WTF? "real-timeness adds overhead"? The whole point of RTOSes is to remove the overhead and turn almost everything over to the application. Lots of RTOSes have only basic memory management and I/O is largely limited to device discovery and initialization -- everything else is left to the application, because only the application can define how long an operation may take. Some telephone switches can even be rebooted while the applications run!
You might want to take a couple of computer classes before you post here again.
The K8's integrated memory controller supports much better memory throughput.
That's fine in theory, but that's not proven by the results in the test. As shown on this page, memory access is pretty even across the board, with the exception of the 570 SLI, which showed remarkably higher latency (almost 50% higher). As they mention in the article, that could be an aberration with the particular board they used, but they also noted that the board was supplied by nVidia, so it should (hopefully) be one that showed good performance.
(a styx album) that had the band name only show up when the light hit it at the right angle
Pretty close, it was a Styx album (Paradise Theater), but the graphic was of a twenties-era woman in profile, supposedly from the cornice of the album's namesake.
One cool thing about vinyl that nobody's mentioned is the interaction you had with the whole process. When I was in college, friends would come over and we'd listen to music, and it was almost an event. You had to get up, choose an album, then remove it from the sleeve with that practiced motion that left you with the edge in your hand and your thumb on the label. Then you'd carefully place it on the turntable, flipping it deftly with two fingers. Then, you'd cue up the tone arm, but before you lowered it, you pulled out the record cleaner. Three drops on the surface, spread them across using the base of the bottle, then gently touch the cleaner to the surface. Let it spin once, then tap the cue button to lower the tonearm and the music would start. And you got to do this about every twenty minutes!
When my friends started getting CD players, we would all ooh and aah, and agree that the music sounded much better (even though it didn't: the players my friends could afford were the cheapies that were amazingly harsh and brittle, with a big 'ol notch filter around 16KHz to take some of the bite out of the highs). Ah, memories...
Hell, buy some new ones -- they're still in production! Along with most of the classic Klipsch speakers: La Scala, Heresy, I see the Cornwall is up to version III. No Belle Klipsch, though.
if demand increases, but you can still only make 1,000,000 of gas a day, you have to adjust your price such that the demand is at 1,000,000 gallons a day again
There's no requirement that you satisfy demand. Increased demand only means increased competition, which makes it possible to increase prices with the assurance that they'll be met. If you raised prices and the market was elastic, then you'd wind up losing sales as the market curbed its consumption. That's why big companies hire economics graduates, to determine the maximum price they can charge for their product based on the demand. GP's point is that the market is inelastic, because people can't reasonably reduce their consumption beyond a certain (relatively small) amount.
just because someone is a an engineer at Porsche doesn't mean they really know what they're talking about
Uh, if they're talking gas-powered engines and fuel requirements, then they really, really, do. Porsche is a pretty small operation and only have a couple dozen engineers, and the ones they have are very good with cars.
Sounds like somebody needs a hybrid! I used to get 40-41MPG when I was spending 90% of my commute on I-80 at 75MPH, I'm getting more like 47 now that I'm just commuting in-town. And the Civic is a good family car, unless your family is large enough to need more than five seats. You'd save (28080 / 41 = 685) about 355 gallons a year, or about $825 at today's prices, even more if it goes back up to $3/gallon. That's about $65 a month, certainly not enough to justify buying one if your current car is paid for, but something to think about when it comes time to replace your ride.
Because that's called a laptop without a battery and is fuckin' pointless.
Uh, I'm gonna have to call bullshit on this one. Your average computer buyer (c.f., Slashdot poster) buys a whole package when they upgrade: CPU, monitor, keyboard, mouse, and often printer. Then they give the old system to Grandma or the kids or it goes into the TV room as the "family" computer. Nobody buys just a new motherboard or CPU tower, because Wal-Mart/Best Buy/CompUSA/Office Max doesn't stock 'em.
Sounds like a performance enhancement to me. Instead of searching through blocks of records, looking for an open slot to stick a new record in, Postgres just sticks the new record in the first available space in the last block. My guess is the "vacuum" process just consolidates the records, eliminating the deleted ones. Compare this with Oracle's "row chaining" issues (where a table row must be split across multiple blocks, necessitating multiple I/Os for a single row).
ISTR that as of PostreSQL 8, the vacuum process is automatic and runs periodically while the database is up.
I think the eternal problem with MySQL is that everyone thinks that just because "SQL" is in the name it's a relational database. It's not. Sure, it's got tables and you can join tables together and use SQL queries, but it wasn't originally designed to do the things that a relational database must do. It was designed to be a quick, easy-to-use database that made developer's lives easier. And from that standpoint, it does well.
FWIW, the commercial database UNIFY used to be pretty much the same thing back in the mid-80s. They had a wicked-fast ISAM database, and then they wrapped that all up in an SQL wrapper. They were a little more concientious, though, so you had guaranteed atomic transactions and rollback capability and more complete SQL support (e.g., nested/correlated subqueries), so it was truly relational (as the term is generally used). Horrible syntax-based optimizer, though (actually, I'm not even convinced it was an optimizer, it was probably just the way their SQL parser interpreted the query).
Maybe some out-of-work geeks just started hanging around M5 to lend a hand? Unpaid, but maybe they could talk Jamie into letting them put "Crewed for M5 (Mythbusters)" on their resume. That's gotta be worth something!
I was under the impression that Python wasn't so much interpreted as compiled to bytecodes which ran in a special run-time environment (kind of a lightweight JVM). Javascript and perl are interpreted, true, but you can't really interact with the interpreter directly (unless that's what the perl debugger is, I actually haven't used that).
The public library in my home town had a couple of Apple ][s (back in the early 80s) with Logo on them. I wrote an assembler routine that let kids print out their graphics (hires screen to a Grappler printer card, those were the days!) and defined a Logo "word" to invoke it. I always wanted to teach a kids programming class there, but by then I was in college so I never had the opportunity. I hope somebody else did.
The same college, now starts teaching kids some RAD languages, Java, C# and whatnot.
This isn't necessarily a bad thing. When I started college, the language du jour was Pascal, so that's what we learned and used for our first-year courses. However, in order to take any of the high-level courses (language design, compiler design, OS design), you had to take the assembly language course.
We also had a file systems design course where we implemented the UNIX v6 filesystems (block devices, inodes and all) in Pascal. Now THAT was nasty.... (At the time. Now I see it as one of the best courses I ever took.)
the good thing about BASIC is not that it was easy
No, the thing about BASIC was that it was interpreted, so you had immediate gratification ("? 'cool!'" and your computer said "cool!"), plus it made debugging pretty easy:
INTEGER OVERFLOW ON LINE 430 > LIST 430 430 LET Z=2^X+1 > ? X X: 16 > 430 IF X CONTINUE
Uh huh. But how does it do that? "Daddy, what's a layout manager?" "Daddy, why do I have to put a dot there?" "Daddy, why do I have to put a square brackets here, but curly braces there?" "Why do I have to use a dollar sign on name here, but not there?"
TCL may have a place (I've used it, but I always wash my hands real well afterwards), but I wouldn't suggest it as a teaching language.
Actually, I understood the GP to mean that lots of people sign up for CS programs, then wind up bailing or failing because they didn't know how much "extraneous" work was involved.
The point of all the math is to make the subject sufficiently academic that it is acceptable to universities
Not at all, the math is there to explain why things work the way they do, and why an O(n log n) routine is better than an O(n^2) one. It's also handy to be able to recognize an NP problem, before you agree to write the solution in two weeks!;)
Seriously, I can't imagine developing software without at least a passing familiarity with discrete math. It's not that hard (I found it easier than trig, but then I suxx0r3d at trig), and it teaches you how to think about the kind of problems that are common in programming (e.g., graph traversal).
move things core to core (or CPU to CPU case being
Does Windows really move things around, or do they just spread themselves around naturally? I mean, I would think that a good scheduler would try to keep CPU/core affinity, but would for the most part schedule processes on the least-loaded processor. Does anyone know if CPU affinity really pays off on typical job mixes, or do caches tend to get flushed too often to make it worthwhile?
Yes, but do the professors record them themselves, then edit them for level and time before making them available? I think that recording lectures and making them available to students taking the course is something that every school should do, but it should be done by the school, and shouldn't take up any of the professor's time. Wire the lecture halls, get some a/v monkey to set the recording levels and set up the whiteboards, then record it and put it up on the student portal. But personally, I don't have a problem with the professor charging chump change (less than a pitcher of beer!) for his lectures if he has to make the extra effort to make them available.
Slashdot Mirror
WTF? "real-timeness adds overhead"? The whole point of RTOSes is to remove the overhead and turn almost everything over to the application. Lots of RTOSes have only basic memory management and I/O is largely limited to device discovery and initialization -- everything else is left to the application, because only the application can define how long an operation may take. Some telephone switches can even be rebooted while the applications run!
You might want to take a couple of computer classes before you post here again.
Heh. That'll teach me to post while I'm in a training class...
(Damn caffiene-deficient SLC tech center!)
One cool thing about vinyl that nobody's mentioned is the interaction you had with the whole process. When I was in college, friends would come over and we'd listen to music, and it was almost an event. You had to get up, choose an album, then remove it from the sleeve with that practiced motion that left you with the edge in your hand and your thumb on the label. Then you'd carefully place it on the turntable, flipping it deftly with two fingers. Then, you'd cue up the tone arm, but before you lowered it, you pulled out the record cleaner. Three drops on the surface, spread them across using the base of the bottle, then gently touch the cleaner to the surface. Let it spin once, then tap the cue button to lower the tonearm and the music would start. And you got to do this about every twenty minutes!
When my friends started getting CD players, we would all ooh and aah, and agree that the music sounded much better (even though it didn't: the players my friends could afford were the cheapies that were amazingly harsh and brittle, with a big 'ol notch filter around 16KHz to take some of the bite out of the highs). Ah, memories...
Hell, buy some new ones -- they're still in production! Along with most of the classic Klipsch speakers: La Scala, Heresy, I see the Cornwall is up to version III. No Belle Klipsch, though.
You need one of those little plastic adapter thingies....
There's no requirement that you satisfy demand. Increased demand only means increased competition, which makes it possible to increase prices with the assurance that they'll be met. If you raised prices and the market was elastic, then you'd wind up losing sales as the market curbed its consumption. That's why big companies hire economics graduates, to determine the maximum price they can charge for their product based on the demand. GP's point is that the market is inelastic, because people can't reasonably reduce their consumption beyond a certain (relatively small) amount.
Sounds like somebody needs a hybrid! I used to get 40-41MPG when I was spending 90% of my commute on I-80 at 75MPH, I'm getting more like 47 now that I'm just commuting in-town. And the Civic is a good family car, unless your family is large enough to need more than five seats. You'd save (28080 / 41 = 685) about 355 gallons a year, or about $825 at today's prices, even more if it goes back up to $3/gallon. That's about $65 a month, certainly not enough to justify buying one if your current car is paid for, but something to think about when it comes time to replace your ride.
Sounds like a performance enhancement to me. Instead of searching through blocks of records, looking for an open slot to stick a new record in, Postgres just sticks the new record in the first available space in the last block. My guess is the "vacuum" process just consolidates the records, eliminating the deleted ones. Compare this with Oracle's "row chaining" issues (where a table row must be split across multiple blocks, necessitating multiple I/Os for a single row).
ISTR that as of PostreSQL 8, the vacuum process is automatic and runs periodically while the database is up.
FWIW, the commercial database UNIFY used to be pretty much the same thing back in the mid-80s. They had a wicked-fast ISAM database, and then they wrapped that all up in an SQL wrapper. They were a little more concientious, though, so you had guaranteed atomic transactions and rollback capability and more complete SQL support (e.g., nested/correlated subqueries), so it was truly relational (as the term is generally used). Horrible syntax-based optimizer, though (actually, I'm not even convinced it was an optimizer, it was probably just the way their SQL parser interpreted the query).
Maybe some out-of-work geeks just started hanging around M5 to lend a hand? Unpaid, but maybe they could talk Jamie into letting them put "Crewed for M5 (Mythbusters)" on their resume. That's gotta be worth something!
Are those the ones that turn into robots when they land? Because I think we definitely need more of those!
LOGO isn't interpreted anymore?
I was under the impression that Python wasn't so much interpreted as compiled to bytecodes which ran in a special run-time environment (kind of a lightweight JVM). Javascript and perl are interpreted, true, but you can't really interact with the interpreter directly (unless that's what the perl debugger is, I actually haven't used that).
The public library in my home town had a couple of Apple ][s (back in the early 80s) with Logo on them. I wrote an assembler routine that let kids print out their graphics (hires screen to a Grappler printer card, those were the days!) and defined a Logo "word" to invoke it. I always wanted to teach a kids programming class there, but by then I was in college so I never had the opportunity. I hope somebody else did.
We also had a file systems design course where we implemented the UNIX v6 filesystems (block devices, inodes and all) in Pascal. Now THAT was nasty.... (At the time. Now I see it as one of the best courses I ever took.)
D'oh! That should read:
> 430 IF X < 16 THEN Z=2^X+1 ELSE Z=-1
> CONTINUE
INTEGER OVERFLOW ON LINE 430
> LIST 430
430 LET Z=2^X+1
> ? X
X: 16
> 430 IF X CONTINUE
This was the great strength of BASIC.
Uh huh. But how does it do that? "Daddy, what's a layout manager?" "Daddy, why do I have to put a dot there?" "Daddy, why do I have to put a square brackets here, but curly braces there?" "Why do I have to use a dollar sign on name here, but not there?"
TCL may have a place (I've used it, but I always wash my hands real well afterwards), but I wouldn't suggest it as a teaching language.
Actually, I understood the GP to mean that lots of people sign up for CS programs, then wind up bailing or failing because they didn't know how much "extraneous" work was involved.
Seriously, I can't imagine developing software without at least a passing familiarity with discrete math. It's not that hard (I found it easier than trig, but then I suxx0r3d at trig), and it teaches you how to think about the kind of problems that are common in programming (e.g., graph traversal).
Yes, but do the professors record them themselves, then edit them for level and time before making them available? I think that recording lectures and making them available to students taking the course is something that every school should do, but it should be done by the school, and shouldn't take up any of the professor's time. Wire the lecture halls, get some a/v monkey to set the recording levels and set up the whiteboards, then record it and put it up on the student portal. But personally, I don't have a problem with the professor charging chump change (less than a pitcher of beer!) for his lectures if he has to make the extra effort to make them available.