Why so little parallelism?
|
From: Eugene Miya on 31 Oct 2006 14:42 >>>If I had a couple billion I would have better uses than building a fab. >> >> Then you have to be satisified with Moose turd pie. >> Or find someone elses facilities. In article <4qpjvfFo7qeaU1(a)individual.net>, Del Cecchi <cecchinospam(a)us.ibm.com> wrote: >Now Eugene, he doesn't need a few billion. He can design his own and >get help with the gory details, as well as access to a world class fab >for much less than a few billion. Actually I know this so long as you don't need performance. Besides we need to keep Applied Materials and Lam Research and all those other firms employed and advancing semiconductors. One computer designer I know is completely self taught, didn't finish high school and sold 256K units. Taught herself how to program FPGAs. >I would be happy to facilitate the transaction. A few million as >starters would be sufficient. Cheap! --
From: Del Cecchi on 31 Oct 2006 18:44 "Eugene Miya" <eugene(a)cse.ucsc.edu> wrote in message news:4547b552$1(a)darkstar... >>>>If I had a couple billion I would have better uses than building a >>>>fab. >>> >>> Then you have to be satisified with Moose turd pie. >>> Or find someone elses facilities. > > In article <4qpjvfFo7qeaU1(a)individual.net>, > Del Cecchi <cecchinospam(a)us.ibm.com> wrote: >>Now Eugene, he doesn't need a few billion. He can design his own and >>get help with the gory details, as well as access to a world class fab >>for much less than a few billion. > > Actually I know this so long as you don't need performance. Performance can be provided. As the hot rodders say, performance costs money. How fast do you want to go? > > Besides we need to keep Applied Materials and Lam Research and all > those > other firms employed and advancing semiconductors. Don't worry, they get theirs. > > One computer designer I know is completely self taught, didn't finish > high school and sold 256K units. Taught herself how to program FPGAs. > > >>I would be happy to facilitate the transaction. A few million as >>starters would be sufficient. > > Cheap! That was for starters. > > --
From: Nick Maclaren on 1 Nov 2006 03:57 In article <4qq5elFof9pgU1(a)individual.net>, "Del Cecchi" <delcecchiofthenorth(a)gmail.com> writes: |> |> >>Now Eugene, he doesn't need a few billion. He can design his own and |> >>get help with the gory details, as well as access to a world class fab |> >>for much less than a few billion. |> > |> > Actually I know this so long as you don't need performance. |> |> Performance can be provided. As the hot rodders say, performance costs |> money. How fast do you want to go? Yup. Fast, cheap, reliable. Pick any two. Regards, Nick Maclaren.
From: Eugene Miya on 1 Nov 2006 10:28 In article <4qq5elFof9pgU1(a)individual.net>, Del Cecchi <delcecchiofthenorth(a)gmail.com> wrote: > >"Eugene Miya" <eugene(a)cse.ucsc.edu> wrote in message >news:4547b552$1(a)darkstar... Del, you have to edit a tad better. Move the attribution to my text. >>>>>If I had a couple billion I would have better uses than building a >>>>>fab. >>>> >>>> Then you have to be satisified with Moose turd pie. >>>> Or find someone elses facilities. >> >> In article <4qpjvfFo7qeaU1(a)individual.net>, >> Del Cecchi <cecchinospam(a)us.ibm.com> wrote: >>>Now Eugene, he doesn't need a few billion. He can design his own and >>>get help with the gory details, as well as access to a world class fab >>>for much less than a few billion. >> >> Actually I know this so long as you don't need performance. > >Performance can be provided. As the hot rodders say, performance costs >money. How fast do you want to go? Well, you guys continue those peta-scale things with the DOE over at LLNL. Let them pay for it. And if something architecturally interesting comes of it, we can sign NDAs. But our budgets are tight due to the perceptions of computing and flight realities. No bucks, no Buck Rogers. >> Besides we need to keep Applied Materials and Lam Research and all >> those other firms employed and advancing semiconductors. > >Don't worry, they get theirs. That's the job. Policy makers are clueless to infrastructure needs. If the world had only seen the attempt by the former Gov. of Penn./Sec. DHS attempt to move businesses from here in a pitiful attempt to help his economy. >>>I would be happy to facilitate the transaction. A few million as >>>starters would be sufficient. >> >> Cheap! >That was for starters. Go for him. Too bad that Amdahl doesn't bother with news groups. --
From: jacko on 1 Nov 2006 13:10
hi back to the OPs question. why so little parallel? it's to do with the language used to program gets turned into an RTL (register transfer language) which specifies all variables as registers, and then this has to be mapped onto a machine which only has so many regs. this is why we use paper for calculations, so that the variable space can fit on it while the short term memory (register space) can only hold so much. making better RTL -> machine code translators which efficiently schedule transfers of variables between cores, to provide modular computation units. it is not a hardware problem per se, but a communications problem. maybe the RTL is not expressive enough. What would you have in an OpenRTL? OpenAPL?? cheers. |