Why so little parallelism?
|
From: Stephen Fuld on 21 Jan 2007 10:53 Peter "Firefly" Lund wrote: snip > (Actually, I can recommend *anything* by Simon Peyton Jones, no matter > the subject. Take a look at his article on Excel, functions, and > usability.) I read this paper and found it quite interesting. I have two questions. Is the functionality described in the paper included in current releases of Excel? The paper said they might be, but I haven't found it touted as a new feature. The second is to think about using the "spreadsheet with functions" metaphor to help make parallel programming easier. Since many people find the spreadsheet metaphor easy to deal with, and it can "automatically" break up the calculations into multiple "threads", this might help people make better use of multi-core processors, etc. It does require some rethinking for most programmers. For example, loops become iterations in space (different cells) not time. But this might be a smaller change than that required to specifically specify locking or message passing. -- - Stephen Fuld (e-mail address disguised to prevent spam)
From: Stephen Fuld on 22 Jan 2007 12:33 Stephen Fuld wrote: > Peter "Firefly" Lund wrote: > > snip > >> (Actually, I can recommend *anything* by Simon Peyton Jones, no matter >> the subject. Take a look at his article on Excel, functions, and >> usability.) > > > I read this paper and found it quite interesting. Sorry for tacking on to my own post, but I thought I should give a link to the paper. One of several is http://research.microsoft.com/~simonpj/papers/excel/excel.pdf -- - Stephen Fuld (e-mail address disguised to prevent spam)
From: Anne & Lynn Wheeler on 31 Jan 2007 11:45 IBM's Chief Architect Says Software is at Dead End http://it.slashdot.org/it/07/01/30/1547235.shtml and Where's The Software To Catch Up To Multicore Computing? http://www.informationweek.com/news/showArticle.jhtml?articleID=197001130 from above: To make this complex architecture useful to even the most advanced scientific simulation application developers, much of the work on the system development is in the programming methodology enablement and corresponding application framework and tooling. .... snip ...
From: Nick Maclaren on 1 Feb 2007 04:05 In article <m3d54v16y5.fsf(a)garlic.com>, Anne & Lynn Wheeler <lynn(a)garlic.com> writes: |> |> IBM's Chief Architect Says Software is at Dead End |> http://it.slashdot.org/it/07/01/30/1547235.shtml |> Where's The Software To Catch Up To Multicore Computing? |> http://www.informationweek.com/news/showArticle.jhtml?articleID=197001130 I am scarcely likely to disagree! |> from above: |> |> To make this complex architecture useful to even the most advanced |> scientific simulation application developers, much of the work on the |> system development is in the programming methodology enablement and |> corresponding application framework and tooling. Yeah. For 30 years now. Progress in many areas, adequate to excellent. Progress in enabling the automatic use of parallelism by 'ordinary' codes written by non-parallel-skilled programmers, essentially nil. Dammit, you and I know that we could make use of almost arbitrary parallelism for matrix codes before 1975 - possibly well before. And the same was true for trivially decomposable problems. In fact, many people did just those things. But where are we now, as far as automatic parallelisation goes? Regards, Nick Maclaren.
From: Joe Seigh on 1 Feb 2007 06:33
Nick Maclaren wrote: > In article <m3d54v16y5.fsf(a)garlic.com>, > Anne & Lynn Wheeler <lynn(a)garlic.com> writes: > |> > |> IBM's Chief Architect Says Software is at Dead End > |> http://it.slashdot.org/it/07/01/30/1547235.shtml > |> Where's The Software To Catch Up To Multicore Computing? > |> http://www.informationweek.com/news/showArticle.jhtml?articleID=197001130 > > I am scarcely likely to disagree! > > |> from above: > |> > |> To make this complex architecture useful to even the most advanced > |> scientific simulation application developers, much of the work on the > |> system development is in the programming methodology enablement and > |> corresponding application framework and tooling. > > Yeah. For 30 years now. Progress in many areas, adequate to excellent. > Progress in enabling the automatic use of parallelism by 'ordinary' > codes written by non-parallel-skilled programmers, essentially nil. > > Dammit, you and I know that we could make use of almost arbitrary > parallelism for matrix codes before 1975 - possibly well before. And > the same was true for trivially decomposable problems. In fact, many > people did just those things. But where are we now, as far as > automatic parallelisation goes? > > Apparently not far enough along for hardware manufacturers who now expect 30 years worth of software to be instantly rewritten to exploit multi-cored cpus because their business model suddenly depends on it. They should know that making up a new lemonade recipe takes time. -- Joe Seigh When you get lemons, you make lemonade. When you get hardware, you make software. |