Also, you don't need to round trip one signal at a time, you can have multiple signals in flight. Which makes this whole discussion rather moot.
Nope! Having several bits flying over the wire is a form of pipelineing (check-out "wave pipelining"), which increases bandwidth, but does not do anything to decrease latency. bringing us back to my original point, which can be illustrated by a fact that good weight-lifters rarely make good boxers...;-)
Yeah, also called PIM, for "processing in memory". Peter Kogge (of Kogge-Stone fast adder fame) is doing research into those in Notre Dame (disclaimer: I used to be affiliated with the same HTMT crowd designing ultra-fast superconductor processing elements for the petaflop computer, thus I know first hand how hard it is to match memory speed to pipelined processing speed...;-) ).
Bringing processors and (large) memories closer to each other does not help much, as, as you mentioned, there is an order of magnitude difference between processor clock speed and memory access speed. The physical reason for this is that to do a certain operation on one pipeline stage in a processor you need to charge a clock line passing through a couple dozen to couple hundred gates; in memory case you have to charge the word line passing through sqrt(1G)=30,000 gates. It takes time (RC, unless one uses superconductors and forgets about R;-) ) and power (CV^2/2).
The only rerasonable solution is, indeed, to make memory blocks smaller and closer to processor elements, making them essentially registers/caches, not RAM.
Oh, and, BTW, in the rather naive picture on the link you sent, the solution will not work that well if you have multiple processors -- you have to make sure that each can talk to other's memories (in SMP case) AND to each other.
But then, the trick is that he did not mention memory latency, only bandwidth! Getting the latter is relatively easy -- just make memory bus wider (as given bus speed), trying to decrease latency will pretty soon make you run into speed-of-light limitation.
Maybe those processors do have enough memory bandwidth to load two of them completely doing SAXPY? Assuming 12 GFLOPS sustained (3 GHz, 2 cores, separate ADD and MUL on each) you need to feed input vectors at 12*8 bytes/double = 96 GB/sec, for, say 1 GHz memory bus it is translated into 96*8=768 memory pins only for input -- well, wider than I've seen on desktop PCs...;-)
When you start doing anything else , the roundtrip time between processors and memory (latency) becomes more important than raw bandwidth.
Well, I would not necessarily agree... People who buy into Lindows do not (usually) download FC2, and they would not get a RedHat boxed set, so no lost sale to RH. OTOH, after they figure out that "this Linux thing" is not as scary and quite useable they might decide to try more advanced distributions and they will have a chance to do this immediately.
Oh, well, I can appreciate the full meaning of your ~/.signature now!;-)
The lesson: Test on production data.;-( I know how it feels, a good friend of mine does similar kind of stuff for (German! European bureocracy!) medical insurance databases, the stories she told me are unbelievable to anyone with more mainstream CS background.
Well, maybe another lesson is that Java is not exactly the right language when you are mostly dealing with syscalls (lseek(), read(), printf some SQL and send it towards the DB), which kind of brings us back to the original topic...
At least the developers didn't try to mmap the archive. Maybe they should've? Then some part of what they were trying to do would shift from Java to highly-optimized (maybe even in asm;-) ) code in the kernel. Not that it is a guaranteed win, but at least deserves some consideration and thought, but the thought process requires of the guy doing it to know what the tradeoffs are, and you get this knowledge only from learning something besides Java.
The real problem is the massive, monolithic archive... A kernel module to mount the archive as a filesystem, anyone? I know, I know, but is not it sad that manufacturing gave us 3GHz processors and 300 GB disks (yeah, they just "print" those, not thinking of each individual Hz or byte) and too many programmers do not stop and think what does it take to handle a 100 GB dataset...
Of course you know your application better than anyone here, and I bet that you did do the right thing, but somethings strikes me as a little bit odd:... since we also didn't start swapping, there were fewer context switches, etc. and the actual bottleneck was I/O
Are you sure we are talking about the same 'threads', as in pthreads/clone()-based, NOT fork()ed-out _processes_? By definition the first kind all share memory space and running 10 of them instead of one should not cause excessive memory consumption/swapping.
Well, of course it is not always true, if you mmap() a 100 GB database file and unleash a whole bunch of threads to work on different (non-overlapping) parts of that you start thrashing your OS's buffer cache really soon and can bring your system to a halt rather effectively...
As a side note, I always assumed that the only reason to use more threads (in your program) than there are CPUs/contexts available is when your problem is I/O bound (you achieve the benefits of asynchronous I/O without explicitly coding it in): number of threads = number of CPUs + number which on average sleeps waiting for I/O to complete.
Ah, filesharing in the good old days... they weren't really any better, but we were a lot younger:)
Yeah, yeah...;-)
Does anyone else still set boot order to C: A: to prevent the accidental loading of an undetected BSV?
As a matter of fact, YES!;-) Actually the last time a BSV tried to bit me was a couple of months ago (!) when I picked up a floppy from the ground on a street corner in some relatively bad area of LA (do not ask me what brought me there, some cultural/artistic thing my GF participated in;-) ), was just curious what was on it... Well, among other things a BSV, but fortunately it had no idea how to infect a Linux box, but I had to forget the floppy in the drive to find out...;-)
You dound like someone working hard towards the degree and sincerely being afraid that you will not be given that if your girlfriend finds out about that other girl, or something...;-)
In The Real Life (TM) degrees are needed to a) make yourself/your parents proud and b) get employment in certain places/positions which do require this IF they like you otherwise.
Do you think anyone will be hiring THE guy any time soon on a position like this (even if he would formally have his degree)?
It is more of a symbolic jesture, really, and if you do that bad in scientific community you quite deserve this. (As an example of the opposite trend, think about the Honorary Degrees -- I guess Linus has one of those;-) -- they are generally "better" than regular ones, but still have more of a symbolic status...)
... viruses used to live in the boot sector of a 5" floppy disk, while keeping the whole bootloader there as well, AND in the worst cases there were a handful of those on a single disk (bootsector);-) I can understand the fun of writing something like this back then, but the "Hey, dude, click here for your document..." is just plain silly...;-)
Maybe because in the past decade people (ones with $$, as in "VCs";-) ) were interested more in running data around those, you know, strings which they called "Internet"?;-) And now they have to run data in that thing which the old guy Maxwell discussed, you know, the "Waves"? And the latter one takes some extra effort in doing DSP?;-)
On a more serious note, businesses want to differentiate themselves from each other (who needs another plain old stand-alone 2 megapixel camera if one can have a 5 MP one in their cellphone?), and plain Si is actually becoming cheaper (if you do not think about the top-of-the-line, 90nm one, of course, you do not need anything like that for your better-than-average consumer appliance).
And, of course, classic systolic arrays research has almost nothing to do with modern FPGAs, we will not see it re-emerging until speed-of-light signal propagation delays on chip become dominant (I have something to say about that too, but will not now...).
I see that we do agree on the most important points, just a bit of nitpicking on my side!;-)
"... wide majority of modern applications." -- yeah, the _modern_ part is what makes me kind of nervous... Is it me getting old, or, as I said to my friend some years ago, "You do not get much out of Java which you can not get from a g++ with an added flag to disable dereference operator"?
My KPilot is bored until I press the HotSync button on my Palm. Hey! This is what HLT instruction is for! Save some electrones until that hardware interrupt arrives!;-)
Do not get me started on that topic - we just don't have the time to go there.;-) Any time I hear someone say "middleware", I feel the urge to run screaming. As you could tell, I used much less respectful word for the thing...;-) I am happy I do not have to actually work on the marketplace when the intimate knowledge of the middleware flavour of the day is required (though I would not mind being paid like those guys;-) ).
Your story about Python (and above-mentioned GTK) was exactly what I would expect from a reasonable educated guy, really, especially since I pretty much hold the same views. One of the coolest systems I ever had to work with was Cadence (big EDA CAD system), most of interesting parts written in their Scheme dialect. Autocad, BTW, too...;-)
I even enterntaned for a while an idea of sitting down and writing a chip layout editor in Python and GTK (around GnomeCanvas), except that the Canvas did not keep the CTA matrices with the object groups (think, cells) and I could not have the same cell on the screen in, say, different orientations.;-) OTOH, the GnomeCanvas low-level implementation deserves to be coded not even in assembly, but in different GPU's assemblies, to bring us back to the original point!;-)
Mastery of assembly is completely pointless for anyone outside of OS kernel, compiler construcution and embedded development...which probably means you.
And why do you think that in a year or two the Si wafer real estate (or, more realistically FPGA real estate, which it is probably right now) will NOT become cheap enough that one who thinks of himself as a 'programmer' now would have to actually write VHDL for the processor part, then assembly glue, and only then port Linux to the "ultra-optimized to track your eye movements while simultaneously scanning your brain waves for the tune you'd like to listen to now" car-dashboard computer?
The line between hardware and software is disappearing, and producing way too many "programmers" who can only code Java is a sure way to the next economic disaster in this country (way worse than.com collapse).
The As I've said before, a good mechanic can tear down and reassemble an internal combustion engine, and that my argument is that a good programmer can do the same on an ASM level should the need arise.!
but, OTOH, you do have a 4-digit/. uid, after all...;-)
Due to the advancee in hardware capabilities Bloat is now excusable - and encouraged. It is cheaper to buy a big ass computer that can crank and hog bits than it is to pay somebody good money to nitpick over every bit of resource that their programs consume.
What if you want to run your app on (God forbid!;-) ) SEVERAL computers? I may be mistaken, but I do not think that you still have the mentality of "The world's market in computers is probably 5 or 6 of them" (food is burning on the stove, no time to get the exact Watson's quote;-) ).
Last time I checked the price difference between a $100 ARM-based unit and $5,000 G5 was still $4,900, so if you intend to run your program on a 1000 units you can as well pay $900,000 to a guy who can port from the latter to the former and keep the cool $4M.
Or if you think about selling your solution (HW+SW) to some millions of customers... (Wow!);-)
If I need to write a GUI app, I'll do it in Python with GTK or QT bindings.
Almost by definition any GUI app spends all its time waiting for the user to press a button (responce time in milliseconds range), so noone in their rigth mind would code one in assembly. Low-level (think BITBLT) parts of the GUI is coded in assembler or maybe even microcode (extreme cases;-) ).
You being able to pick an efficient library and sensible approach to the task at hand in no way invalidates the granny-post's point that businesses relying too much on levels and levels of crud (bought for big $$ from the companies employing people who have not learned assembly in college, not to mention high school;-) ) tend to produce code bloat.
Well, this is the second part of the equation... To my mind the premise of the original article was not exactly that everyone should code in assembly (and reading "assembly language" three times in one 4-line paragraph was driving me crasy, he should've used ASM or something;-) ), but that one should LEARN different things...
To my mind, any programmer (I would not refer to the ivory towers of CS here;-) ) should more or less know both LISP and assembly, the latter one for the reasons discussed in the arcticle, the former one if only because it kind of relates to what can be computed AT ALL, _not_ how efficiently it can be done. (see, for example, chapter three )I guess) of today's story about \Omega;-) ).
They do dumb things like multithread something that is CPU bound...
Not that I disagree with your point (actually I do agree with it alot!;-) ), but, I'd guess, almost a half of the people thinking in multithreading terms in this world (definitely half the total IQ, not to offend anyone;-) ) use it to solve CPU-bounded problems on SMP boxes/multithreaded CPUs (Tera SMT; Intel's Hyperthreading on massive (ASCI) scale, things like that...)
When you just start throwing processors (or thread contexts) on the problem, you will find soon enough that any problem is I/O bound...;-)
I'm all for free speech, but the freedom of a group or indivduals aren't absolute...
Of course it's not, it can be argued if a freedom of an individual is absolute or not (more often reasonable people agree that it still shouldn't infringe or in anyway threaten the freedoms of others, but there is no such thing as a "freedom of a group", only of the individuals in this group...
... which used to be Tera, Inc. they are in Seattle as well, I am wondering if THAT connection is more important than the MS one.
Not that Burton Smith (CTO of Tera, err., Cray, and overall VERY cool guy -- I used to work with him on a project) and Thomas Sterling ("farther of Beowulf", really!;-) ) work on the next-gen supercomputer (see, e.g.,
here, I am wonderng what the implications are...
Paul B.
Basically, the market can decide if Playboy Radio on XM is something they want to support or not... but nobody is forced to listen to that, you can't even accidently tune it on an XM device unless you're paying a monthly fee and then an extra monthly fee for that one channel.
What if someone (out of the kindness of their hearts!) would decide to give away XM radio receivers to anyone asking? No, I do not mean anyone associated with the apparently indecent ones on this mentioned radio program, just _someone_? Will your argument still hold? Can you still give your old 386 to a kid knowing that it can be used for indecent purposes?
And, after all, where is the boundary between "public" and "private" in something neither of us can hold in our hands or show to others and say "this is mine!"?
... having my mod points but rather decided to participate in the discussion...
A *big guy* with a *big antenna* is not too much different from equally *big* guy with almost as big a truck plowing over your land, but that issue was somehow settled in, oh, so pre-last-century...
I fail to see how what happened is a case of free-speech. Asking for decency during one particular type of broadcast is not the same as supressing free speech or censorship.
The classic example of possible cause for supressing free speech is "shouting 'Fire!' in the full theater", which puts others in the situation of some "clear and eminent danger". PLEASE tell me what clear danger comes out of the broadcasting of the aforementioned boob of Ms. Jackson?
If you can not, a bonus question for you: How "one particular type of broadcast" is different from *THAT* other one?;-) Paul B.
P.S. I can understand (thgough not necessarily agree with the existance) of a Gov't body impartially providing the applicants licenses on a 'first come, first served" basis, but the amount of the discussion of J.J.'s tit in this context makes me wonder if it is the/, I am reading...
Slashdot Mirror
No, really... First time in the history of mankind... ;-)
Paul B.
Also, you don't need to round trip one signal at a time, you can have multiple signals in flight. Which makes this whole discussion rather moot.
;-)
Nope! Having several bits flying over the wire is a form of pipelineing (check-out "wave pipelining"), which increases bandwidth, but does not do anything to decrease latency. bringing us back to my original point, which can be illustrated by a fact that good weight-lifters rarely make good boxers...
Paul B.
Yeah, also called PIM, for "processing in memory". Peter Kogge (of Kogge-Stone fast adder fame) is doing research into those in Notre Dame (disclaimer: I used to be affiliated with the same HTMT crowd designing ultra-fast superconductor processing elements for the petaflop computer, thus I know first hand how hard it is to match memory speed to pipelined processing speed... ;-) ).
;-) ) and power (CV^2/2).
Bringing processors and (large) memories closer to each other does not help much, as, as you mentioned, there is an order of magnitude difference between processor clock speed and memory access speed. The physical reason for this is that to do a certain operation on one pipeline stage in a processor you need to charge a clock line passing through a couple dozen to couple hundred gates; in memory case you have to charge the word line passing through sqrt(1G)=30,000 gates. It takes time (RC, unless one uses superconductors and forgets about R
The only rerasonable solution is, indeed, to make memory blocks smaller and closer to processor elements, making them essentially registers/caches, not RAM.
Oh, and, BTW, in the rather naive picture on the link you sent, the solution will not work that well if you have multiple processors -- you have to make sure that each can talk to other's memories (in SMP case) AND to each other.
Paul B.
But then, the trick is that he did not mention memory latency, only bandwidth! Getting the latter is relatively easy -- just make memory bus wider (as given bus speed), trying to decrease latency will pretty soon make you run into speed-of-light limitation.
;-)
Maybe those processors do have enough memory bandwidth to load two of them completely doing SAXPY? Assuming 12 GFLOPS sustained (3 GHz, 2 cores, separate ADD and MUL on each) you need to feed input vectors at 12*8 bytes/double = 96 GB/sec, for, say 1 GHz memory bus it is translated into 96*8=768 memory pins only for input -- well, wider than I've seen on desktop PCs...
When you start doing anything else , the roundtrip time between processors and memory (latency) becomes more important than raw bandwidth.
Paul B.
Well, I would not necessarily agree... People who buy into Lindows do not (usually) download FC2, and they would not get a RedHat boxed set, so no lost sale to RH. OTOH, after they figure out that "this Linux thing" is not as scary and quite useable they might decide to try more advanced distributions and they will have a chance to do this immediately.
Paul B.
Oh, well, I can appreciate the full meaning of your ~/.signature now! ;-)
;-( I know how it feels, a good friend of mine does similar kind of stuff for (German! European bureocracy!) medical insurance databases, the stories she told me are unbelievable to anyone with more mainstream CS background.
;-) ) code in the kernel. Not that it is a guaranteed win, but at least deserves some consideration and thought, but the thought process requires of the guy doing it to know what the tradeoffs are, and you get this knowledge only from learning something besides Java.
The lesson: Test on production data.
Well, maybe another lesson is that Java is not exactly the right language when you are mostly dealing with syscalls (lseek(), read(), printf some SQL and send it towards the DB), which kind of brings us back to the original topic...
At least the developers didn't try to mmap the archive. Maybe they should've? Then some part of what they were trying to do would shift from Java to highly-optimized (maybe even in asm
The real problem is the massive, monolithic archive... A kernel module to mount the archive as a filesystem, anyone? I know, I know, but is not it sad that manufacturing gave us 3GHz processors and 300 GB disks (yeah, they just "print" those, not thinking of each individual Hz or byte) and too many programmers do not stop and think what does it take to handle a 100 GB dataset...
Paul B.
Of course you know your application better than anyone here, and I bet that you did do the right thing, but somethings strikes me as a little bit odd:
Are you sure we are talking about the same 'threads', as in pthreads/clone()-based, NOT fork()ed-out _processes_? By definition the first kind all share memory space and running 10 of them instead of one should not cause excessive memory consumption/swapping.
Well, of course it is not always true, if you mmap() a 100 GB database file and unleash a whole bunch of threads to work on different (non-overlapping) parts of that you start thrashing your OS's buffer cache really soon and can bring your system to a halt rather effectively...
As a side note, I always assumed that the only reason to use more threads (in your program) than there are CPUs/contexts available is when your problem is I/O bound (you achieve the benefits of asynchronous I/O without explicitly coding it in): number of threads = number of CPUs + number which on average sleeps waiting for I/O to complete.
Paul B.
Ah, filesharing in the good old days... they weren't really any better, but we were a lot younger :)
;-)
;-) Actually the last time a BSV tried to bit me was a couple of months ago (!) when I picked up a floppy from the ground on a street corner in some relatively bad area of LA (do not ask me what brought me there, some cultural/artistic thing my GF participated in ;-) ), was just curious what was on it... Well, among other things a BSV, but fortunately it had no idea how to infect a Linux box, but I had to forget the floppy in the drive to find out... ;-)
Yeah, yeah...
Does anyone else still set boot order to C: A: to prevent the accidental loading of an undetected BSV?
As a matter of fact, YES!
Paul B.
You dound like someone working hard towards the degree and sincerely being afraid that you will not be given that if your girlfriend finds out about that other girl, or something... ;-)
;-) -- they are generally "better" than regular ones, but still have more of a symbolic status...)
In The Real Life (TM) degrees are needed to a) make yourself/your parents proud and b) get employment in certain places/positions which do require this IF they like you otherwise.
Do you think anyone will be hiring THE guy any time soon on a position like this (even if he would formally have his degree)?
It is more of a symbolic jesture, really, and if you do that bad in scientific community you quite deserve this. (As an example of the opposite trend, think about the Honorary Degrees -- I guess Linus has one of those
Paul B.
... viruses used to live in the boot sector of a 5" floppy disk, while keeping the whole bootloader there as well, AND in the worst cases there were a handful of those on a single disk (bootsector) ;-) I can understand the fun of writing something like this back then, but the "Hey, dude, click here for your document..." is just plain silly... ;-)
Paul B.
Maybe because in the past decade people (ones with $$, as in "VCs" ;-) ) were interested more in running data around those, you know, strings which they called "Internet"? ;-) And now they have to run data in that thing which the old guy Maxwell discussed, you know, the "Waves"? And the latter one takes some extra effort in doing DSP? ;-)
On a more serious note, businesses want to differentiate themselves from each other (who needs another plain old stand-alone 2 megapixel camera if one can have a 5 MP one in their cellphone?), and plain Si is actually becoming cheaper (if you do not think about the top-of-the-line, 90nm one, of course, you do not need anything like that for your better-than-average consumer appliance).
And, of course, classic systolic arrays research has almost nothing to do with modern FPGAs, we will not see it re-emerging until speed-of-light signal propagation delays on chip become dominant (I have something to say about that too, but will not now...).
Paul B.
I see that we do agree on the most important points, just a bit of nitpicking on my side! ;-)
;-)
;-) Any time I hear someone say "middleware", I feel the urge to run screaming. As you could tell, I used much less respectful word for the thing... ;-) I am happy I do not have to actually work on the marketplace when the intimate knowledge of the middleware flavour of the day is required (though I would not mind being paid like those guys ;-) ).
;-)
;-) OTOH, the GnomeCanvas low-level implementation deserves to be coded not even in assembly, but in different GPU's assemblies, to bring us back to the original point! ;-)
"... wide majority of modern applications." -- yeah, the _modern_ part is what makes me kind of nervous... Is it me getting old, or, as I said to my friend some years ago, "You do not get much out of Java which you can not get from a g++ with an added flag to disable dereference operator"?
My KPilot is bored until I press the HotSync button on my Palm. Hey! This is what HLT instruction is for! Save some electrones until that hardware interrupt arrives!
Do not get me started on that topic - we just don't have the time to go there.
Your story about Python (and above-mentioned GTK) was exactly what I would expect from a reasonable educated guy, really, especially since I pretty much hold the same views. One of the coolest systems I ever had to work with was Cadence (big EDA CAD system), most of interesting parts written in their Scheme dialect. Autocad, BTW, too...
I even enterntaned for a while an idea of sitting down and writing a chip layout editor in Python and GTK (around GnomeCanvas), except that the Canvas did not keep the CTA matrices with the object groups (think, cells) and I could not have the same cell on the screen in, say, different orientations.
Anyway, good luck!
Paul B.
Mastery of assembly is completely pointless for anyone outside of OS kernel, compiler construcution and embedded development...which probably means you.
.com collapse).
And why do you think that in a year or two the Si wafer real estate (or, more realistically FPGA real estate, which it is probably right now) will NOT become cheap enough that one who thinks of himself as a 'programmer' now would have to actually write VHDL for the processor part, then assembly glue, and only then port Linux to the "ultra-optimized to track your eye movements while simultaneously scanning your brain waves for the tune you'd like to listen to now" car-dashboard computer?
The line between hardware and software is disappearing, and producing way too many "programmers" who can only code Java is a sure way to the next economic disaster in this country (way worse than
Are you sure it is not happening?
Paul B.
The As I've said before, a good mechanic can tear down and reassemble an internal combustion engine, and that my argument is that a good programmer can do the same on an ASM level should the need arise.!
/. uid, after all... ;-)
but, OTOH, you do have a 4-digit
Due to the advancee in hardware capabilities Bloat is now excusable - and encouraged. It is cheaper to buy a big ass computer that can crank and hog bits than it is to pay somebody good money to nitpick over every bit of resource that their programs consume.
;-) ) SEVERAL computers? I may be mistaken, but I do not think that you still have the mentality of "The world's market in computers is probably 5 or 6 of them" (food is burning on the stove, no time to get the exact Watson's quote ;-) ).
;-)
What if you want to run your app on (God forbid!
Last time I checked the price difference between a $100 ARM-based unit and $5,000 G5 was still $4,900, so if you intend to run your program on a 1000 units you can as well pay $900,000 to a guy who can port from the latter to the former and keep the cool $4M.
Or if you think about selling your solution (HW+SW) to some millions of customers... (Wow!)
Paul B.
If I need to write a GUI app, I'll do it in Python with GTK or QT bindings.
;-) ).
;-) ) tend to produce code bloat.
Almost by definition any GUI app spends all its time waiting for the user to press a button (responce time in milliseconds range), so noone in their rigth mind would code one in assembly. Low-level (think BITBLT) parts of the GUI is coded in assembler or maybe even microcode (extreme cases
You being able to pick an efficient library and sensible approach to the task at hand in no way invalidates the granny-post's point that businesses relying too much on levels and levels of crud (bought for big $$ from the companies employing people who have not learned assembly in college, not to mention high school
Paul B.
Where are my mod points when I need 'em... It is actually FUNNY (+1)
Paul B.
Well, this is the second part of the equation... To my mind the premise of the original article was not exactly that everyone should code in assembly (and reading "assembly language" three times in one 4-line paragraph was driving me crasy, he should've used ASM or something ;-) ), but that one should LEARN different things...
;-) ) should more or less know both LISP and assembly, the latter one for the reasons discussed in the arcticle, the former one if only because it kind of relates to what can be computed AT ALL, _not_ how efficiently it can be done. (see, for example, chapter three )I guess) of today's story about \Omega ;-) ).
To my mind, any programmer (I would not refer to the ivory towers of CS here
Paul B.
They do dumb things like multithread something that is CPU bound...
;-) ), but, I'd guess, almost a half of the people thinking in multithreading terms in this world (definitely half the total IQ, not to offend anyone ;-) ) use it to solve CPU-bounded problems on SMP boxes/multithreaded CPUs (Tera SMT; Intel's Hyperthreading on massive (ASCI) scale, things like that...)
;-)
Not that I disagree with your point (actually I do agree with it alot!
When you just start throwing processors (or thread contexts) on the problem, you will find soon enough that any problem is I/O bound...
Paul B.
I'm all for free speech, but the freedom of a group or indivduals aren't absolute...
Of course it's not, it can be argued if a freedom of an individual is absolute or not (more often reasonable people agree that it still shouldn't infringe or in anyway threaten the freedoms of others, but there is no such thing as a "freedom of a group", only of the individuals in this group...
Just nitpicking.
Paul B.
... which used to be Tera, Inc. they are in Seattle as well, I am wondering if THAT connection is more important than the MS one. Not that Burton Smith (CTO of Tera, err., Cray, and overall VERY cool guy -- I used to work with him on a project) and Thomas Sterling ("farther of Beowulf", really! ;-) ) work on the next-gen supercomputer (see, e.g.,
here, I am wonderng what the implications are...
Paul B.
Basically, the market can decide if Playboy Radio on XM is something they want to support or not... but nobody is forced to listen to that, you can't even accidently tune it on an XM device unless you're paying a monthly fee and then an extra monthly fee for that one channel.
What if someone (out of the kindness of their hearts!) would decide to give away XM radio receivers to anyone asking? No, I do not mean anyone associated with the apparently indecent ones on this mentioned radio program, just _someone_? Will your argument still hold? Can you still give your old 386 to a kid knowing that it can be used for indecent purposes?
And, after all, where is the boundary between "public" and "private" in something neither of us can hold in our hands or show to others and say "this is mine!"?
Paul B.
.. to get this reaction from /.?
Or from the democats, for that matter?...
Paul B.
... having my mod points but rather decided to participate in the discussion...
A *big guy* with a *big antenna* is not too much different from equally *big* guy with almost as big a truck plowing over your land, but that issue was somehow settled in, oh, so pre-last-century...
Paul B.
I fail to see how what happened is a case of free-speech. Asking for decency during one particular type of broadcast is not the same as supressing free speech or censorship.
;-)
/, I am reading...
The classic example of possible cause for supressing free speech is "shouting 'Fire!' in the full theater", which puts others in the situation of some "clear and eminent danger". PLEASE tell me what clear danger comes out of the broadcasting of the aforementioned boob of Ms. Jackson?
If you can not, a bonus question for you: How "one particular type of broadcast" is different from *THAT* other one?
Paul B.
P.S. I can understand (thgough not necessarily agree with the existance) of a Gov't body impartially providing the applicants licenses on a 'first come, first served" basis, but the amount of the discussion of J.J.'s tit in this context makes me wonder if it is the