I knew of a doctor that used to spend a lot of time in a decompression chamber (working with BENDS patients and such). He would run his BBS from there, and I'd call in occasionally. At least he found a way to pass the time.
In particular you might try patch-2.4.18-pre9-mjc2.bz2 , which include O(1) scheduler, preempt, and Rik's Rmap vm (among other things), and has been working solid for a number of people. At least, it is worth testing out to see if it helps any.
To build it, get the
linux-2.4.17.tar.gz kernel,
patch it to linux-2.4.18-pre9, then patch again with patch-2.4.18-pre9-mjc2. Then build and use the kernel. Check recent (ie. 2002 ) kernel archives to read discussion of this and other related patches, if desired.
Hmmm, well I got that fact wrong; I seem to remember it being implied that he brought the Bible back. Were there hints as to the other books. I am beginning to worry that I may be wrong in my earlier assessment, so perhaps I'll have to watch it again (someday).
The Japanese military alone murdered *far* more civilians (predominantly in China and south-east Asia), in ways that are at least as gruesome and calculated as Hiroshima and Nagasaki. To say the bombing was one of the worst single acts in human history, is only to say that WWII itself was one of the worst times in human history (hardly an arguable statement).
The vast majority of deaths in that war were non-combatants... Nearly 100,000,000 total deaths by some estimates. If anything, the bombs took attention from the horrific attrocities which the Japanese military and government perpetrated, and which they have never even had to officially acknowledge.
When I was learning C, years ago, I came across these contest entries (this was probably around 1989). I learned a LOT (!!) from these pieces of code. Many of them are staggering in their cleverness and produced many "Aha!" moments when I thought about them. If I were teaching a C course, I would definitely use these as examples (later in the course).
I remember Leo Schwab (ewhac on/.) describing how Kim Sachs used to work; If Jim wanted to change the color of something, he would manually repaint each pixel. Then, if he wanted to change it back, he would repaint each pixel again. He wouldn't just do a simple palette change.
Now that may sound foolish, but if you saw Jim's work you'd see that he was very adept at blending and texturing with colors, and using dithering techniques, and so by changing the color of a few pixels, he would then repaint the surrounding pixels so that it looked just right. A simple palette change would not have had the same effect, because each pixel color was predicated on its neighbor. In short, it took a LOT of patience and mouse clicks (in fact, he used to talk about how many mice he wore out from clicking)
I just graduated from a Stanford graduate program last year; I noticed a fair amount of cheating (usually students "grouping" together to do take home exams or do projects, as well as the occasional wandering eye during test taking). This school also had an Honor system, and many of the students were top caliber. But being almost the best was apparently not good enough; some students desired to get the TOP score, and cheated in spite of the honor system.
I did it the hard way, and was nowhere near the top of my class. But I daresay I am much more self assured as a result; and when I interview a straight-A drone for a job, I know to really ask specifics about what they did in school, and how they did it. Also, I check for imagination and free thinking. That seems to help expose the "overacheaters". (Granted most students probably do not cheat, but I wouldn't be surprised if the amount of graduating cheaters is between 20-30%; a shamefully high amount, if true)
However, in my mind at least, the reputation of Stanford (and by extension, other top-tier schools) has dropped because I have seen how prevalent cheating can be, and when I hire one person, without their six buddies to help out, I'm not assured I'll be getting full value.
You would trust the site that keeps your password stored on the site in plaintext, and had a break-in which compromised all the passwords? You are more trusting than I...
Are these benchmark comaprisons for P4 systems each with equal configurations (other than one running DDR and one running RDRAM)? If so, it may be indicative of poor DDR support in P4 chipsets; in high end systems, one could use multi-channel DDR (installed in pairs, for example) for higher bandwidth, although I don't know that any motherboards support this.
In any case, without more specifics, it is hard to know whether those results are a reliable basis for comparison. But, your point is well taken; matrix math can really make use of memory throughput, and is not as sensitive to latency issues, depending on the problem and its implementation.
But, I think DDR can deliver equal or higher throughput (at least for a while), if chipsets and motherboards were designed for this.
You're not taking into account that more modern chips use smaller dies, and less voltage, thus producing less overall heat output for the same MHZ. However, smaller die means the heat that is there may be more concentrated.
So in general, it may be possible to underclock the.13 micron die chips and have them run without a fan. It is simply a matter of how much to underclock, and whether the chip and motherboard support it (and whether your heatsink is adequate).
Because vendors want features, FEATURES, *FEATURES*! They dump in XFS, low-latency, JFS, ALSA, etc. etc. which is fine if THEY are willing to support it. The mainstream kernel would rather wait to accept patches when they have a clearer idea of how maintainable things will be in the long term. Eventually, many or most of the popular and useful patches do get in. Linus believes very much in the vendor model; it allows him to focus more on future development rather than current featuritis.
Which is actually a good thing. Now patching from 2.4.18 is not predicated on how you got to 2.4.18 (although those running 2.4.18-rc4 have to patch to 2.4.18 proper, before patching to 2.4.19).
Depending on luminance levels, contrast, etc. the eye can perceive at much higher rates than 40Hz. Film appears "okay" at 24fps because the film itself has motion blur, and because most people are used to it (and filmakers work within it's limitations). But I work in a vision research lab where we have a 240Hz monitor (120 Hz for each eye with high-speed shutter stereo). For some things, it does matter.
Just out of curiousity, I'm building two new Athlon systems, basically identical. However, I was planning to underclock one of them to reduce energy use and heat production. Can this only be done by changing the FSB? Or can the processor multiplier also be lowered? I assume a multiplier locked processor means locked from going both up AND down? (The chips are Athlon XP 1700+ btw, which I assume are multiplier locked)
I assume Apple (like myself) is waiting for deployment of serial ATA technology; this will get around the current size restrictions as well as offer other improvements. I had hoped it would be available by now, but it seems it will be another year or so before it is even targeted for high end consumer level products.
See my other post, but I actually am agreeing with you in part. "digital" refers to the nature of the data (whether storage or processing, etc.), as a collection or sequence of digits, however encoded. But the underlying encoding doesn't need to be discrete, and even if it is, that doesn't imply digital. (ie. what you are calling "discrete values", I am calling "digits", I think) Anyway, I think we are in agreement, although I may be playing a little loose with terminology (I'm a software guy:)
The distances are not fixed; the signal is pulse width modulated, and is read back and filtered to produce an NTSC signal (simplified, but basically accurate). It is never treated as a sequence of DIGITS, and thus is not digital. When these things were designed, it just wasn't feasible to make it digital . But that illustrates my point; I explained to my friend that the pits and valleys were discrete (ie. two distinct levels), but that didn't make the system as a whole digital. Oh well; I hadn't realized it either until I studied a little bit of signal theory, and learned that laserdisc video didn't work like CD audio.
It reminds me of another time when a different group of aquaintances (with computer degrees) couldn't understand how to receive digital radio over their TV cable (using the DMX box, or whatever it was called.) The argument was "But the wire is analog!":) In that same group, an electrical engineer didn't understand how radio signals propagated through the air (he claimed it was "an unexplained phenomenon"; and he wasn't speaking in abstract terms, he didn't even understand that there are at least MODELS of how the phenomenon must work (photons, excitement, etc.)) I've since stopped hanging out with those folks; it made my college degree seem so much less valuable.
I'll go one step further and say that even if there are discrete levels, it is still not necessarily digital. Digital implies processing as digits (usually groups of discrete levels), but discrete levels, or modulated square wave frequencies, etc. are all analog by default.
I had a discussion of this way back when with an otherwise very bright computer guy, who just couldn't understand that laserdiscs (pre-DVD; the big ones that movies came on) were NOT digitally encoded. He thought that the discrete nature of the encoding (pits and valleys) meant it must be.
Slashdot Mirror
"If any of you [in the audience] are in Marketing, or Advertising... kill yourselves."
I pay sales tax on used CD's or books at a store, unless I am mistaken (and I may be; perhaps I never payed attention).
I knew of a doctor that used to spend a lot of time in a decompression chamber (working with BENDS patients and such). He would run his BBS from there, and I'd call in occasionally. At least he found a way to pass the time.
To build it, get the linux-2.4.17.tar.gz kernel, patch it to linux-2.4.18-pre9, then patch again with patch-2.4.18-pre9-mjc2. Then build and use the kernel. Check recent (ie. 2002 ) kernel archives to read discussion of this and other related patches, if desired.
Hmmm, well I got that fact wrong; I seem to remember it being implied that he brought the Bible back. Were there hints as to the other books. I am beginning to worry that I may be wrong in my earlier assessment, so perhaps I'll have to watch it again (someday).
The Japanese military alone murdered *far* more civilians (predominantly in China and south-east Asia), in ways that are at least as gruesome and calculated as Hiroshima and Nagasaki. To say the bombing was one of the worst single acts in human history, is only to say that WWII itself was one of the worst times in human history (hardly an arguable statement).
The vast majority of deaths in that war were non-combatants... Nearly 100,000,000 total deaths by some estimates. If anything, the bombs took attention from the horrific attrocities which the Japanese military and government perpetrated, and which they have never even had to officially acknowledge.
I hated the end of that movie; he only takes *one* book back to the future (hmmm...). Can you guess which one he took?
When I was learning C, years ago, I came across these contest entries (this was probably around 1989). I learned a LOT (!!) from these pieces of code. Many of them are staggering in their cleverness and produced many "Aha!" moments when I thought about them. If I were teaching a C course, I would definitely use these as examples (later in the course).
I remember Leo Schwab (ewhac on /.) describing how Kim Sachs used to work; If Jim wanted to change the color of something, he would manually repaint each pixel. Then, if he wanted to change it back, he would repaint each pixel again. He wouldn't just do a simple palette change.
Now that may sound foolish, but if you saw Jim's work you'd see that he was very adept at blending and texturing with colors, and using dithering techniques, and so by changing the color of a few pixels, he would then repaint the surrounding pixels so that it looked just right. A simple palette change would not have had the same effect, because each pixel color was predicated on its neighbor. In short, it took a LOT of patience and mouse clicks (in fact, he used to talk about how many mice he wore out from clicking)
You spelled "innovation" wrong. Although given that it was a GWB quote, perhaps you were quoting him accurately.
I just graduated from a Stanford graduate program last year; I noticed a fair amount of cheating (usually students "grouping" together to do take home exams or do projects, as well as the occasional wandering eye during test taking). This school also had an Honor system, and many of the students were top caliber. But being almost the best was apparently not good enough; some students desired to get the TOP score, and cheated in spite of the honor system.
I did it the hard way, and was nowhere near the top of my class. But I daresay I am much more self assured as a result; and when I interview a straight-A drone for a job, I know to really ask specifics about what they did in school, and how they did it. Also, I check for imagination and free thinking. That seems to help expose the "overacheaters". (Granted most students probably do not cheat, but I wouldn't be surprised if the amount of graduating cheaters is between 20-30%; a shamefully high amount, if true)
However, in my mind at least, the reputation of Stanford (and by extension, other top-tier schools) has dropped because I have seen how prevalent cheating can be, and when I hire one person, without their six buddies to help out, I'm not assured I'll be getting full value.
You would trust the site that keeps your password stored on the site in plaintext, and had a break-in which compromised all the passwords? You are more trusting than I...
Are these benchmark comaprisons for P4 systems each with equal configurations (other than one running DDR and one running RDRAM)? If so, it may be indicative of poor DDR support in P4 chipsets; in high end systems, one could use multi-channel DDR (installed in pairs, for example) for higher bandwidth, although I don't know that any motherboards support this.
In any case, without more specifics, it is hard to know whether those results are a reliable basis for comparison. But, your point is well taken; matrix math can really make use of memory throughput, and is not as sensitive to latency issues, depending on the problem and its implementation.
But, I think DDR can deliver equal or higher throughput (at least for a while), if chipsets and motherboards were designed for this.
You're not taking into account that more modern chips use smaller dies, and less voltage, thus producing less overall heat output for the same MHZ. However, smaller die means the heat that is there may be more concentrated.
.13 micron die chips and have them run without a fan. It is simply a matter of how much to underclock, and whether the chip and motherboard support it (and whether your heatsink is adequate).
So in general, it may be possible to underclock the
Because vendors want features, FEATURES, *FEATURES*! They dump in XFS, low-latency, JFS, ALSA, etc. etc. which is fine if THEY are willing to support it. The mainstream kernel would rather wait to accept patches when they have a clearer idea of how maintainable things will be in the long term. Eventually, many or most of the popular and useful patches do get in. Linus believes very much in the vendor model; it allows him to focus more on future development rather than current featuritis.
Which is actually a good thing. Now patching from 2.4.18 is not predicated on how you got to 2.4.18 (although those running 2.4.18-rc4 have to patch to 2.4.18 proper, before patching to 2.4.19).
Depending on luminance levels, contrast, etc. the eye can perceive at much higher rates than 40Hz. Film appears "okay" at 24fps because the film itself has motion blur, and because most people are used to it (and filmakers work within it's limitations). But I work in a vision research lab where we have a 240Hz monitor (120 Hz for each eye with high-speed shutter stereo). For some things, it does matter.
Just out of curiousity, I'm building two new Athlon systems, basically identical. However, I was planning to underclock one of them to reduce energy use and heat production. Can this only be done by changing the FSB? Or can the processor multiplier also be lowered? I assume a multiplier locked processor means locked from going both up AND down? (The chips are Athlon XP 1700+ btw, which I assume are multiplier locked)
Thanks for any info.
Texas A&M (and I'm not trolling...)
I assume Apple (like myself) is waiting for deployment of serial ATA technology; this will get around the current size restrictions as well as offer other improvements. I had hoped it would be available by now, but it seems it will be another year or so before it is even targeted for high end consumer level products.
See my other post, but I actually am agreeing with you in part. "digital" refers to the nature of the data (whether storage or processing, etc.), as a collection or sequence of digits, however encoded. But the underlying encoding doesn't need to be discrete, and even if it is, that doesn't imply digital. (ie. what you are calling "discrete values", I am calling "digits", I think) Anyway, I think we are in agreement, although I may be playing a little loose with terminology (I'm a software guy :)
The distances are not fixed; the signal is pulse width modulated, and is read back and filtered to produce an NTSC signal (simplified, but basically accurate). It is never treated as a sequence of DIGITS, and thus is not digital. When these things were designed, it just wasn't feasible to make it digital . But that illustrates my point; I explained to my friend that the pits and valleys were discrete (ie. two distinct levels), but that didn't make the system as a whole digital. Oh well; I hadn't realized it either until I studied a little bit of signal theory, and learned that laserdisc video didn't work like CD audio.
:) In that same group, an electrical engineer didn't understand how radio signals propagated through the air (he claimed it was "an unexplained phenomenon"; and he wasn't speaking in abstract terms, he didn't even understand that there are at least MODELS of how the phenomenon must work (photons, excitement, etc.)) I've since stopped hanging out with those folks; it made my college degree seem so much less valuable.
It reminds me of another time when a different group of aquaintances (with computer degrees) couldn't understand how to receive digital radio over their TV cable (using the DMX box, or whatever it was called.) The argument was "But the wire is analog!"
I'll go one step further and say that even if there are discrete levels, it is still not necessarily digital. Digital implies processing as digits (usually groups of discrete levels), but discrete levels, or modulated square wave frequencies, etc. are all analog by default.
I had a discussion of this way back when with an otherwise very bright computer guy, who just couldn't understand that laserdiscs (pre-DVD; the big ones that movies came on) were NOT digitally encoded. He thought that the discrete nature of the encoding (pits and valleys) meant it must be.
I read that as "how fed up the world is", not "how fucked up the world is." Both interpretations seem correct to me.
Stanislaus Lem would probably disagree (if he weren't dead).