He has lots of other cool ideas, such as using magnetic propulsion to levitate satelites by having them capture slugs fired from the ground, reverse their course, and fire them downwards again (== upwards momentum on both halves of the equation). I'd draw ascii-art, but your loving slashdot team have deemed that "lame" so you get none.
Satelites that levitate above the poles by tacking with solar sails against gravity (higher the "orbit", smaller the sail needed).
His book "indistinguishable from magic" is basically the worlds best anti-patent protection. more power (postumously) to him.
I'm from there, and in many more ways than highlighted by the article, Sweden fails to excell. NB: comments are about the government, not the people, who I miss.
For some reason, the media and government are a bunch of wishy-washy whiners, incapable of seeing far beyond the end of their noses. I think the problem is that politics isn't really a road to fame and power there (egalitarian society, dontchano), so the people who end up running for politics are well meaning incompetents.(*) You know where you get by good intentions.
I could very well see them putting this sort of levvy on blank CDs and then be suprised when sales plummet. It's like the government doesn't realize they exist in a global economy.
Not that it is the worst country in the world, but there is a reason I don't live there anymore.
BTW: Can't say that I've ever come accross that much swedish erotica, in much the same way that swedish fish aren't terribly popular there. Gott-o-Blandat, on the other hand, rocks. Salt-o-Blandat even more so.
(*) with some exceptions. Apparently a girl I went to high-school with is the Green Party's spokesperson. Sharp as a tack, that one.
Darwin awards apparently went to some guy (who alone with the rest of the nominees, exists only in the minds of the awards comittee) who *drank* the stuff.
The article went into gruesome detail about ruptured organs, nitrogen narcosis, and the bends...
I'll respond, since there still isn't an "incorrect" moderation option.
Linux is actually REALLY bad about avoiding disk io. Perhaps if you ran from a ramdisk root and unmounted the disk. There was also talk about running IP filters with a halted kernel, which would seem to as IO-less as you could expect to get. however, in these cases, you really don't need a disk at all, as you could net-boot the computer.
There are a couple of patches out there for 2.4.20, (google for morton+laptop mode) and there's also noflushd. The fundamental problem is that evey n seconds or so, all dirty buffers get written to disk. The only scenario in which you have no dirty buffers would be... when the laptop does almost nothing. Which isn't very useful. These patches turn off automatic flushing until the disk spins up otherwise, or all memory is taken up with dirty buffers. One of them is smarter about it than the other, but I forget which.
DCT, actually, and the rouding and quantiziation of coefficents is lossy, yes, but nothing compared to the havoc wreaked by the psycho-acoustic modelling.
Perhaps that's what you meant by throwing away. Losses are incurred in many places. The whole DCT and quantitize phase is very similar to jpegs', BTW.
allows you do do image composition in the browser.
If I have a nice rendered title image (as slashdot, up in the corner), I can render that as just the letters, using the alpha channel to anti-alias the letters against the backdrop -- which could be anything. This way, if I have several backdrops (blue, green, some picture), I only need one rendered image. Thus, I only need to render the logical items, and can specify them seperately to have my browser, which compositions them.
The alternative is to assemble each physical area on the screen into one image. Now I have to worry about those lining up with the rest of the page -- which typically leads to hardcoding pixel sizes, using invisible spacer gifs, and any number of hacks. These hacks don't help maintainability or accessability.
If I had a single mod point I would pun-ish you!!!
You forgot the best one:
If I said you had a beautiful body...
would you hold it against me?
That has got to be one of the funniest. sketches. evar. I know several of the pythons think that the fish slapping dance (which i have never seen...) is the best, but besides the english-hungarian phrasebook (this sketch), the world-championships in hide-and-seek, the...
Oh my! I just realize (sp?) that I have MP withdrawal. I need a fix of absurdity. STAT!
I bought the 23" sony, with the same res. I liked it, but didn't love it. Analysis of my usage habits suggested that I was paying a premium for the 16x9 formfactor but wasn't using the extra pixels that got me.
So I returned it and bought the dell 20". Much cheaper (close to $2k difference), and I only gave up 320 pixels in width, which I never used anyways.
The sony WAS better for DVDs but that was about it. I only wish the Dell had more flexible physical adjustments -- especially heightwise. I'm *this* close to attacking the mounting bracket with a hacksaw to make it lower, but not just yet.
This will likely be the last LCD I own, as by the the time display technology have made 4MP displays common, I'm sure oleds will finally have arrived. You just can't reliably make an LCD that big. The yeild just isn't up to making it worthwhile. Unless they start selling high-defect versions cheaper.
Marketing droids (who I HOPE read this forum -- idiots to ignore so many early adopters): I would buy a high-defect 30" display for use as an entertainment center: the defects don't distract from movies that much, and since you won't use it for long coding sessions, shouldn't be too much of a bother for couch-surfing. But it would have to cost in the $1K range.
The compatibility problem is that the arrays get it wrong.
Given animals aliased to cats, it would be statically type safe for me to say animals[0]=new Dog();. But this causes a runtime error. Runtime errors on assignment!?! Array subtyping is NOT a feature (although I would have to think for a bit what to replace it with before removing it).
Foo<Bar> isn' a subtype of Foo (which in this context should be read Foo<Object>), is it? So why should it be a subclass?
That would require covariance in the type parameter implying subtyping, and a clear counter-argument is void set_element(<param>). For Foo<Bar>, it accepts only Bar, but for Foo it accepts any Object. Thus, treating a Foo<Bar> as a Foo will cause runtime errors.
Java has already made this mistake with arrays.
They so should have done the Right Thing and taken a more powerful generics system which allows mixins to be expressed.
I must admit I was suprised that all the tests were by the proprietary manufacturers. I'm assuming that this means that each testing utility will use undocumented info to optimize the testing.
Which may be a good thing.
Here's why this is not 100% a happy answer: I do have a machine standing idle, but that is the same machine that the drive was behaving poorly in before, so I either have to [temporarily] decommission a real machine to run good tests, or run proprietary tests in the iffy machine (pentium 75, fwitw, and the cheapest that could be found in glasgow at the time: "Bits and PCs" home assembled brand)
However, I _was_ really hoping for a fairly cross-platform utility that simulated real-world create/read/write cycles. After all, all drives (as far as I'm concerned, should have made this clear in the parent) speak IDE, so the tester should just send a stream of IDE commands and verify that they are correct.
Basically, I want something that I can start before leaving the house in the morning, and come home to a verdict: clean, acceptable, marginal, sucks.
But since I haven't yet dl'ed the maxtor utils (should have said which, sorry) and tried 'em out, perhaps I am merely grousing about non-issues. Wouldn't be the first time.
Recommend some good ones, appart from the ones that can be run automatically from fdisk (badblocks...?), please.
I have an old 20G drive that was losing data in an older system. I'm looking for some stress test to figure out whether it was the MB/Chipset or the drive.
ah. Cool. So the problem is that the binary interface is computed over the optimized code, rather than the unoptimized code.
I agree: this seems like a bug (not necessarily in GCC, perhaps in the language spec). The optimized code should provide (and require) the same interface (and semantics) as the optimized code.
But then, I'm just playing armchair language designer here. C has a long history of getting by pretty well by doing what is useful, rather than "right".
nope. I'm still confused. How could not inlining a function have a semantic effect? I mean, inlining _copies_ the function, so it should be completely transparent.
Don't forget the oh-so-cool Fastest FFT In The West.
IIRC, it will partically evaluate the code against the known size of the input, and I think also do some data-driven special-casing.
Basically, it beats the pants off standard-library FFTs.
While I'm at it, responding to grandparent:
most scientific software still doesn't use such libraries
Would you care to elaborate? I mean, if you're not writing against known ultra-optimized libraries, what business do you have expecting your software to run fast? That's like compiling with gcc instead of intel's compiler. I would expect that most scientists DO care enough to use the fastest libraries at hand, and put some effort into identifying bottlenecks.
Or perhaps you were implying that most scientific software is too esoteric to benefit from fast linear-algebra? From what I recall from my physics courses, it was pretty much ALL linear algebra: vectors/ matrices/ determinants / eigen values.
But there's like three people in the world who actually use altivec. Hardly optimizing for the common case... (that said, I do run mplayer which I think relies on similar SIMD instructions on the x86 to provide realtime img post-processing).
Wouldn't it be cheaper for all concerned for apple to seriously subsidize a PCI photoshop accelerator (basically a card full of DSPs and RAM with a wide bus)?
I'm thinking back to the days when I would read BYTE and to drool over this black (really cool, in those days) 486/Pentium box filled to the brim with i960 cards. By a company called... erm.. Microway? With their compiler, it could reach a peak of 1 gigaflop. Back in the early ninety's. Do some reverse Moore's law extrapolation to see how impressive those numbers were.
Now if you're going to cater to heavy number crunching, it seems that's the way to go. Just admit that the CPU is there to deal with the UI, and let the DSPs do the effects.
So you're the guy keeping Monster Cable and other witchdoctors in business. I have one of those green markers, for sale.. cheap!:-)
I can't agree at all. As long as you're not running over powercords scavenged from your grandmothers old lamps, then every other part of your system has a greater impact. Speakers are the most important factor, both visbly and audibly. Money spent on a system with crap speakers is money wasted. Buy the speakers you need, the components you can afford, and steal cables from your local building site (once again, as long as that isn't where the old lamp cords hang out).
Cat 5 does make great wiring, as does the heavy-gauge single-copper-core grounding wire you can get at home despot.
I'm lucky if my wireless gets 300kb. (bits!). Enough to stream mp3s, but not much else. I dunno what the problem is, but it isn't range (less than 5 m).
Last year, I had three different computing environments: Home, University, and (part-time) professional programming.
The university was a unix environment administered by smart admins, who while draconian at times, would apologise weeks in advance for downtime.
At home, I had several linux boxen administered by a guy whose sysadmin skill were at best lacking -- me.
Unfortunately, the company saw fit to make that same sysadmin take care of installing software on my windows box.
I'll let you guess which of these enviroments was the most stable and productive. My linux box sure looks nice, but it crashed twice on me yesterday.
I can't tell you how many hours of productivity I've lost doing sysadmin stuff on windows: installing oracle, deinstalling oracle, connecting to team streams, applying patches...
It boggles my mind that people WANT to do that sort of stuff.
Slashdot Mirror
He has lots of other cool ideas, such as using magnetic propulsion to levitate satelites by having them capture slugs fired from the ground, reverse their course, and fire them downwards again (== upwards momentum on both halves of the equation). I'd draw ascii-art, but your loving slashdot team have deemed that "lame" so you get none.
Satelites that levitate above the poles by tacking with solar sails against gravity (higher the "orbit", smaller the sail needed).
His book "indistinguishable from magic" is basically the worlds best anti-patent protection. more power (postumously) to him.
I'm from there, and in many more ways than highlighted by the article, Sweden fails to excell. NB: comments are about the government, not the people, who I miss.
For some reason, the media and government are a bunch of wishy-washy whiners, incapable of seeing far beyond the end of their noses. I think the problem is that politics isn't really a road to fame and power there (egalitarian society, dontchano), so the people who end up running for politics are well meaning incompetents.(*) You know where you get by good intentions.
I could very well see them putting this sort of levvy on blank CDs and then be suprised when sales plummet. It's like the government doesn't realize they exist in a global economy.
Not that it is the worst country in the world, but there is a reason I don't live there anymore.
BTW: Can't say that I've ever come accross that much swedish erotica, in much the same way that swedish fish aren't terribly popular there. Gott-o-Blandat, on the other hand, rocks. Salt-o-Blandat even more so.
(*) with some exceptions. Apparently a girl I went to high-school with is the Green Party's spokesperson. Sharp as a tack, that one.
Ok, o2 is very reactive, but doesn't it need heat to kick-start it? I mean, otherwise all the O2 in the air would be reacting, right?
Or is it a matter of concentrations? (kinda like the math we did in chemistry about ions and osmosis)
Darwin awards apparently went to some guy (who alone with the rest of the nominees, exists only in the minds of the awards comittee) who *drank* the stuff.
The article went into gruesome detail about ruptured organs, nitrogen narcosis, and the bends...
It took me several minutes of surfing to figure out that your client is for the windows platform. Perhaps you could state that clearly, upfront?
Just a suggestion.
I'll respond, since there still isn't an "incorrect" moderation option.
Linux is actually REALLY bad about avoiding disk io. Perhaps if you ran from a ramdisk root and unmounted the disk. There was also talk about running IP filters with a halted kernel, which would seem to as IO-less as you could expect to get. however, in these cases, you really don't need a disk at all, as you could net-boot the computer.
There are a couple of patches out there for 2.4.20, (google for morton+laptop mode) and there's also noflushd. The fundamental problem is that evey n seconds or so, all dirty buffers get written to disk. The only scenario in which you have no dirty buffers would be... when the laptop does almost nothing. Which isn't very useful. These patches turn off automatic flushing until the disk spins up otherwise, or all memory is taken up with dirty buffers. One of them is smarter about it than the other, but I forget which.
DCT, actually, and the rouding and quantiziation of coefficents is lossy, yes, but nothing compared to the havoc wreaked by the psycho-acoustic modelling.
Perhaps that's what you meant by throwing away. Losses are incurred in many places. The whole DCT and quantitize phase is very similar to jpegs', BTW.
I actually read about some guy who was trying to rip a beachboys CD in his 52X drive and the disc shattered, completely fsking up the drive.
allows you do do image composition in the browser.
If I have a nice rendered title image (as slashdot, up in the corner), I can render that as just the letters, using the alpha channel to anti-alias the letters against the backdrop -- which could be anything. This way, if I have several backdrops (blue, green, some picture), I only need one rendered image. Thus, I only need to render the logical items, and can specify them seperately to have my browser, which compositions them.
The alternative is to assemble each physical area on the screen into one image. Now I have to worry about those lining up with the rest of the page -- which typically leads to hardcoding pixel sizes, using invisible spacer gifs, and any number of hacks. These hacks don't help maintainability or accessability.
So not important, but useful.
You forgot the best one:
That has got to be one of the funniest. sketches. evar. I know several of the pythons think that the fish slapping dance (which i have never seen...) is the best, but besides the english-hungarian phrasebook (this sketch), the world-championships in hide-and-seek, the...
Oh my! I just realize (sp?) that I have MP withdrawal. I need a fix of absurdity. STAT!
I bought the 23" sony, with the same res. I liked it, but didn't love it. Analysis of my usage habits suggested that I was paying a premium for the 16x9 formfactor but wasn't using the extra pixels that got me.
So I returned it and bought the dell 20". Much cheaper (close to $2k difference), and I only gave up 320 pixels in width, which I never used anyways.
The sony WAS better for DVDs but that was about it. I only wish the Dell had more flexible physical adjustments -- especially heightwise. I'm *this* close to attacking the mounting bracket with a hacksaw to make it lower, but not just yet.
This will likely be the last LCD I own, as by the the time display technology have made 4MP displays common, I'm sure oleds will finally have arrived. You just can't reliably make an LCD that big. The yeild just isn't up to making it worthwhile. Unless they start selling high-defect versions cheaper.
Marketing droids (who I HOPE read this forum -- idiots to ignore so many early adopters): I would buy a high-defect 30" display for use as an entertainment center: the defects don't distract from movies that much, and since you won't use it for long coding sessions, shouldn't be too much of a bother for couch-surfing. But it would have to cost in the $1K range.
The compatibility problem is that the arrays get it wrong.
Given animals aliased to cats, it would be statically type safe for me to say animals[0]=new Dog();. But this causes a runtime error. Runtime errors on assignment!?! Array subtyping is NOT a feature (although I would have to think for a bit what to replace it with before removing it).
but... but...
Foo<Bar> isn' a subtype of Foo (which in this context should be read Foo<Object>), is it? So why should it be a subclass?
That would require covariance in the type parameter implying subtyping, and a clear counter-argument is
void set_element(<param>). For Foo<Bar>, it accepts only Bar, but for Foo it accepts any Object. Thus, treating a Foo<Bar> as a Foo will cause runtime errors.
Java has already made this mistake with arrays.
They so should have done the Right Thing and taken a more powerful generics system which allows mixins to be expressed.
RH 9 has a woefully sucky kernel, and it WILL cause skipping under consistent IO load.
Thanks to all responses.
I must admit I was suprised that all the tests were by the proprietary manufacturers. I'm assuming that this means that each testing utility will use undocumented info to optimize the testing.
Which may be a good thing.
Here's why this is not 100% a happy answer: I do have a machine standing idle, but that is the same machine that the drive was behaving poorly in before, so I either have to [temporarily] decommission a real machine to run good tests, or run proprietary tests in the iffy machine (pentium 75, fwitw, and the cheapest that could be found in glasgow at the time: "Bits and PCs" home assembled brand)
However, I _was_ really hoping for a fairly cross-platform utility that simulated real-world create/read/write cycles. After all, all drives (as far as I'm concerned, should have made this clear in the parent) speak IDE, so the tester should just send a stream of IDE commands and verify that they are correct.
Basically, I want something that I can start before leaving the house in the morning, and come home to a verdict: clean, acceptable, marginal, sucks.
But since I haven't yet dl'ed the maxtor utils (should have said which, sorry) and tried 'em out, perhaps I am merely grousing about non-issues. Wouldn't be the first time.
Disk diagnostics?
Recommend some good ones, appart from the ones that can be run automatically from fdisk (badblocks...?), please.
I have an old 20G drive that was losing data in an older system. I'm looking for some stress test to figure out whether it was the MB/Chipset or the drive.
let me take the opportunity to advertise my favorite. research paper. evar.
Pessimal Algorithms and Simplexity analysis
Only tangentially related. but good.
ah. Cool. So the problem is that the binary interface is computed over the optimized code, rather than the unoptimized code.
I agree: this seems like a bug (not necessarily in GCC, perhaps in the language spec). The optimized code should provide (and require) the same interface (and semantics) as the optimized code.
But then, I'm just playing armchair language designer here. C has a long history of getting by pretty well by doing what is useful, rather than "right".
nope. I'm still confused. How could not inlining a function have a semantic effect? I mean, inlining _copies_ the function, so it should be completely transparent.
Obviously I'm wrong. How?
IIRC, it will partically evaluate the code against the known size of the input, and I think also do some data-driven special-casing.
Basically, it beats the pants off standard-library FFTs.
While I'm at it, responding to grandparent:
Would you care to elaborate? I mean, if you're not writing against known ultra-optimized libraries, what business do you have expecting your software to run fast? That's like compiling with gcc instead of intel's compiler. I would expect that most scientists DO care enough to use the fastest libraries at hand, and put some effort into identifying bottlenecks.
Or perhaps you were implying that most scientific software is too esoteric to benefit from fast linear-algebra? From what I recall from my physics courses, it was pretty much ALL linear algebra: vectors/ matrices/ determinants / eigen values.
In summarium, this position confuses me.
ya.
But there's like three people in the world who actually use altivec. Hardly optimizing for the common case... (that said, I do run mplayer which I think relies on similar SIMD instructions on the x86 to provide realtime img post-processing).
Wouldn't it be cheaper for all concerned for apple to seriously subsidize a PCI photoshop accelerator (basically a card full of DSPs and RAM with a wide bus)?
I'm thinking back to the days when I would read BYTE and to drool over this black (really cool, in those days) 486/Pentium box filled to the brim with i960 cards. By a company called... erm.. Microway? With their compiler, it could reach a peak of 1 gigaflop. Back in the early ninety's. Do some reverse Moore's law extrapolation to see how impressive those numbers were.
Now if you're going to cater to heavy number crunching, it seems that's the way to go. Just admit that the CPU is there to deal with the UI, and let the DSPs do the effects.
Has anyone used one of those machines?
thanks for the tip. Will try when I get home.
So you're the guy keeping Monster Cable and other witchdoctors in business. I have one of those green markers, for sale.. cheap! :-)
I can't agree at all. As long as you're not running over powercords scavenged from your grandmothers old lamps, then every other part of your system has a greater impact. Speakers are the most important factor, both visbly and audibly. Money spent on a system with crap speakers is money wasted. Buy the speakers you need, the components you can afford, and steal cables from your local building site (once again, as long as that isn't where the old lamp cords hang out).
Cat 5 does make great wiring, as does the heavy-gauge single-copper-core grounding wire you can get at home despot.
I'm lucky if my wireless gets 300kb. (bits!). Enough to stream mp3s, but not much else. I dunno what the problem is, but it isn't range (less than 5 m).
Last year, I had three different computing environments: Home, University, and (part-time) professional programming.
The university was a unix environment administered by smart admins, who while draconian at times, would apologise weeks in advance for downtime.
At home, I had several linux boxen administered by a guy whose sysadmin skill were at best lacking -- me.
Unfortunately, the company saw fit to make that same sysadmin take care of installing software on my windows box.
I'll let you guess which of these enviroments was the most stable and productive. My linux box sure looks nice, but it crashed twice on me yesterday.
I can't tell you how many hours of productivity I've lost doing sysadmin stuff on windows: installing oracle, deinstalling oracle, connecting to team streams, applying patches...
It boggles my mind that people WANT to do that sort of stuff.