AMD's 45nm technology compared against Intel's
in [Chips]
|
From: Del Cecchi` on 23 Dec 2008 22:47 Del Cecchi` wrote: snip followup to my own post. boooo > > Transmeta had no process technology. I am not sure which foundry they > used. A quick search says the first was TSMC at 130 nm, followed by a > switch to Fujitsu at 90. Turns out they started with IBM at 180 nm. Went to TSMC at 180 and 130 then Fujitsu at 90.
From: Robert Myers on 23 Dec 2008 23:49 On Dec 23, 10:43 pm, Del Cecchi` <dcecchinos...(a)att.net> wrote: > Robert Myers wrote: > > On Dec 23, 8:45 pm, Del Cecchi` <dcecchinos...(a)att.net> wrote: > > > Transmeta had no process technology. I am not sure which foundry they > used. A quick search says the first was TSMC at 130 nm, followed by a > switch to Fujitsu at 90. > Transmeta had power management patents. It was apparent that leakage (and hence, power consumption) was a big problem. > > > I spoke of knowledge as shown in csiphc. > > You generalized and did not qualify. > If there is anyone in this group who knows anything about solid state physics, no one has shown it [here], so what we've had [here] is an endless repetition of content-free polemics. This is *not* fun. > > I don't envy Intel, and never did. Please leave me out of this. > You're relatively new to this catfight [here]. > > > I think the learning curve applies to Semiconductors more than to > automobiles or grand pianos. > Ok. Everyone has an opinion. I think fluid mechanics the most sublime of the subdisciplines of physics. > > That's determined by design wins. AMD ain't doing so hot, and arguing > > here isn't going to change that. > > It isn't determined by design wins. It is determined by the behavior of > the part and stuff like die size. Design wins are influenced by many > things, and technical superiority is only one factor. > No. The "real question" is the real question, and the real question isn't who gets bragging rights by some measure proposed in a Usenet group, but by design wins. > less importance. IBM had a monopoly in punch cards and almost one in > the 360 era and we know how much help that was 20 years later. > That was my point. Why do we spend so much time worrying about Intel's "monopoly?" Robert.
From: Del Cecchi` on 24 Dec 2008 15:35 Robert Myers wrote: snip > > If there is anyone in this group who knows anything about > solid state physics, no one has shown it [here], so what we've had > [here] is an endless repetition of content-free polemics. > > This is *not* fun. I know some stuff about solid state physics. I didn't realize that was under discussion. > > >>I don't envy Intel, and never did. Please leave me out of this. >> > > You're relatively new to this catfight [here]. If this thread is supposed to be a catfight, then I won't be new, I'll be gone. > >> >>I think the learning curve applies to Semiconductors more than to >>automobiles or grand pianos. >> > > Ok. Everyone has an opinion. I think fluid mechanics the most > sublime of the subdisciplines of physics. so? Is that some kind of sarcastic rejoinder? > The applicability of the learning curve to semiconductor manufacturing is based on that manufacturing being a batch process in which only part of the resulting products are functional, along with the manufacturing facility being the dominant cost rather than materials and labor. > snip >> > > No. The "real question" is the real question, and the real question > isn't who gets bragging rights by some measure proposed in a Usenet > group, but by design wins. If you are talking finances or stock price or market share you are correct. I thought we were discussing technology. Sorry to intrude in the rock throwing. My bad. > > >>less importance. IBM had a monopoly in punch cards and almost one in >>the 360 era and we know how much help that was 20 years later. >> > > That was my point. Why do we spend so much time worrying about > Intel's "monopoly?" I don't worry about it at all. And I never have. Maybe some people do because they have AMD stock, or they object to some property of Intel or its processors. > > Robert. del
From: Robert Myers on 24 Dec 2008 19:55 On Dec 24, 3:35 pm, Del Cecchi` <dcecchinos...(a)att.net> wrote: > Robert Myers wrote: > > snip > > > > > If there is anyone in this group who knows anything about > > solid state physics, no one has shown it [here], so what we've had > > [here] is an endless repetition of content-free polemics. > > > This is *not* fun. > > I know some stuff about solid state physics. I didn't realize that was > under discussion. > Your sensitivity on this point surprises me. You never seemed the type to worry about how smart people thought you were or how much they thought you knew. > > >>I think the learning curve applies to Semiconductors more than to > >>automobiles or grand pianos. > > > Ok. Everyone has an opinion. I think fluid mechanics the most > > sublime of the subdisciplines of physics. > > so? Is that some kind of sarcastic rejoinder? > It's a comment on the very human tendency to insist on the uniqueness of ones own interests and enterprises. Computer science and engineering, whose days are largely now past, has had more than its share of self-aggrandizing tropes, which is what I take the uniqueness of the "learning curve" to be. Stand back about ten feet, and it's no different from autos, airplanes, tires, or grand pianos. I should have talked more about tires, which in some ways are *really* similar to microprocessors in terms of how much witchcraft is involved and how badly things can go wrong with *truly* catastrophic consequences. And the physics, even just the continuum mechanics, are complicated and subtle. The math is hard and not well understood. What's different about microprocessors is that they've had better press. > > > No. The "real question" is the real question, and the real question > > isn't who gets bragging rights by some measure proposed in a Usenet > > group, but by design wins. > > If you are talking finances or stock price or market share you are > correct. I thought we were discussing technology. Sorry to intrude in > the rock throwing. My bad. > What's the technology for and what's the measure of success? Commercial airplane manufacturers use cost per seat mile as a standin for likely success in the marketplace. Ultimately, it's who has the full order book and makes a profit that wins. Is there any engineer who does not live by some similar measure of success ultimately related to money? The constant annoyance here is that people pick some hobby horse or other (latency is the local favorite), claim that *it* is the most important measure of chip superiority, and hammer away at it, oblivious to the fact that Intel sells more processors than anyone else in the face of their logic. That, of course, leads to the screwball conclusion that Intel isn't really good at anything except marketing and market manipulation, another favorite local hobby horse. I'm actually embarrassed to be discussing this kind of stuff with you, Del, but it has been the meat and potatoes of the la la land you are posting in at the moment. Technological decisions almost invariably come down to money, especially if you allow yourself to translate risk to human life and health to dollars, which designers increasingly do. What other measure means anything? The nodding approval of the local AMD obsessives? Robert.
From: Del Cecchi` on 24 Dec 2008 22:54
Robert Myers wrote: > On Dec 24, 3:35 pm, Del Cecchi` <dcecchinos...(a)att.net> wrote: > >>Robert Myers wrote: >> >>snip >> >> >> >> >>>If there is anyone in this group who knows anything about >>>solid state physics, no one has shown it [here], so what we've had >>>[here] is an endless repetition of content-free polemics. >> >>>This is *not* fun. >> >>I know some stuff about solid state physics. I didn't realize that was >>under discussion. >> > > Your sensitivity on this point surprises me. You never seemed the > type to worry about how smart people thought you were or how much they > thought you knew. whatever. You are the one said no one has shown such knowledge. > > >>>>I think the learning curve applies to Semiconductors more than to >>>>automobiles or grand pianos. >> >>>Ok. Everyone has an opinion. I think fluid mechanics the most >>>sublime of the subdisciplines of physics. >> >>so? Is that some kind of sarcastic rejoinder? >> > > It's a comment on the very human tendency to insist on the uniqueness > of ones own interests and enterprises. Computer science and > engineering, whose days are largely now past, has had more than its > share of self-aggrandizing tropes, which is what I take the uniqueness > of the "learning curve" to be. Stand back about ten feet, and it's no > different from autos, airplanes, tires, or grand pianos. > > I should have talked more about tires, which in some ways are *really* > similar to microprocessors in terms of how much witchcraft is involved > and how badly things can go wrong with *truly* catastrophic > consequences. And the physics, even just the continuum mechanics, are > complicated and subtle. The math is hard and not well understood. > What's different about microprocessors is that they've had better > press. So that is why chips that start out costing many dollars are eventually sold for many times less even without shrinks etc. And with process improvements, chips that used to cost 50 dollars now cost 50 cents. If GM had pulled that off with cars we would all be better off. And I could have a baby grand for the price of the change on top of my dresser if Steinway had done likewise. Here is my attempt at an explanation. I am sure there are better ones out there. A 300mm wafer has about 70,000 mm2 of area. If a chip uses 100 mm2, there are possibly about 700 chips on the wafer. Processing the wafer costs a few thousand dollars. Call it 3500, including the depreciation on the fab, equipment, labor, materials. When manufacturing starts on that 100 mm chip, yield sucks and you only get a few good ones per wafer so they cost hundreds of dollars each to make. Once they have built them for a while they get good at it and yield is well above 50 percent and then each chip would only cost 10 dollars. And if the fab depreciation is paid off, the wafers are cheaper too. It is not microprocessors that are unique, but the batch fabrication of semiconductors. The notion of "learning curve" goes back to TI and discrete transistors at least. > > >>>No. The "real question" is the real question, and the real question >>>isn't who gets bragging rights by some measure proposed in a Usenet >>>group, but by design wins. >> >>If you are talking finances or stock price or market share you are >>correct. I thought we were discussing technology. Sorry to intrude in >>the rock throwing. My bad. >> > > What's the technology for and what's the measure of success? > > Commercial airplane manufacturers use cost per seat mile as a standin > for likely success in the marketplace. Ultimately, it's who has the > full order book and makes a profit that wins. > > Is there any engineer who does not live by some similar measure of > success ultimately related to money? > > The constant annoyance here is that people pick some hobby horse or > other (latency is the local favorite), claim that *it* is the most > important measure of chip superiority, and hammer away at it, > oblivious to the fact that Intel sells more processors than anyone > else in the face of their logic. That, of course, leads to the > screwball conclusion that Intel isn't really good at anything except > marketing and market manipulation, another favorite local hobby horse. You must admit Intel has done some market manipulation on occasion. > > I'm actually embarrassed to be discussing this kind of stuff with you, > Del, but it has been the meat and potatoes of the la la land you are > posting in at the moment. > > Technological decisions almost invariably come down to money, > especially if you allow yourself to translate risk to human life and > health to dollars, which designers increasingly do. What other > measure means anything? The nodding approval of the local AMD > obsessives? > > Robert. Of course money drives decisions on technology. I never said it didn't. What I did say is that discussions of the relative merit of designs can be independent of market success. GM sold a lot of Vegas and Ford sold a lot of Pintos. Should we declare the Vega or the Pinto superior to the Mercedes (or a Volvo if you think that is a better example) because Ford and GM had a much greater market share? Or the classic VHS/Beta battle? You might also note that Government makes the tradeoff of human life vrs money much more than the microprocessor vendors do. We got told that an intersection near our house wouldn't get a traffic light until three people were killed. And the government is only now starting to put up barriers to keep cars from crossing narrow medians on four lane highways in very selected areas. Although I don't quite know how that got in there, since decisions on memory interfaces or other aspects of processor design don't seem to involve risk to human life that I can see. Both companies make fine processors, any of which will probably work fine for anything I want to do. We now return you to the regularly scheduled cage fighting match. del |