Why so little parallelism?
|
From: Bill Todd on 16 Dec 2006 21:09 BDH wrote: >> |> > |> That's an unsolved problem? >> |> > >> |> > Yes. >> |> >> |> Well, you shouldn't have told him. The best way of getting an unsolved >> |> problem solved is to present it as homework. (Cf. George Dantzig.) >> >> Well, yes, but given his attitude, I don't think that telling him will >> make any difference! I will be impressed if he can pull it off, but >> suspect that he will merely reinvent one of the solutions for the easy >> cases. > > Yeah, whatever. On my list of choices for things to do when I have time > it goes. Yes, the actual solution takes little time, but even writing > it down can take a while. I've heard of 'Napoleon complexes', but I believe this is the first time I've ever encountered a 'Fermat complex'. - bill
From: Anne & Lynn Wheeler on 18 Dec 2006 13:32 eugene(a)cse.ucsc.edu (Eugene Miya) writes: > I know it's sitting at the Museum. email not long before we got told that the project was being transferred and we weren't suppose to work on anything with more than four processors. it turned out that there was major product announcement later, but it was by kingston, not us, and we never did do any scale-up announcements. if you have anything in the computer museum, it didn't from us. previous cluster-in-a-rack/medusa refs: http://www.garlic.com/~lynn/2006w.html#13 IBM sues maker of Intel-based Mainframe clones http://www.garlic.com/~lynn/2006w.html#14 IBM sues maker of Intel-based Mainframe clones http://www.garlic.com/~lynn/2006w.html#20 cluster-in-a-rack http://www.garlic.com/~lynn/2006w.html#26 Why so little parallelism? http://www.garlic.com/~lynn/2006w.html#38 Why so little parallelism? http://www.garlic.com/~lynn/2006w.html#39 Why so little parallelism? http://www.garlic.com/~lynn/2006w.html#40 Why so little parallelism? http://www.garlic.com/~lynn/2006w.html#41 Why so little parallelism? as several previous references, here is old references to the meeting at oracle http://www.garlic.com/~lynn/95.html#13 http://www.garlic.com/~lynn/96.html#15 From: wheeler Date: 29 January 1992, 16:28:48 PST Subject: "cluster computing" I believe we got a charter last week to do generalized "cluster computing" with both horizontal growth and availability. We are going full steam ahead now with plans for major product announce directions in 6-8 weeks. Hester gave Larry Ellison (pres. of Oracle) some general "technology directions". I'm now in the middle with nailing down overall ... total computing system environment ... for that year end time frame (database, horizontal growth, availability, connectivity, function, fileserving, applications, system management, enterprise wide services, etc, etc). I wasn't able to make the LLNL meeting tues & weds. this week ... but XXXXX and YYYYY came by this afternoon (after the meeting). YYYYY had already put together pictures of the visionary direction (i.e. for LLNL national storage center) titled "DATAstore" with NSC providing a generalized fabric switch/router with lots of things connected to it ... both directly & fully-meshed and in price/performance hierarchy ... that had HA/6000 as the controlling central "brains". He effectively said I can get a generalized NSC switch/router built off combining current NSC/DX technology (including the RISC/6000 SLA interface) and their HiPPI switch by 2nd qtr. By ye he should have for me a generalized switch fabric (called UNIswitch) that has variety of "port" boards * Sonet, * FDDI, * ESCON, * FCS, * serial HiPPI, * parallel HiPPI, * NSC/DX In theory, anything coming in any port ... can come out any other port. Also, YYYYY has built into the "switch fabric" a "security" cross-matrix function that can limit who can talk to who (i.e. otherwise the default fabric is fully-meshed environment, everybody can talk to everybody). I can use this for the HA "I/O fencing" function ... which is absolutely necessary for going greater than two-way. XXXXX brought up the fact that we have a larger "scope" here and that immediately there are about a dozen large "hot Unitree" activities going on at the moment and that (at least) we three will have to coordinate. One of them is the current LLNL physical data repository technical testbed ... but there are two other production environments at LLNL that have to be addressed in parallel with this work ... and there are another 9 or so locations that we also have to address. In addition, both NSC and DISCOS have been having some fairly close dealings with Cornell ... both Cornell proper and also with regard to the bid on the NSF stuff. Also the Grummen SI stuff for Nasa/Huntsville came up. ZZZZZ was also in town visiting Almaden about some multi-media stuff .... and I invited him to sit in on the meeting with YYYYY and XXXXX. That gave us the opportunity to discuss a whole other series of opportunities (like at Cargil(sp?)). The tie-in at Discos is interesting since General Atomics also operates the UCSD supercomputing center ... and at least two of the papers at last fall SOSP on multi-media filesystem requirements were from UCSD (XXXXX knows the people doing the work). Also in the discussions with XXXXX about Unitree development we covered various things that WWWWW (LLNL) had brought up in the past couple days (off line) and the Cummings Group stuff (NQS-exec, network caching, log-structured filesystem, etc). XXXXX wants to have two 3-way meetings now ... one between WWWWW, XXXXX and me ... in addition to the 3-way (or possibly 4-way) meeting between Cummings, XXXXX, and me. This is all the visionary stuff that we sort of ran thru for the total computing environment that we would like to have put together for next year (hardware, software, distributed, networking, system management, commercial, technical, filesystems, information management). Effectively YYYYY, XXXXX, and I came out of the meeting with ground-work platform for both hardware & software to take over the whole worlds' computing environment. Little grandiose, but we will be chipping away at it in nice manageable business justified "chunks/deliverables". This is consistent with an overall theme and a series of whitepapers that we have an outside consultant working on (was one of the founders of Infoworld and excellent "tech writer") ... talking about the computing vision associated with "cluster computing" (which includes the MEDUSA stuff ... and HA/MEDUSA being base for HA/6000 scaleup). .... snip ... and of course, we were producing a product ... misc. past posts mentioning http://www.garlic.com/~lynn/subtopic.html#hacmp the reference to enterprise wide services was part of our 3-tier architecture ... misc. recent postings http://www.garlic.com/~lynn/2006u.html#55 What's a mainframe? http://www.garlic.com/~lynn/2006v.html#10 What's a mainframe? http://www.garlic.com/~lynn/2006v.html#14 In Search of Stupidity http://www.garlic.com/~lynn/2006v.html#35 What's a mainframe? and past collected postings mentioning 3-tier http://www.garlic.com/~lynn/subnetwork.html#3tier for other relational drift and scale-up http://www.garlic.com/~lynn/2004d.html#6 Memory Affinity and couple other old rdbms/oracle references: http://www.garlic.com/~lynn/2004o.html#40 Facilities "owned" by MVS http://www.garlic.com/~lynn/2000e.html#49 How did Oracle get started? and, of course, misc. and sundry posts about system/r http://www.garlic.com/~lynn/subtopic.html#systemr part of ha/cmp scaleup was work on distributed lock manager ... misc past posts: http://www.garlic.com/~lynn/2000.html#64 distributed locking patents http://www.garlic.com/~lynn/2000g.html#32 Multitasking and resource sharing http://www.garlic.com/~lynn/2001.html#40 Disk drive behavior http://www.garlic.com/~lynn/2001c.html#66 KI-10 vs. IBM at Rutgers http://www.garlic.com/~lynn/2001e.html#2 Block oriented I/O over IP http://www.garlic.com/~lynn/2001e.html#4 Block oriented I/O over IP http://www.garlic.com/~lynn/2001f.html#22 Early AIX including AIX/370 http://www.garlic.com/~lynn/2001i.html#21 3745 and SNI http://www.garlic.com/~lynn/2001i.html#30 IBM OS Timeline? http://www.garlic.com/~lynn/2001j.html#17 I hate Compaq http://www.garlic.com/~lynn/2001j.html#47 OT - Internet Explorer V6.0 http://www.garlic.com/~lynn/2001k.html#5 OT - Internet Explorer V6.0 http://www.garlic.com/~lynn/2001k.html#18 HP-UX will not be ported to Alpha (no surprise)exit http://www.garlic.com/~lynn/2001l.html#5 mainframe question http://www.garlic.com/~lynn/2001l.html#8 mainframe question http://www.garlic.com/~lynn/2001l.html#17 mainframe question http://www.garlic.com/~lynn/2001n.html#23 Alpha vs. Itanic: facts vs. FUD http://www.garlic.com/~lynn/2002b.html#36 windows XP and HAL: The CP/M way still works in 2002 http://www.garlic.com/~lynn/2002b.html#37 Poor Man's clustering idea http://www.garlic.com/~lynn/2002d.html#31 2 questions: diag 68 and calling convention http://www.garlic.com/~lynn/2002e.html#67 Blade architectures http://www.garlic.com/~lynn/2002e.html#71 Blade architectures http://www.garlic.com/~lynn/2002f.html#1 Blade architectures http://www.garlic.com/~lynn/2002f.html#4 Blade architectures http://www.garlic.com/~lynn/2002f.html#5 Blade architectures http://www.garlic.com/~lynn/2002f.html#6 Blade architectures http://www.garlic.com/~lynn/2002f.html#17 Blade architectures http://www.garlic.com/~lynn/2002k.html#8 Avoiding JCL Space Abends http://www.garlic.com/~lynn/2002m.html#21 Original K & R C Compilers http://www.garlic.com/~lynn/2002n.html#27 why does wait state exist? http://www.garlic.com/~lynn/2002o.html#14 Home mainframes http://www.garlic.com/~lynn/2003c.html#53 HASP assembly: What the heck is an MVT ABEND 422? http://www.garlic.com/~lynn/2003d.html#2 IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM http://www.garlic.com/~lynn/2003d.html#8 IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM http://www.garlic.com/~lynn/2003d.html#54 Filesystems http://www.garlic.com/~lynn/2003h.html#35 UNIX on LINUX on VM/ESA or z/VM http://www.garlic.com/~lynn/2003i.html#70 A few Z990 Gee-Wiz stats http://www.garlic.com/~lynn/2003k.html#10 What is timesharing, anyway? http://www.garlic.com/~lynn/2003k.html#17 Dealing with complexity http://www.garlic.com/~lynn/2004c.html#53 defination of terms: "Application Server" vs. "Transaction Server" http://www.garlic.com/~lynn/2004d.html#72 ibm mainframe or unix http://www.garlic.com/~lynn/2004i.html#1 Hard disk architecture: are outer cylinders still faster than inner cylinders? http://www.garlic.com/~lynn/2004i.html#2 New Method for Authenticated Public Key Exchange without Digital Certificates http://www.garlic.com/~lynn/2004i.html#8 Hard disk architecture: are outer cylinders still faster than inner cylinders? http://www.garlic.com/~lynn/2004m.html#0 Specifying all biz rules in relational data http://www.garlic.com/~lynn/2004m.html#5 Tera http://www.garlic.com/~lynn/2004q.html#10 [Lit.] Buffer overruns http://www.garlic.com/~lynn/2004q.html#37 A Glimpse into PC Development Philosophy http://www.garlic.com/~lynn/2004q.html#70 CAS and LL/SC http://www.garlic.com/~lynn/2004q.html#71 will there every be another commerically signficant new ISA? http://www.garlic.com/~lynn/2005.html#40 clusters vs shared-memory (was: Re: CAS and LL/SC (was Re: High Level Assembler for MVS & VM & VSE)) http://www.garlic.com/~lynn/2005.html#55 Foreign key in Oracle Sql http://www.garlic.com/~lynn/2005b.html#1 Foreign key in Oracle Sql http://www.garlic.com/~lynn/2005f.html#18 Is Supercomputing Possible? http://www.garlic.com/~lynn/2005f.html#32 the relational model of data objects *and* program objects http://www.garlic.com/~lynn/2005h.html#26 Crash detection by OS http://www.garlic.com/~lynn/2005h.html#28 Crash detection by OS http://www.garlic.com/~lynn/2005i.html#42 Development as Configuration http://www.garlic.com/~lynn/2005m.html#8 IBM's mini computers--lack thereof http://www.garlic.com/~lynn/2005q.html#49 What ever happened to Tandem and NonStop OS ? http://www.garlic.com/~lynn/2005r.html#23 OS's with loadable filesystem support? http://www.garlic.com/~lynn/2005u.html#38 Mainframe Applications and Records Keeping? http://www.garlic.com/~lynn/2006b.html#8 Free to good home: IBM RT UNIX http://www.garlic.com/~lynn/2006c.html#8 IBM 610 workstation computer http://www.garlic.com/~lynn/2006c.html#41 IBM 610 workstation computer http://www.garlic.com/~lynn/2006d.html#14 IBM 610 workstation computer http://www.garlic.com/~lynn/2006j.html#20 virtual memory http://www.garlic.com/~lynn/2006o.html#24 computational model of transactions http://www.garlic.com/~lynn/2006o.html#32 When Does Folklore Begin??? http://www.garlic.com/~lynn/2006o.html#33 When Does Folklore Begin??? http://www.garlic.com/~lynn/2006o.html#62 Greatest Software, System R
From: Nicholas King on 19 Dec 2006 06:11 BDH wrote: >> |> > |> That's an unsolved problem? >> |> > >> |> > Yes. >> |> >> |> Well, you shouldn't have told him. The best way of getting an unsolved >> |> problem solved is to present it as homework. (Cf. George Dantzig.) >> >> Well, yes, but given his attitude, I don't think that telling him will >> make any difference! I will be impressed if he can pull it off, but >> suspect that he will merely reinvent one of the solutions for the easy >> cases. > > Yeah, whatever. On my list of choices for things to do when I have time > it goes. Yes, the actual solution takes little time, but even writing > it down can take a while. > Writing down the solution is part of solving it. If you can't solve it and write it down with 20 minutes then you can simply claim arbitrary extra time to solve it whilst you are writing it down. However personally i think your claim is a load of bullshit since i could easily just give you any problem from the class of P-Complete languages which are thought to be inherently sequential. http://en.wikipedia.org/wiki/P-complete
From: ChrisQuayle on 19 Dec 2006 10:44 Eugene Miya wrote: > Whose books? Foley and Van Dam, etc.? > Newman and Sproull? I have Foley & Van Dam, Hearn & Baker, Burger & Gillies and others, but the problem with all these books is that they assume a machine with lots of resources. Being a complete newbie to graphics programming and having bought the books on graphics, I was soon out buying s/h books on math for revision - stuff I hadn't looked at for years. If you are working with low throughput processors, this leads on to books on approximations and techniques thereof. Books from the 60's/70's/80's, when graphics was really hard work in performance terms. This leads one to stuff like cordic etc, but none of this is as fast as scaled integer arithmetic ^ 2 and lookup tables for the transcendental functions. Oh yes, and ABE books is your friend :-)... > > Are these pixels aliased or anti-aliased? > For a lot of embedded work, you don't have the cpu throughput to support anti aliasing, so it's one bit per pixel. You get round the problem of aliasing by designing the screen layout to avoid visible aliasing, or choose shapes that don't cause the problem in the first place. > I have friends who are looking to hire graphics talent, but they are > finding that the talent pool is drying up because the perception is that > graphics is a "solved" problem. They are not even finding it in India > and China. The normal computing channels have guys who think they don't > know graphics and art, and they more than enough graphics arts people, > but fewer and fewer in the technical coding aspects. > Well, perhaps the problem of graphics is solved in terms of low cost hardware accelerated boards and libraries to drive them, available at low cost to all, but one would hope that this doesn't mean the end of basic research... Chris
From: BDH on 20 Dec 2006 09:20
> Writing down the solution is part of solving it. If you can't solve it > and write it down with 20 minutes then you can simply claim arbitrary > extra time to solve it whilst you are writing it down. You wouldn't say somebody needs to write an algorithm in assembly for it to be done. You shouldn't say somebody needs to write it in any programming language for it to be done. Why should you say somebody should have to write it in English? > However personally i think your claim is a load of bullshit since i > could easily just give you any problem from the class of P-Complete > languages which are thought to be inherently sequential. > > http://en.wikipedia.org/wiki/P-complete Well that is an interesting link. I didn't know there was a parallelization counterpart to NP-complete! No, I can't see a way to parallelize those P-complete problems, at least with non-quantum computers that can physically exist. But I thought "I can't do something if it's impossible" was implied. |