The future of CPU based computing, mini clusters.
|
Prev: Maximum ammount of local variables in cg shaders ?
Next: Online Exams for Certification, Free Practice Exams, Study Material, Dumps
From: Terje Mathisen on 2 Nov 2009 02:18 Bernd Paysan wrote: > Terje Mathisen wrote: > Hm, digging out the publication leads me here: > > http://www.physics.berkeley.edu/research/zettl/pdf/361.NanoLet.9- > Begtrup.pdf > >> I.e. this might well work at some point in time, but not in "2-4 >> years", more like 10-20. > > I don't think one can plan the availability of unobtanium. The current :-) > state of research is to find a material that is suitable as memristor. > You can find one tomorrow, or not in 1000 years. Zettl's nanotubes with > embedded iron shuttles look much better than HPs titanium dioxide > memristors. They are also probably a lot harder to make ;-). The key point is that even if they do find the proper (easy to manufacture, stable, cheap etc) material today, it will still take a lot more than "2-4 years" to make it commonly available in the form of embedded memory. Terje -- - <Terje.Mathisen at tmsw.no> "almost all programming can be viewed as an exercise in caching"
From: Bernd Paysan on 2 Nov 2009 05:41 Terje Mathisen wrote: > The key point is that even if they do find the proper (easy to > manufacture, stable, cheap etc) material today, it will still take a lot > more than "2-4 years" to make it commonly available in the form of > embedded memory. Well, I'm a bit more optimistic about that. "Easy to manufacture" means "fits into the current process flow", so it's essentially just another process step - memristor inserted into a via - in a standard process. If you ask the process guys "please grow multi-walled carbon nanotubes of 200nm length with embedded iron shuttles in every via hole", then you'll get your multi-year estimate with no guarantee to success, but if it's really just something you can create with a standard CVD process step (like the titanium oxide HP wants to use), it's not rocket science, and can be ported to other fabs as well. -- Bernd Paysan "If you want it done right, you have to do it yourself" http://www.jwdt.com/~paysan/
From: ChrisQ on 2 Nov 2009 06:55 nmm1(a)cam.ac.uk wrote: > In article <vBZGm.2971$6O1.1139(a)newsfe08.ams2>, > ChrisQ <meru(a)devnull.com> wrote: >>> Yes, but I wasn't cheating in that way - the machine I used did NOT >>> memory map registers - it had 128 registers, and used a special >>> register to allow register indexing. >> 128 registers provides a bit more space then the pdp11's 8, but how did >> your code work ?. The registers must have been sequential in some >> accessable address space for the program counter to access them, or you >> were doing something really clever like indirect execution through the >> index ?. Not familiar with that architecture makes it more difficult guess. >> >> Just how did it work ? :-)... > > After 40+ years, I should have to check up on which algorithm it was > and reinvent the method. Basically, it was something that needed > a lot of combinatoric calculations on a small number of counts. > > And, yes, you could index through the registers. The loading and > unloading was manual, of course. > Right, i've forgotten half the stuff I ever knew about the pdp11 and vax and that's only 20 years ago. Still, there does seem to be *real* progress in computer architecture, despite the retentiveness of the major players. It seems that Tilera processors will be on server motherboards within a year. A 100 cpu 'computing surface', if you like. No fpu, but i'm sure ways will be found to utilise them for hpc none the less. Link is at: http://www.theregister.co.uk/2009/11/02/tilera_quanta_servers/ How could you use that ?. It will be commodity prices as well... Regards, Chris
From: nmm1 on 2 Nov 2009 07:48 In article <1588054.EvxWjJku6o(a)elfi.zetex.de>, Bernd Paysan <bernd.paysan(a)gmx.de> wrote: >Terje Mathisen wrote: >> The key point is that even if they do find the proper (easy to >> manufacture, stable, cheap etc) material today, it will still take a lot >> more than "2-4 years" to make it commonly available in the form of >> embedded memory. > >Well, I'm a bit more optimistic about that. "Easy to manufacture" means >"fits into the current process flow", so it's essentially just another >process step - memristor inserted into a via - in a standard process. Well, yes, if you don't mind it not working or fouling up everything else. There's more to engineering than just getting parts to fit together. >If >you ask the process guys "please grow multi-walled carbon nanotubes of 200nm >length with embedded iron shuttles in every via hole", then you'll get your >multi-year estimate with no guarantee to success, but if it's really just >something you can create with a standard CVD process step (like the titanium >oxide HP wants to use), it's not rocket science, and can be ported to other >fabs as well. Rocket science is almost trivial; rocket engineering is hard. Quite a lot of things are like that. Regards, Nick Maclaren.
From: Robert Myers on 2 Nov 2009 09:00
On Nov 2, 7:48 am, n...(a)cam.ac.uk wrote: > Rocket science is almost trivial; rocket engineering is hard. Quite > a lot of things are like that. If X engineering is hard, it's almost because X science is missing or incomplete. That is, in my not so humble opinion, particularly true of rocket science, which is anything but trivial. Robert. |