why the hell would I talk about increasing voltage. of course im talking about an alternative
worrying about english gets in the way of doing the real technical stuff with the most efficiency possible. there is no free lunch. energy you spend constructing your sentences can be used for your technical engineering shit.
Sun just announced that the UltraSparc IV+ went 90 nm and gained a quite a lot frequenzy wise.
From 1.2 GHz to 1.8 GHz by doing the 130 nm to 90 nm dance we are all talking about.
So.. What is Texas Instrument's secret of doing a 50% increase in speed? Was their former fab really bad and their new really great, or is their 1.8 GHz claim just a lie or wishful thinking?
I think the secret here is that 1.2 GHz wasn't pushing the 130 nm envelope.
worrying about english gets in the way of doing the real technical stuff with the most efficiency possible. there is no free lunch. energy you spend constructing your sentences can be used for your technical engineering shit.
I hate to break it to you Nr9, but you need to live and work in a world populated by humans. The sooner you learn that the energy you put into inter-personal communication is valuable, the sooner you'll find that you have a career path. If you can't see that then its your loss, not anyone else's.
Nonetheless this is an interesting discussion, so let's get back to it. The processors that have "done well" on 90 nm seem to be those that are not pushing the frequency limits. Intel's Dothan, AMD's first 90nm, IBM's 2 GHz 970FX, and the new UltraSparc all demonstrate considerably improved thermal performances. The trouble only seems to occur with the frequency leaders -- Prescott and the 2.5+ GHz 970FX. This actually supports Nr9's statements (allowing for his tendency to make them sound extreme): the frequency scaling benefits of at 90 nm seem to be non-existant on the 2.5+ GHz chips. This makes me seriously wonder about a couple of high profile high clock rate chips that are supposed to be showing up in 2005/2006...
What is Prescott's speed? I mean, if Intel gets to 3.* GHZ at 90nm and IBM gets to 2.5+ GHZ then I would have to say 2.5 GHZ isn't the 90nm barrier Nr9 is claiming. There are obviously a number of factors that are involved that allow such a wide range of speeds across many different 90nm chips.
And your comment about the trouble with frequency leaders: aren't the problems almost always exclusively with the frequency leaders?
When engineers don't communicate and document requirements properly, the project cost can easily spiral out of control and/or the end product doesn't meet customer expectations. And good communications skills are absolutely essential to this. As the size of the project goes up, the importance of team communications goes up exponentially.
And this communications skill includes inter-personal communications as well as written skills. I've seen too many examples of badly written requirements that are misinterpreted by the design engineer who goes down the wrong path (very efficiently, by the way, but totally wasted effort). The end result is a product that doesn't work the way the customer needed, so a redesign or a quick but inelegant fix is implemented. This kind of mistakes cost money, schedule, and customer satisfaction, and could have been easily avoided if the requirements are properly written without ambiguity.
Nonetheless this is an interesting discussion, so let's get back to it. The processors that have "done well" on 90 nm seem to be those that are not pushing the frequency limits. Intel's Dothan, AMD's first 90nm, IBM's 2 GHz 970FX, and the new UltraSparc all demonstrate considerably improved thermal performances. The trouble only seems to occur with the frequency leaders -- Prescott and the 2.5+ GHz 970FX. This actually supports Nr9's statements (allowing for his tendency to make them sound extreme): the frequency scaling benefits of at 90 nm seem to be non-existant on the 2.5+ GHz chips. This makes me seriously wonder about a couple of high profile high clock rate chips that are supposed to be showing up in 2005/2006...
Chip design / manufacturing isn't my area, but the way I see it is:
If there really is a Mhz wall, that means that the low end will shortly bump into the high end making single chips from each manufacturer low price commodity items - roughly comperable to other manufacturer's chips in performance and price. A single chip becomes a almost fixed unit of computational power that has a relatively low marketplace value.
Which means the focus turns almost exclusively to the memory system and each company's implementation. Which is what we are seeing with IBM,Sony, and Toshiba's Cell technology. And regardless of what one thinks of the specifics of Cell, a company like Apple will have to plan to move to something similar in concept if they don't want to get left behind.
The first Cell workstations are supposed to be released by the end of this year - I'm not sure if they are still on target for making that release time. But they should give us a idea of the type of system Apple could move to in the not to distant future.
Mark my words, there will not be a 5GHz processor for at least the next five years and there probably will never be.
I'm not an engineer so I will stay out of that conversation. However, using simple common sense: In 1999, I don't think anyone was able to accurately predict the specs of todays CPUs. During the past five years, there must have been some innovations that were expected and some that just came from nowhere. If Nr9 can accurately predict the evolution of the PowerPCs in the year 2009, he must have some magical powers or something. Use that power to buy stocks or predict the results of the Presidential election.
It is amazing that this thread still goes on and in that way. Not only because Nr9 stated nothing but the obvious trend in CPU industry, but also because of his rude and offensive manner when address the others here and his unusual level of negativism. A big NO from me.
Codenamed "Panther" the US IV+ will be available in mid 2005 and will _start_ at 1.8 GHz and move past 2 GHz later on. It seems to me, and obviously TI too, that 90 nm won't have to be the head ache that it's been for others. Pretty bold statements though considering what happened to the competition.
At the end of the day you really haven't said much here that hasn't been pretty well known for quite some time and you have consistently failed to go into any form of technical detail, Nr9. Even Intel acknowledged long ago power usage was getting too high. What nobody planned on was that leakage would jump so much at the 90 nm process. Every other company except Intel had multi-core plans for the 90 nm process and even Intel had them for the 65 nm process simply because they needed something to do with the transistors.
All that has really been realised lately is scaling is now hard but that has been pretty obvious given Intel's tearing up their old roadmaps.
Nothing you've said makes me think you have any inside knowledge at all and in fact much of what you've said is publicly available from journals or conference papers. In fact most is available from enthusiast sites and possible to make up claims for.
Quote:
Originally posted by Nr9
Right, do you even have an enginnering degree? You don't understand that unless you go to a liberal arts university, you dont need to write much do you?
Chemical in fact so process engineering and design is my business I'm guessing you however don't, or if you do you're most certainly very junior (I started serious reports that went beyond the mathematics of a problem in 2nd year so that'd make you 1st if that).
If ever you actually get into process design or any form of substantial project work, which most engineering is, you will spend your life in reports and presentations. It's how you convince managers that your ideas are not only technically feasible but worth carrying out. For instance in 3 weeks time I'll submit a feasibility study for a piece of equipment that'll equate to around 600 pages, excluding appendices (done by a team of 5). Everything from the economics, environmental impacts, OH&S, operating procedures, staffing, and detailed outlines of the equipment.
Around 60 - 70% of it is spent discussing the problem definition and key aspects of concern and assumptions and why the technical discussion and your ideas are actually worth the paper they're written on. That's engineering. If you aren't writing reports on your work then you aren't doing much of substance and nobody much cares about your work Engineers write reports, and management write more, just a fact of life. If all you can do is crunch numbers then you're next to useless as an engineer these days. Computers can solve mathematics these days so engineering has moved far more towards interpretation and proper methodologies of problem solving and analysis. All of which requires a good ability to communicate, particularly with the less technically minded. Just wait until you have to work with some community groups
I mean, if Intel gets to 3.* GHZ at 90nm and IBM gets to 2.5+ GHZ then I would have to say 2.5 GHZ isn't the 90nm barrier Nr9 is claiming.
No, there is no exact number to qualify as an absolute limit. What we're talking about is that crazy overclockers can make PPC970fx run at, say, 5GHz right now, but only in a can of liquid Nitrogen. Since IBM won't even consider shipping insanely overclocked chips, it roughly equals IBM having stuck at 2.5GHz. They may reach 3GHz or so for mass production of PPC970fx, but they are still having problems producing them even at 2.5GHz in quantities sufficient for Apple. You know what I mean: if those 3GHz chips turn out to be too hot, too few or too expensive, it will be just as though IBM haven't made a single chip running at 3GHz. Nobody gives a damn about a theoretical clock speed of a given chip - it only matters what clock speed the chip is running at on your desktop.
Quote:
There are obviously a number of factors that are involved that allow such a wide range of speeds across many different 90nm chips.
Yes, there are too many factors, only some of which relate to engineering (like materials, process, fab cleanness, design/process tweaking, etc.). There are also economical issues, like cost, profit, contracts you can't break, etc. And there are marketing issues, like pressure from sales guys, the urge to be the first, the need to be the best, etc.
History knows some examples when products were manufactured at a planned loss, or only to suppress competitors, or only to keep a customer happy. There are too many questions to ask, but what I want to know is what IBM engineers are doing to increase the performance/power ratio of 970fx right now. If it is not already obsolete in terms of further improvements.
If all you can do is crunch numbers then you're next to useless as an engineer these days. Computers can solve mathematics these days so engineering has moved far more towards interpretation and proper methodologies of problem solving and analysis. All of which requires a good ability to communicate, particularly with the less technically minded. Just wait until you have to work with some community groups
computers can't decide theories
im sorry you have to work in that kind of environment
i work in a research group and everyone around me knows what im talkign about
Nr9... MISTER IBM, the most misunderstood person around the world
Well, you may find it funny, I admit. But still: has Nr9 said anything so outrageously stupid that people like you mock him through 4 (FOUR!) pages of the thread?! You either believe or not, you either agree or argue, but why laugh like mad? Am I missing something?
Well, you may find it funny, I admit. But still: has Nr9 said anything so outrageously stupid that people like you mock him through 4 (FOUR!) pages of the thread?! You either believe or not, you either agree or argue, but why laugh like mad? Am I missing something?
What's about his Ars-Technica account... you get the idea
Comments
if you can't see that , you are dumb
why the hell would I talk about increasing voltage. of course im talking about an alternative
worrying about english gets in the way of doing the real technical stuff with the most efficiency possible. there is no free lunch. energy you spend constructing your sentences can be used for your technical engineering shit.
Originally posted by snoopy
Do I detect a slightly milder Nr9? . . .
Originally posted by Henriok
Sun just announced that the UltraSparc IV+ went 90 nm and gained a quite a lot frequenzy wise.
From 1.2 GHz to 1.8 GHz by doing the 130 nm to 90 nm dance we are all talking about.
So.. What is Texas Instrument's secret of doing a 50% increase in speed? Was their former fab really bad and their new really great, or is their 1.8 GHz claim just a lie or wishful thinking?
I think the secret here is that 1.2 GHz wasn't pushing the 130 nm envelope.
Originally posted by Nr9
worrying about english gets in the way of doing the real technical stuff with the most efficiency possible. there is no free lunch. energy you spend constructing your sentences can be used for your technical engineering shit.
I hate to break it to you Nr9, but you need to live and work in a world populated by humans. The sooner you learn that the energy you put into inter-personal communication is valuable, the sooner you'll find that you have a career path. If you can't see that then its your loss, not anyone else's.
Nonetheless this is an interesting discussion, so let's get back to it. The processors that have "done well" on 90 nm seem to be those that are not pushing the frequency limits. Intel's Dothan, AMD's first 90nm, IBM's 2 GHz 970FX, and the new UltraSparc all demonstrate considerably improved thermal performances. The trouble only seems to occur with the frequency leaders -- Prescott and the 2.5+ GHz 970FX. This actually supports Nr9's statements (allowing for his tendency to make them sound extreme): the frequency scaling benefits of at 90 nm seem to be non-existant on the 2.5+ GHz chips. This makes me seriously wonder about a couple of high profile high clock rate chips that are supposed to be showing up in 2005/2006...
What is Prescott's speed? I mean, if Intel gets to 3.* GHZ at 90nm and IBM gets to 2.5+ GHZ then I would have to say 2.5 GHZ isn't the 90nm barrier Nr9 is claiming. There are obviously a number of factors that are involved that allow such a wide range of speeds across many different 90nm chips.
And your comment about the trouble with frequency leaders: aren't the problems almost always exclusively with the frequency leaders?
And this communications skill includes inter-personal communications as well as written skills. I've seen too many examples of badly written requirements that are misinterpreted by the design engineer who goes down the wrong path (very efficiently, by the way, but totally wasted effort). The end result is a product that doesn't work the way the customer needed, so a redesign or a quick but inelegant fix is implemented. This kind of mistakes cost money, schedule, and customer satisfaction, and could have been easily avoided if the requirements are properly written without ambiguity.
Originally posted by Programmer
Nonetheless this is an interesting discussion, so let's get back to it. The processors that have "done well" on 90 nm seem to be those that are not pushing the frequency limits. Intel's Dothan, AMD's first 90nm, IBM's 2 GHz 970FX, and the new UltraSparc all demonstrate considerably improved thermal performances. The trouble only seems to occur with the frequency leaders -- Prescott and the 2.5+ GHz 970FX. This actually supports Nr9's statements (allowing for his tendency to make them sound extreme): the frequency scaling benefits of at 90 nm seem to be non-existant on the 2.5+ GHz chips. This makes me seriously wonder about a couple of high profile high clock rate chips that are supposed to be showing up in 2005/2006...
Chip design / manufacturing isn't my area, but the way I see it is:
If there really is a Mhz wall, that means that the low end will shortly bump into the high end making single chips from each manufacturer low price commodity items - roughly comperable to other manufacturer's chips in performance and price. A single chip becomes a almost fixed unit of computational power that has a relatively low marketplace value.
Which means the focus turns almost exclusively to the memory system and each company's implementation. Which is what we are seeing with IBM,Sony, and Toshiba's Cell technology. And regardless of what one thinks of the specifics of Cell, a company like Apple will have to plan to move to something similar in concept if they don't want to get left behind.
The first Cell workstations are supposed to be released by the end of this year - I'm not sure if they are still on target for making that release time. But they should give us a idea of the type of system Apple could move to in the not to distant future.
Originally posted by Nr9
Mark my words, there will not be a 5GHz processor for at least the next five years and there probably will never be.
I'm not an engineer so I will stay out of that conversation. However, using simple common sense: In 1999, I don't think anyone was able to accurately predict the specs of todays CPUs. During the past five years, there must have been some innovations that were expected and some that just came from nowhere. If Nr9 can accurately predict the evolution of the PowerPCs in the year 2009, he must have some magical powers or something. Use that power to buy stocks or predict the results of the Presidential election.
We should change this thread name to "Nr9: easy target".
Originally posted by Henriok
Sun just announced the UltraSparc IV+...
More Info,
Codenamed "Panther" the US IV+ will be available in mid 2005 and will _start_ at 1.8 GHz and move past 2 GHz later on. It seems to me, and obviously TI too, that 90 nm won't have to be the head ache that it's been for others. Pretty bold statements though considering what happened to the competition.
All that has really been realised lately is scaling is now hard but that has been pretty obvious given Intel's tearing up their old roadmaps.
Nothing you've said makes me think you have any inside knowledge at all and in fact much of what you've said is publicly available from journals or conference papers. In fact most is available from enthusiast sites and possible to make up claims for.
Originally posted by Nr9
Right, do you even have an enginnering degree? You don't understand that unless you go to a liberal arts university, you dont need to write much do you?
Chemical in fact so process engineering and design is my business
If ever you actually get into process design or any form of substantial project work, which most engineering is, you will spend your life in reports and presentations. It's how you convince managers that your ideas are not only technically feasible but worth carrying out. For instance in 3 weeks time I'll submit a feasibility study for a piece of equipment that'll equate to around 600 pages, excluding appendices (done by a team of 5). Everything from the economics, environmental impacts, OH&S, operating procedures, staffing, and detailed outlines of the equipment.
Around 60 - 70% of it is spent discussing the problem definition and key aspects of concern and assumptions and why the technical discussion and your ideas are actually worth the paper they're written on. That's engineering. If you aren't writing reports on your work then you aren't doing much of substance and nobody much cares about your work
Originally posted by bunge
I mean, if Intel gets to 3.* GHZ at 90nm and IBM gets to 2.5+ GHZ then I would have to say 2.5 GHZ isn't the 90nm barrier Nr9 is claiming.
No, there is no exact number to qualify as an absolute limit. What we're talking about is that crazy overclockers can make PPC970fx run at, say, 5GHz right now, but only in a can of liquid Nitrogen. Since IBM won't even consider shipping insanely overclocked chips, it roughly equals IBM having stuck at 2.5GHz. They may reach 3GHz or so for mass production of PPC970fx, but they are still having problems producing them even at 2.5GHz in quantities sufficient for Apple. You know what I mean: if those 3GHz chips turn out to be too hot, too few or too expensive, it will be just as though IBM haven't made a single chip running at 3GHz. Nobody gives a damn about a theoretical clock speed of a given chip - it only matters what clock speed the chip is running at on your desktop.
There are obviously a number of factors that are involved that allow such a wide range of speeds across many different 90nm chips.
Yes, there are too many factors, only some of which relate to engineering (like materials, process, fab cleanness, design/process tweaking, etc.). There are also economical issues, like cost, profit, contracts you can't break, etc. And there are marketing issues, like pressure from sales guys, the urge to be the first, the need to be the best, etc.
History knows some examples when products were manufactured at a planned loss, or only to suppress competitors, or only to keep a customer happy. There are too many questions to ask, but what I want to know is what IBM engineers are doing to increase the performance/power ratio of 970fx right now. If it is not already obsolete in terms of further improvements.
Originally posted by onlooker
I think Telomar just made Nr9 look like Sen. John Edwards did last night.
But not quite as bad as Dubya on Thursday.
Originally posted by Telomar
If all you can do is crunch numbers then you're next to useless as an engineer these days. Computers can solve mathematics these days so engineering has moved far more towards interpretation and proper methodologies of problem solving and analysis. All of which requires a good ability to communicate, particularly with the less technically minded. Just wait until you have to work with some community groups
computers can't decide theories
im sorry you have to work in that kind of environment
i work in a research group and everyone around me knows what im talkign about
Originally posted by Nr9
i work in a research group and everyone around me knows what im talkign about
Nr9... MISTER IBM, the most misunderstood person around the world
Originally posted by Fat Freddy
Nr9... MISTER IBM, the most misunderstood person around the world
Well, you may find it funny, I admit. But still: has Nr9 said anything so outrageously stupid that people like you mock him through 4 (FOUR!) pages of the thread?! You either believe or not, you either agree or argue, but why laugh like mad? Am I missing something?
Originally posted by costique
Well, you may find it funny, I admit. But still: has Nr9 said anything so outrageously stupid that people like you mock him through 4 (FOUR!) pages of the thread?! You either believe or not, you either agree or argue, but why laugh like mad? Am I missing something?
What's about his Ars-Technica account... you get the idea