watching others walk away from them technology wise.
Yes, I'd like to see evidence as well. Unfortunately this industry is buried under NDAs and trade secrets. They don't publish convenient web links to their innermost secrets so that us rumor mongers can point at them. I'd settle (happily) for clear evidence that you are correct as well... but I don't see anybody shipping significant clock rate bumps based on the 90nm migration. Intel 0.4 GHz, AMD 0.0 GHz, IBM 0.5 GHz. Wow, IBM stumbled and fell flat but still turned in the best improvement (yes, this is facetious comment).
And one last comment: I interpret "the limit" that Nr9 speaks of as something other than a speed barrier like the speed of sound or the speed of light. In engineering, something like processor design is a huge set of trade-offs, product requirements, and goals. In this very complex design space there is a vague boundry of what makes a possible and acceptable
design. The available technology governs this design space, and physics governs the technology. It is simply not correct to say that 2.55178 GHz is the limit for the PowerPC. It is not even correct to say that about the PowerPC 970FX on process XYZ at voltage V, because it depends on what kind of system parameters, operating conditions, software, etc etc etc. So instead the limit being reached is really just a recognition by processor designers that they are flogging a dead horse to try and achieve their (very complex set of) goals, and new avenues to improve processor design must be followed.