Notes on two presentations by Gordon Bell ca. 1998
|
From: nmm1 on 2 Feb 2010 17:28 In article <E8Wdne4baN0-AvXWnZ2dnUVZ_g2dnZ2d(a)supernews.com>, Gavin Scott <gavin(a)allegro.com> wrote: > >I'm curious how many of the "C problems" go away with functional >styles of problem description, or whether the hard ones just mutate >into a different form. My experience (and that of others) is that there are two issues that are often confused: A good half of "C problems" go away when moving to a saner imperative language, especially one with checking. A lot go away even moving to Fortran. Of the remaining ones, perhaps half are eliminated by moving to a functional language, and the others just mutate. For example, "dangling pointer" problems often end up as accessing the wrong version of an object or memory leaks. Regards, Nick Maclaren.
From: Eugene Miya on 2 Feb 2010 17:43 In article <E8Wdne4baN0-AvXWnZ2dnUVZ_g2dnZ2d(a)supernews.com>, Gavin Scott <gavin(a)allegro.com> wrote: >> Museum heading > >I haven't been over to the museum in ages, but I was really sad to >see visible storage close. It's just going to its next stage. Scale has its problems. All kinds of discussion is going on in various channels. >That was perhaps the best museum experience >ever, having all those famous machines right *there*. All of the >designs for the $50,000,000 version that I saw looked like a big >walk-through coffee-table book. Lots of pretty colors to engage the >visual cortex while your feet get sore, and lots of text that people >will gloss over. Well, I am not certain that I would call it the best museum. I like the Exploratorium (a climbing partner is chief scientist), and I visited the Deutsches Museum 4 years ago, just came from the lesser TM Wien, I have a "hat" representing the Air and Space Museum on the West Coast, Duxford near Nick is very impressive. Every museum involves some amount of compromise. The Smithsonian's hands are tied, their charter places them in a stange place. CHM is now is more oriented toward the general public and especially that part of the public computer industry which promises funding toward endowment and simple stories which the public thinks it can relate to (once called the Kiddy museum in Boston). For instance, while I was not a fan of VLIW, I felt the Museum's purpose was to real hardware in contrast to gum flapping, and if it ran, so much the better. That's not likely ever to happen, but the hardware, and the story are somewhat there. It's very tempting and easy to stay with the big boys. But interesting stuff in the collection is not on display. On one hand, I would want the visitor to be irked at the petty aspects of computers. I sometimes wonder if the future is going to survive its past. >It's going to be hard to beat the warehouse + enthusiastic docent >model I think. Many museums have that. Baby boomers docent all over the place. >F# >class language with support on par with C# and Visual Basic. This >may well make it the most widely used functional programming >language before very long. Really? >I found it an interesting language, and can definitely see a lot of >advantages in trying to extract parallelism out of something like >this rather than with the usual C-derived stuff. Do you think MS has the right compiler wienners to generate code? >I'm curious how many of the "C problems" go away with functional >styles of problem description, or whether the hard ones just mutate >into a different form. My impression for most languages these days is toward mutation. That's not necessarily bad, just different. I'd compare it to LISP attempts, APL, and some of the other functional languages if I had time. -- Looking for an H-912 (container).
From: Robert Myers on 2 Feb 2010 17:46 On Feb 2, 5:20 pm, eug...(a)cse.ucsc.edu (Eugene Miya) wrote: > > Sid Fernbach, for whom one of the IEEE awards is given, started a briefly > class classification of supercomputers, and outsiders wondered what they > VAX, the mainframe, their IBM PC rated on his scale where a Cray 1 was a > "Class 6" machine. Since I had Sid handy, I had a chance to ask him: > Oh, they are "class 1/2". They barely even rate. They aren't supercomputers. Such a comment would require a defense. As it stands, I disagree with it. There are vast holes in my knowledge and experience of computers, but I went practically straight from a Cray to a PC (admittedly aided by an expansion bus card running a NS320xx). My current PC is hands-down more powerful than the Cray ever was to me as a user. If there are differences, they are differences that have obsessed computer professionals, like the ability to justify their budgets by serving the needs of dozens or hundreds of computer users simultaneously. The Cray had mainframe-like characteristics (like peripheral processors) that PC's don't currently have, but those differences don't matter to me as a user, and the larger range of software, more immediate access to data and superior graphics, along with enough power to do interesting things, make the PC seem like a big step up to me. The one thing wrong that I see is that the step up to whatever is difficult or even nearly impossible. When you get to the point where more horsepower is required, even the step to a personal beowulf cluster is painful--and even that might not have been so had Intel not made the decision it made about infiniband. As it stands right now, the old advice--wait a few years--is a long way from running out of steam. The publicity hounds at the national labs will run you over with their spectacular graphics? That's a given. Robert.
From: Eugene Miya on 2 Feb 2010 18:12 In article <57f7bab3-3dda-4fef-8e1a-6c1f72f46191(a)h2g2000yqj.googlegroups.com>, Robert Myers <rbmyersusa(a)gmail.com> wrote: >Anyone who used the old machines can remember how much science could which old machines? PDP-8s? Tube machines? >be done with them. Can we now do billions of times more science? And you are measuring science how? I saw Barbara Mikulski during the State of the Union. And that's what this reminded me. >As far as I'm concerned, it all went wrong when computer rooms became >warehouses. Go hang Lynn. He worked for IBM. And you think integrated circuitry is much better? I'd worry more about the distractions of software. -- Looking for an H-912 (container). ------------ And now a word from our sponsor ------------------ For a quality usenet news server, try DNEWS, easy to install, fast, efficient and reliable. For home servers or carrier class installations with millions of users it will allow you to grow! ---- See http://netwinsite.com/sponsor/sponsor_dnews.htm ----
From: Eugene Miya on 2 Feb 2010 18:26
I have to run. This is it for the day. In article <062e5d6d-6a1b-49f0-9b34-a09be3cf32c4(a)h2g2000yqj.googlegroups.com>, Robert Myers <rbmyersusa(a)gmail.com> wrote: >On Feb 2, 5:20=A0pm, eug...(a)cse.ucsc.edu (Eugene Miya) wrote: >> Sid Fernbach, for whom one of the IEEE awards is given, started a briefly >> class classification of supercomputers, and outsiders wondered what they >> VAX, the mainframe, their IBM PC rated on his scale where a Cray 1 was a >> "Class 6" machine. =A0Since I had Sid handy, I had a chance to ask him: >> Oh, they are "class 1/2". =A0They barely even rate. >> "They aren't supercomputers." > >Such a comment would require a defense. As it stands, I disagree with >it. You can say whatever you think. 1) Sid's dead. 2) No one, not even the DOE, uses his scale. 3) Sid said lots of other things PC users would howl over. VAX users as well (however neither CTSS nor NLTSS was planned for the VAX architecture, but they insisted all their tools and their AEC/ERDA/DOE environment were superior). >There are vast holes in my knowledge and experience of computers, but >I went practically straight from a Cray to a PC (admittedly aided by >an expansion bus card running a NS320xx). My current PC is hands-down >more powerful than the Cray ever was to me as a user. That's the nature of hindsight. You had a stand alone Cray? How luck of you! You likely really mean shared Cray. It would also depend whether you were in the COS or similar batch side of the world vs. the insistent interactive CTSS side of the world. >If there are differences, they are differences that have obsessed >computer professionals, like the ability to justify their budgets by >serving the needs of dozens or hundreds of computer users >simultaneously. The Cray had mainframe-like characteristics (like >peripheral processors) that PC's don't currently have, but those >differences don't matter to me as a user, and the larger range of >software, more immediate access to data and superior graphics, along >with enough power to do interesting things, make the PC seem like a >big step up to me. "And the great thing about the Alto is that it doesn't run faster at night..." to quote a paper. >The one thing wrong that I see is that the step up to whatever is >difficult or even nearly impossible. When you get to the point where >more horsepower is required, even the step to a personal beowulf >cluster is painful--and even that might not have been so had Intel not >made the decision it made about infiniband. > >As it stands right now, the old advice--wait a few years--is a long >way from running out of steam. The publicity hounds at the national >labs will run you over with their spectacular graphics? That's a >given. So I have to get this quote from another machine with the data base from an anthropologist working at LANL: I interviewed one experimental physicist, a widely respected senior researcher, who rather scathingly likened ASCI simulations to a well-known Microsoft screen saver: You know the one with the fish on it? So you're watching it, and you see that the bubbles from the fish are going up to the surface of the water. And it all looks just fine, the fish and the bubbles - until you look at the bubbles. The small ones are rising faster than the big ones. Is that right? Of course it's not right. In reality, the large bubbles always rise faster. But if you had never conducted an experiment to verify the picture you were seeing, you'd never know that what the screen saver was telling you is all wrong. Well, ASCI is just like that, it's one big expensive screen saver. -- Looking for an H-912 (container). |