Under some workloads, I can go along with the assertion that a MOSIX cluster is just like having a big machine with a lot of CPU's. It seems to be great for those workloads and I would love to try it out. Those loads tend to be multiple long running (more than a few seconds) and not multithreaded. For MOSIX to be most efficient, there also needs to be fewer jobs than there are CPUs to run them.
Other workloads, however, will not benefit from MOSIX. These statements are based on reading the docs a couple weeks back, not on actual experience.
Speaking from experience, you are pretty much correct. Jobs that use lots of CPU, but have little IO are good for mosix clusters, but jobs that have high IO are bad. The mosix filesystem and other things can partly get around the IO problems if the users plan carefully, but mostly they just want to start 30 jobs and forget about it for a few days.
There is no reason that a mosix cluster can't be combined with a batch/queueing system. This lets lazy/stupid users run their CPU bound jobs and lets mosix distribute them, but more savy users can script their IO jobs to run on particular machines and use local disk for IO.
It took a few months for the users of the cluster I setup to get trained into what jobs work well, and which kill the cluster. The problem is that launching 40 "good" jobs on a single machine is not a problem, because they just shoot out to the other nodes, but launching 40 "bad" jobs on a single machine will make that machine almost unusable.
This can have adverse effects on the cluster if the good jobs were started from the overloaded machine; for example the good jobs might have to check back with their originating machine every few minutes to update a checkpoint file.
Basically, mosix isn't some magic bullet to solve machine limitations, but it is a very cheap and effective way to solve certain problems.
Thanks, that answered many of my questions. I can certainly see how somebody could have committeed Crusade to death. I can just see the spec sheet now: (semi-)attractive female alien character, (semi-)evil corporate guy, potentially-cool but way overdone guy, etc., the Vorlon ambassador guy with the mysterious past was popular, so every character should have a mysterious past!
The characters in B5 started out a bit 2-dimensional, but I think that was on purpose. As the motives for all of the characters became clear they took on much more depth, and of course the characters changed in response to the changing situtations, etc.
That reminds me, I still have one unwatched episode of Crusade left on my Tivo, from sometime last year, that I still haven't watched. I saw it the first time around, but for as amazing as (the first four seasons of) B5 was, Crusade mostly sucked.
I think part of the problem with the fifth season of B5, Crusade, and to some degree Legend of the Rangers, was kind of a too-much-of-a-good-thing syndrome. For example, in B5, [what's her name, the woman commander] gives a nice speech before attacking the Earth fleet, about being death incarnate and all of that. The speech was a bit cliche if taken out of context, but for the situation and the character it worked well. Mostly because through the show, she only gave a few speeches like that, maybe one or two a season. When they happened they were pretty cool, but then there are characters on Crusade that only talk in fancy speeches. It's too much, they lose all impact and end up just being cheesy Shatneresque spewing.
I am sure JMS still has it in him to make a good show, but I think he really needs to go back and watch the third and fourth seasons of B5, and maybe In the Beginning, and see what the balance was that really made them work. Don't just take the best bits and try to only use those in a show.
I am still running two of these (533mhz 164SX) machines as general login and computing servers. For some reason people really like to use xdm on these old alphas from their Windows boxes. I even setup a nice dual processor Intel machine with loads of memory running Debian and the latest gnome and kde, but nobody seems interested in using that.
The old turbochannel alphas had some pretty serious reliability problems (a 90 day warranty on a $7000 computer!?) I had most of the DEC components (i.e., not 3rd party stuff, like disks) on my two turbochannel alphas replaced several times under maintenaince before getting to board revisions that could last more than 6 months.
However, the PCI based alphas I have seem to be totally bullet proof. I think in the whole time I have been running them, once lost a disk, which one can hardly blame DEC/Compaq for.
For integer stuff the 164SX machines are bit slow, probably comparable to a 350mhz PII, but for floating point, they are probably better than a 700mhz PIII (though I haven't benchmarked these thngs in years, so I may be remembering wrong). Of course they don't compare in any way to a $50 1Ghz Duron.
Some of the early turbochannel alphas, had badges saying DECstation on the front. The Personel DECStation 25 (a 25mhz mips R3000) and the Alpha 300 series even shared the same box. I think it is an Alpha 400 series maybe, that I am recalling with that label. I don't think the Alphastation and Alphaserver names were used until the PCI based systems.
All of my old DECstations (mips and alpha) are sitting in a storage room because it is too much bureaucracy to throw them away, but they just aren't worth pulling out and playing with.
I am probably getting some of the names confused, but this is definately a problem that Digital (aka DEC) perpetuated with their constant renaming of stuff. Talk about a company being run into the ground by poor marketing. Since I have been using it, their Unix OS has been called OSF/1, Digital Unix, and Tru64. At some point in the mid-90s, DEC decided that years of name recognition and reputation under the name DEC was too much, so they wanted everybody to call them Digital.
I'll start by saying that I find 2.4 to be very stable, and to perform mostly ok on 1 and 2 way machines. My laptop, desktop, and 2-way server stay up until I decide to reboot them. Actually, I brought my 2-way server down for a disk upgrade today for the first time since early November when I installed 2.4.14.
Having said that, there are some serious issues with 2.4 on some 8-way 8GB machines that I manage. They have been running 2.4.13-ac7 since November, because that is the last kernel that is usable for me (-ac11 would probably be ok). Newer kernels have terrible behavior under the intense IO load these machines go through. They get 14-30 days of uptime, and then hang or get resource starved or something and have to be rebooted.
I think part of the issue is that there simply aren't that many people running 8-way boxes, so bugs aren't found as easy, this is of course on top of having 8-way SMP being much more complex than a defacto single user, single processor desktop machine. To make it even worse, the machines are pushed hard. They move around GBs of data every day, and often will run for extended periods with loads over 25.
Of course, it is still mostly ok. While the machines are working they mostly work fine. Of course 20 days of uptime is totally unacceptable. I have an alpha running Tru64 pushing 300 days of uptime, and the last time it was down was due to a drive failure, not an OS problem.
My only remaining issue with Linux on "small" machines is an oscillation problem in IO. Data will fill up all available memory before being written to disk, and then everything from memory will be written out, and then memory fills up again before anything new is written to disk. This is a bit inefficient, and the machine's responsiveness at the memory-full part of the cycle is poor.
What are my options though? I guess I could try FreeBSD, but a bit of lurking on their lists and forums reveals plenty of problems there, too. Do I switch and hope things get better, or wait out 2.4 and hope it comes around soon? Aside from a few nasty bugs in some releases, pretty much each successive 2.4 kernel has been better than the previous one, at least on small systems.
Several years ago I was having a hard lockup problem with Tru64 (Digital Unix, at the time) and that was very scary. It took time to get the problem escalated to the OS engineers, instead of just sending an e-mail to lkm. Even then I could only hope that the issue was being addressed, but I had no way to know if anybody was doing anything about it or not. (Turned out to be an bug in the NFS server that would cause the machine to lockup when serving to AIX.) For all of its problems though, it is extremely reassuring for me to be able to monitor the development process of Linux through the linux-kernel-mailing list, and other specialized lists. If I feel that people aren't aware of some problem I am experiencing, I can raise the issue. I am not in the dark about what is happening, and what fixes are being made. I know what changes have gone into each kernel update, so I know if there is a chance of it fixing my problems.
Not the Wellcome Trust Centre for Human Genetics
on
3rd Chromosome Deciphered
·
· Score: 3, Informative
Once again, a bit shortsighted, because I can't believe this is a design limitation. The Treo comes in two models 900/1900 (US) and 900/1800 (Europe/Asia). There are 900/1800/1900 tri-band phones out there for much cheaper than $400...
I can mostly deal with it being a GSM only device, and not have TDMA or CDMA, but calling it a dual-band world phone is pretty much an outright lie. Yeah, its a "world phone" if you never travel across oceans.
I still might get one if the service plans are good, but being able to go to the UK and stick a £10 Virgin pay as you go sim card in it would make the thing a true winner.
we came in and said there should be a platform that's identical with millions and millions of machines, and the bios of that should be open to everybody to use
I kind of see what BG is saying here, that the free software movement couldn't have taken off without open, commodity hardware. This is wrong. The free software movement, as it came out of RMS and the MIT AI lab was a direct response to proprietary, closed hardware and software. The free software movement grew out of the tradition of open access to software and tools on very non-standard mini computer hardware.
In the early 80s when the FSF was founded, it was not clear yet that the IBM PC would be such a dominant force in the computing world. Commodity home machines aren't even mentioned in RMS's initial announcement. In fact, he is talking about replacing the system on very expensive, practically custom built machines, which were only found in universities and big businesses.
Sure, the pervasiveness of computing has been a major boost for free software, but this is a base rate issue (i.e., there are x free software users out of n*x computer users).
This is serious, and hopefully not too offtopic. I purchase lots of desktop machines, and for myself I can build exactly what I want from whatever pieces I think are best. However, for some random new post-doc I just want a decent machine that is easy for me to buy, and easy to get fixed if it breaks.
I have always bought Dells, because they make doing educational institution purchases incredibly easy, and if I need service I just call one place. I can customize the computers I am buying, and their prices are reasonable.
I am finding myself in the position of having to buy a very fast computer for somebody else.
The problem I am running into is that Dell does not sell Athlons. I can buy a 1.8Ghz P4 from them for about $1900 fully loaded. I can also build myself a dual Athlon 1.2Ghz for the same price, and the Athlon is much faster.
So my question, is there a reputable and reliable company which sells customizable Athlon machines for a reasonable price?
I am seeing these hits too. Since 18/Sep/2001:07:27:25 -0600 (it is now 09:16) I have been hit by 120 different machines. 105 of them are on my class B, 128.138, 14 more just start with 128, and only one is from a totally different address.
Perhaps I should contact the admins at my site who are in charge of the offending machines.
I used this manual for an Ericsson R320AT GSM phone. My phone certainly isn't a GSM phone, but the address book commands seem to work.
The commands also worked on my friend's Motorolla, which is a GSM phone. On that one, we needed to send an AT+CPBS="ME" to tell the phone to use the numbers stored in the phone, as opposed to those stored in the sim card.
BTW, I had to run unix2dos on the phonebooklist file in my previous message, because the phone expects lines to be terminated with a CRLF, instead of the Unix LF only.
I was looking around for something to allow me to manage the phone book in my Ericsson R280L. I tried Kandy, part of KDE, but it didn't work well enough to do what I wanted. I finally just converted my address book from my palm into text, copied the numbers I wanted into emacs and created a list that looks like:
I connected to the phone with kermit and made sure it was ready to talk, and then ran something like
foreach i (`cat phonenumberlist`)
echo $i >/dev/ttyS0
sleep 1
end
The sleep 1 is important, because the phone couldn't take entries any faster. A friends Motorolla could connect over IR, but it needed sleep 5 after each entry.
Managing my phone list with emacs, sed, awk, and sort turned out to be much simpler than doing it with some click happy windows program I tried. I think it was TrueSync, or something that I got from Yahoo to synch my palm with Yahoo calendar. It claimed to be able to sync to my phone, but it only had two modes of operation, do nothing, and erase all numbers in the phone.
Perhaps they just handed over the encrypted data to the NSA who promptly cracked it. Now, how do you use this in court without revealing that it was NSAs monster cracker that did all the work.
That actually isn't too outlandish. If you recall, during WWII the allies occasionally chose to let soldiers and civilians die rather than reveal that they could read the German's codes (and new where the next attack/bombing was coming).
When the allies had information from decrypted information that revealed the location of ships, they would always send a spotter plane over the ships before attacking to give the German's a plausible explanation for the allies knowing the ships' location.
Properly used and implemented PGP is uncrackable by brute force. Regardless of the size of the NSA's monster cracker they couldn't brute force PGP unless they have some secret knowledge. An as yet unknown (to us) flaw in PGP, or an advance in mathematics that allows for fast factoring of large numbers is something that I expect the NSA would readily kill people to keep secret.
As soon as knowledge of a secret like that got out, people would stop using PGP (or whatever), and the secret would become worthless.
I ran a connection over a dry loop for a few years when I lived in student family housing at my school. At the time, the only connectivity choice was a 9600baud ISN (serial) connection. So, for $3 a month or something the University's telecom group connected a dry pair between my apartment and my office on campus.
I used a pair of short haul modems to run a 38400 SLIP connection over the dry loop. This is nothing compared to the speed of DSL technology, but at the time $150 an end was about all I could afford, and DSL stuff wasn't available at a consumer level.
This setup worked great until bridge construction (the over a river type of bridge) caused the connection to be rerouted, and it never worked right again. A kind telecom employee took pity on me and I used an illicit second phone line to dial into my office for another year or so, until somebody noticed the connection on the switchboard. After that I was limited to dialing in on my main phone line and getting a cell phone for voice. (Paying the university for a second phone line was more expensive than getting a cell phone.)
Of course, now a few years later the apartments I lived in have 10/100 connections onto the University backbone.
BTW, if anybody is interested in buying some used short haul modems, let me know...
the weight of a good book in their hand, and honestly have some kind of tactile fixation with page turning.
I consider myself someone who reads lots of books, and I completely disagree with this statement. I think the people who say they like the feel of a book in there hand have never tried any type of e-reader. They weight of enough books to last me for a two week trip is not pleasant. I would much rather put a few books on my Palmpilot, which I have with anyways, than carry around an extra few pounds of paper.
I have been reading books on my Palmpilot for several years now, and I am completely addicted to it. I even have a Palm III with the old low contrast screen, so I would probably like it more if I moved to a V or 500 with a proper display.
I think people who don't like reading e-books have never tried it. (This is making the assumption that the books these people want to read are available in an usable format. I can completely understand people not wanting to read e-books because they have no interest in 100+ year old stuff from the Gutenberg project or whatever annoying thing the publishers have decided to make available.)
No, the hours are taken from people who have increased their recording time beyond a certain limit, whether by increasing the A or B drive size. A Tivo purchased with 60 hours is "really" a 72 hour Tivo with 12 hours reserved, so when 2.0 gets installed on a hacked 96 hour Tivo it notices that it isn't reserving 12 hours, so it grabs them and turns the Tivo into an 84 hour Tivo. The max number of hours it takes is 12, but the amount scales down with smaller sized units. If you haven't hacked your Tivo, it won't take any time from you.
They have not specified what the extra hours are for, but I am pretty convinced it is for some type of targetted ads. 12 hours could never be filled over the phone line, so it has to be something sucked down from the airwaves, which is either pay-per-view or targeted ads, and ads seems more realistic.
Tivo has been told, and I hope they understand, that the users won't mind targetted ads as long as they in no way impact our viewing, as soon as we are forced to watch ads then all "well behaved" hacking stops, and we all learn how to hack the guide data and cancel our subscriptions.
I have a Tivo, and I know they collect button press and viewing data on me, and I am fine with that (mostly because they told me up front they are doing it).
I was very amused when I read about how they interpreted the button press data: They think if I always fastforward through the commercials on the Simpsons (for example) then the commercials on the Simpsons must really suck, but if I rarely fastforward through the commercials on Friends, then those must be really good, well targeted commercials. In reality it is completely the opposite. If I am actually paying attention to a show (like the Simpsons) then I will always fastforward through the commercials, but if the show is nearly meaningless to me (like Friends) then I don't fastforward through the commercials because I am probably in the kitchen washing dishes or something where I can't even see the TV.
I wish the best to Tivo, and I hope they can pull huge amounts of money from the networks and advertisers for all of my "data", but I reserve the right to laugh at their interpretation.
I think the preceding post sums up the situtation better than any other I have seen. The tests are not perfect, of course they have flaws, but they are better than anything else out there.
The whole school system should be setup to push each student to the maximum of their potential, regardless of how high or low it is. Sucks if my kid isn't as smart as that other one, but don't stop teaching calculus in high school, just because some of the kids won't do well in it.
the connection between what the SAT actually tests and anything that actually matters, such as intelligence
Actually it is quite high. I don't have the statistics handy, but the SAT is basically an intelligence test.
The SAT is not even a particularly good predictor of good grades in college
This may be the case, but there is nothing better.
good grades in college is a pretty bad predictor of success in life
Success in life is a bit harder to define than good grades in college, but if "success" is measured as salary and responsibility, then there is not much better than grades in college to predict future success.
(re: all of this good and better business, just because something is the best, doesn't mean it is good, there is simply nothing that is closer to perfection. i.e. all operating systems suck, some just suck less than others.)
If we had a good test for intelligence
There are good tests for intelligence, people just don't like what they measure. Intelligence is one of the most reliable psychological characteristics that can be measured about a person. The correlation between a test given to an individual at one time point, and at a later time point is about 0.80, which is very high for psychological characteristics.
This is getting back to the point of the original post, it isn't that intelligence tests are bad, it is that people don't like what they say, which is that not all people are the same, some are smarter, some are much smarter, and there isn't much that anybody can do about it.
What I learned from doing very well on tests was that they didn't test anything that meant anything.
This doesn't matter. The test is designed to be a valid indicator of how well (as defined by grades) a person will do in college. If it asked about yak farming, because that was the best indicator of future performance, then that would be a valid question.
This all might seem silly, and no test can capture all of the complexities of an individual, but that is not the point. All of these personal, anecdotal exceptions to the relevancy of SATs don't matter.
You can disagree with the philosophy behind the following statements, but right now they reflect the situation in education as it exists: Currently colleges must do something to limit who can come in and who cannot. Colleges consider successful students to be those who go on to graduate from the college with high marks. The best indicator of how well somebody will do in college is a combination of their high school GPA and SAT or ACT score.
Now, as for the anecdotes, of course these statements don't apply exactly to you. This is where college admissions boards have an extremely difficult and possibly impossible job. They are trying to apply broad statistical principles to individuals, and many of them get it horribly wrong. I have seen lists of applicants with a (literal) cutoff line drawn at 1200 (or wherever). This is completely the wrong way to go about it. There is no valid reason that an 1190 with a 3.8 gpa is worse than a 1210 with a 3.7 gpa. The problem with the lines, is that somebody is always on the border. It is easy to see that a 900/2.1 might not be suitable for a particular institution, but what about the people between 1200 and 900?
We don't like the SAT test, since its results disagree with our vision of a fair and perfect world. Therefore, the test is defective and must be eliminated.
Yes, this is exactly it. These tests, and standards based admissions, and GPAs, and grades themselves all suggest that people are different. Because so many people take literally the statement "all men are created equal" they think that these tests are somehow evil. Of course what "all men are created equal" means, is that all people should be given equal opportunities. Everybody is allowed to go to high school, but some people blow it there (or, through lack of ability, fail to do well), so they are then not given equal access to college. This is how a merit based system (such as our educational process) is supposed to work.
I am not saying this is a perfect world, of course people are not given equal opportunities, which is the whole point of the civil rights movement. The SATs and other standardized tests do not judge equally across different minority groups, and the reasons for this are not understood. Is it possible to quantify hundreds of years of keeping a group down into a few bonus points on a test?
Many Americans seem to have a problem with the suggestion that some people are smarter than others, and that some people have a better shot at doing well in society than others. The logical conclusion to this thinking (because we can only make people smarter to a limited degree) is to make everybody dumber (which is much easier than making people smarter).
But the problem is that once you've done this, everyone who wants to do well on your test will start studying the things that are on the tests. Over time, the cumulative effect is that your corelation is distorted.
The people who study and work hard to perform better on the SATs will tend to get better scores on the SATs than those who slacked off. Guess which two skills are extremely important in relation to getting good grades in college? Studying and working hard. In general (always exceptions) people who study and work hard in college make better grades than those who slack off.
I have a list of words that have appeared on SAT tests in the past, and you have a different list. We take the same test, and just by chance 10% of my words were on the test, but 20% of yours, so you do a little better (because otherwise we are pretty equal in our abilities). This is called error, and anybody using the results of the test should take this into account. If we both took the test again (another version) maybe I would do a little better because this time my words are all over the place. The admission committees knows this, and won't (unless they are stupid) think that somebody who scored an 1150 is clearly a better student than somebody who scored an 1130, though they both will probably do better in school than somebody who scored a 1000.
There is also an article about FastShip in Scientific American which explains a bit about the hull design.
Of course the article is old (10/97?) and states that service between Philedelphia and Europe should start in 2000. I guess they are a bit behind their earlier estimates. The computerized photo on ZZZ is has more detail than the computerized photo at SciAm, so I guess they have done something in 3.5 years.
Re:Difficult to install Linux on a laptop, try Win
on
Linux Is Going Down
·
· Score: 2
RTFM is all real nice to say and everything but when the drivers aren't available it makes no difference in what order I (don't) install them.
<sarcasm>
Hello tech support, I tried not-installing the card stick non-driver both before and after I installed the special Sony hardware detector, but it still doesn't work.
<\sarcasm>
Also, (this is a serious question) how do I install Windows 2000 on a device that does not have a cd-rom or any pre-existing OS on the hard disk?
Slashdot Mirror
Other workloads, however, will not benefit from MOSIX. These statements are based on reading the docs a couple weeks back, not on actual experience.
Speaking from experience, you are pretty much correct. Jobs that use lots of CPU, but have little IO are good for mosix clusters, but jobs that have high IO are bad. The mosix filesystem and other things can partly get around the IO problems if the users plan carefully, but mostly they just want to start 30 jobs and forget about it for a few days.
There is no reason that a mosix cluster can't be combined with a batch/queueing system. This lets lazy/stupid users run their CPU bound jobs and lets mosix distribute them, but more savy users can script their IO jobs to run on particular machines and use local disk for IO.
It took a few months for the users of the cluster I setup to get trained into what jobs work well, and which kill the cluster. The problem is that launching 40 "good" jobs on a single machine is not a problem, because they just shoot out to the other nodes, but launching 40 "bad" jobs on a single machine will make that machine almost unusable.
This can have adverse effects on the cluster if the good jobs were started from the overloaded machine; for example the good jobs might have to check back with their originating machine every few minutes to update a checkpoint file.
Basically, mosix isn't some magic bullet to solve machine limitations, but it is a very cheap and effective way to solve certain problems.
Thanks, that answered many of my questions. I can certainly see how somebody could have committeed Crusade to death. I can just see the spec sheet now: (semi-)attractive female alien character, (semi-)evil corporate guy, potentially-cool but way overdone guy, etc., the Vorlon ambassador guy with the mysterious past was popular, so every character should have a mysterious past!
The characters in B5 started out a bit 2-dimensional, but I think that was on purpose. As the motives for all of the characters became clear they took on much more depth, and of course the characters changed in response to the changing situtations, etc.
That reminds me, I still have one unwatched episode of Crusade left on my Tivo, from sometime last year, that I still haven't watched. I saw it the first time around, but for as amazing as (the first four seasons of) B5 was, Crusade mostly sucked.
I think part of the problem with the fifth season of B5, Crusade, and to some degree Legend of the Rangers, was kind of a too-much-of-a-good-thing syndrome. For example, in B5, [what's her name, the woman commander] gives a nice speech before attacking the Earth fleet, about being death incarnate and all of that. The speech was a bit cliche if taken out of context, but for the situation and the character it worked well. Mostly because through the show, she only gave a few speeches like that, maybe one or two a season. When they happened they were pretty cool, but then there are characters on Crusade that only talk in fancy speeches. It's too much, they lose all impact and end up just being cheesy Shatneresque spewing.
I am sure JMS still has it in him to make a good show, but I think he really needs to go back and watch the third and fourth seasons of B5, and maybe In the Beginning, and see what the balance was that really made them work. Don't just take the best bits and try to only use those in a show.
I am still running two of these (533mhz 164SX) machines as general login and computing servers. For some reason people really like to use xdm on these old alphas from their Windows boxes. I even setup a nice dual processor Intel machine with loads of memory running Debian and the latest gnome and kde, but nobody seems interested in using that.
The old turbochannel alphas had some pretty serious reliability problems (a 90 day warranty on a $7000 computer!?) I had most of the DEC components (i.e., not 3rd party stuff, like disks) on my two turbochannel alphas replaced several times under maintenaince before getting to board revisions that could last more than 6 months.
However, the PCI based alphas I have seem to be totally bullet proof. I think in the whole time I have been running them, once lost a disk, which one can hardly blame DEC/Compaq for.
For integer stuff the 164SX machines are bit slow, probably comparable to a 350mhz PII, but for floating point, they are probably better than a 700mhz PIII (though I haven't benchmarked these thngs in years, so I may be remembering wrong). Of course they don't compare in any way to a $50 1Ghz Duron.
Some of the early turbochannel alphas, had badges saying DECstation on the front. The Personel DECStation 25 (a 25mhz mips R3000) and the Alpha 300 series even shared the same box. I think it is an Alpha 400 series maybe, that I am recalling with that label. I don't think the Alphastation and Alphaserver names were used until the PCI based systems.
All of my old DECstations (mips and alpha) are sitting in a storage room because it is too much bureaucracy to throw them away, but they just aren't worth pulling out and playing with.
I am probably getting some of the names confused, but this is definately a problem that Digital (aka DEC) perpetuated with their constant renaming of stuff. Talk about a company being run into the ground by poor marketing. Since I have been using it, their Unix OS has been called OSF/1, Digital Unix, and Tru64. At some point in the mid-90s, DEC decided that years of name recognition and reputation under the name DEC was too much, so they wanted everybody to call them Digital.
Having said that, there are some serious issues with 2.4 on some 8-way 8GB machines that I manage. They have been running 2.4.13-ac7 since November, because that is the last kernel that is usable for me (-ac11 would probably be ok). Newer kernels have terrible behavior under the intense IO load these machines go through. They get 14-30 days of uptime, and then hang or get resource starved or something and have to be rebooted.
I think part of the issue is that there simply aren't that many people running 8-way boxes, so bugs aren't found as easy, this is of course on top of having 8-way SMP being much more complex than a defacto single user, single processor desktop machine. To make it even worse, the machines are pushed hard. They move around GBs of data every day, and often will run for extended periods with loads over 25.
Of course, it is still mostly ok. While the machines are working they mostly work fine. Of course 20 days of uptime is totally unacceptable. I have an alpha running Tru64 pushing 300 days of uptime, and the last time it was down was due to a drive failure, not an OS problem.
My only remaining issue with Linux on "small" machines is an oscillation problem in IO. Data will fill up all available memory before being written to disk, and then everything from memory will be written out, and then memory fills up again before anything new is written to disk. This is a bit inefficient, and the machine's responsiveness at the memory-full part of the cycle is poor.
What are my options though? I guess I could try FreeBSD, but a bit of lurking on their lists and forums reveals plenty of problems there, too. Do I switch and hope things get better, or wait out 2.4 and hope it comes around soon? Aside from a few nasty bugs in some releases, pretty much each successive 2.4 kernel has been better than the previous one, at least on small systems.
Several years ago I was having a hard lockup problem with Tru64 (Digital Unix, at the time) and that was very scary. It took time to get the problem escalated to the OS engineers, instead of just sending an e-mail to lkm. Even then I could only hope that the issue was being addressed, but I had no way to know if anybody was doing anything about it or not. (Turned out to be an bug in the NFS server that would cause the machine to lockup when serving to AIX.) For all of its problems though, it is extremely reassuring for me to be able to monitor the development process of Linux through the linux-kernel-mailing list, and other specialized lists. If I feel that people aren't aware of some problem I am experiencing, I can raise the issue. I am not in the dark about what is happening, and what fixes are being made. I know what changes have gone into each kernel update, so I know if there is a chance of it fixing my problems.
The WTCHG is actually a cool place that looks for genes for complex diseases, writes useful software, and are heavily invested in using Linux as a scientific computing platform.
They have an 86 cpu Mosix/Linux cluster, and two 8 CPU, 8GB machines running Linux.
I can mostly deal with it being a GSM only device, and not have TDMA or CDMA, but calling it a dual-band world phone is pretty much an outright lie. Yeah, its a "world phone" if you never travel across oceans.
I still might get one if the service plans are good, but being able to go to the UK and stick a £10 Virgin pay as you go sim card in it would make the thing a true winner.
we came in and said there should be a platform that's identical with millions and millions of machines, and the bios of that should be open to everybody to use
I kind of see what BG is saying here, that the free software movement couldn't have taken off without open, commodity hardware. This is wrong. The free software movement, as it came out of RMS and the MIT AI lab was a direct response to proprietary, closed hardware and software. The free software movement grew out of the tradition of open access to software and tools on very non-standard mini computer hardware.
In the early 80s when the FSF was founded, it was not clear yet that the IBM PC would be such a dominant force in the computing world. Commodity home machines aren't even mentioned in RMS's initial announcement. In fact, he is talking about replacing the system on very expensive, practically custom built machines, which were only found in universities and big businesses.
Sure, the pervasiveness of computing has been a major boost for free software, but this is a base rate issue (i.e., there are x free software users out of n*x computer users).
I have always bought Dells, because they make doing educational institution purchases incredibly easy, and if I need service I just call one place. I can customize the computers I am buying, and their prices are reasonable.
I am finding myself in the position of having to buy a very fast computer for somebody else. The problem I am running into is that Dell does not sell Athlons. I can buy a 1.8Ghz P4 from them for about $1900 fully loaded. I can also build myself a dual Athlon 1.2Ghz for the same price, and the Athlon is much faster.
So my question, is there a reputable and reliable company which sells customizable Athlon machines for a reasonable price?
I am seeing these hits too. Since 18/Sep/2001:07:27:25 -0600 (it is now 09:16) I have been hit by 120 different machines. 105 of them are on my class B, 128.138, 14 more just start with 128, and only one is from a totally different address.
Perhaps I should contact the admins at my site who are in charge of the offending machines.
I used this manual for an Ericsson R320AT GSM phone. My phone certainly isn't a GSM phone, but the address book commands seem to work.
The commands also worked on my friend's Motorolla, which is a GSM phone. On that one, we needed to send an AT+CPBS="ME" to tell the phone to use the numbers stored in the phone, as opposed to those stored in the sim card.
BTW, I had to run unix2dos on the phonebooklist file in my previous message, because the phone expects lines to be terminated with a CRLF, instead of the Unix LF only.I was looking around for something to allow me to manage the phone book in my Ericsson R280L. I tried Kandy, part of KDE, but it didn't work well enough to do what I wanted. I finally just converted my address book from my palm into text, copied the numbers I wanted into emacs and created a list that looks like:
I connected to the phone with kermit and made sure it was ready to talk, and then ran something like
The sleep 1 is important, because the phone couldn't take entries any faster. A friends Motorolla could connect over IR, but it needed sleep 5 after each entry.Managing my phone list with emacs, sed, awk, and sort turned out to be much simpler than doing it with some click happy windows program I tried. I think it was TrueSync, or something that I got from Yahoo to synch my palm with Yahoo calendar. It claimed to be able to sync to my phone, but it only had two modes of operation, do nothing, and erase all numbers in the phone.
That actually isn't too outlandish. If you recall, during WWII the allies occasionally chose to let soldiers and civilians die rather than reveal that they could read the German's codes (and new where the next attack/bombing was coming).
When the allies had information from decrypted information that revealed the location of ships, they would always send a spotter plane over the ships before attacking to give the German's a plausible explanation for the allies knowing the ships' location.
Properly used and implemented PGP is uncrackable by brute force. Regardless of the size of the NSA's monster cracker they couldn't brute force PGP unless they have some secret knowledge. An as yet unknown (to us) flaw in PGP, or an advance in mathematics that allows for fast factoring of large numbers is something that I expect the NSA would readily kill people to keep secret.
As soon as knowledge of a secret like that got out, people would stop using PGP (or whatever), and the secret would become worthless.
I ran a connection over a dry loop for a few years when I lived in student family housing at my school. At the time, the only connectivity choice was a 9600baud ISN (serial) connection. So, for $3 a month or something the University's telecom group connected a dry pair between my apartment and my office on campus.
I used a pair of short haul modems to run a 38400 SLIP connection over the dry loop. This is nothing compared to the speed of DSL technology, but at the time $150 an end was about all I could afford, and DSL stuff wasn't available at a consumer level.This setup worked great until bridge construction (the over a river type of bridge) caused the connection to be rerouted, and it never worked right again. A kind telecom employee took pity on me and I used an illicit second phone line to dial into my office for another year or so, until somebody noticed the connection on the switchboard. After that I was limited to dialing in on my main phone line and getting a cell phone for voice. (Paying the university for a second phone line was more expensive than getting a cell phone.)
Of course, now a few years later the apartments I lived in have 10/100 connections onto the University backbone.
BTW, if anybody is interested in buying some used short haul modems, let me know...
I consider myself someone who reads lots of books, and I completely disagree with this statement. I think the people who say they like the feel of a book in there hand have never tried any type of e-reader. They weight of enough books to last me for a two week trip is not pleasant. I would much rather put a few books on my Palmpilot, which I have with anyways, than carry around an extra few pounds of paper.
I have been reading books on my Palmpilot for several years now, and I am completely addicted to it. I even have a Palm III with the old low contrast screen, so I would probably like it more if I moved to a V or 500 with a proper display.
I think people who don't like reading e-books have never tried it. (This is making the assumption that the books these people want to read are available in an usable format. I can completely understand people not wanting to read e-books because they have no interest in 100+ year old stuff from the Gutenberg project or whatever annoying thing the publishers have decided to make available.)
They have not specified what the extra hours are for, but I am pretty convinced it is for some type of targetted ads. 12 hours could never be filled over the phone line, so it has to be something sucked down from the airwaves, which is either pay-per-view or targeted ads, and ads seems more realistic.
Tivo has been told, and I hope they understand, that the users won't mind targetted ads as long as they in no way impact our viewing, as soon as we are forced to watch ads then all "well behaved" hacking stops, and we all learn how to hack the guide data and cancel our subscriptions.
I was very amused when I read about how they interpreted the button press data: They think if I always fastforward through the commercials on the Simpsons (for example) then the commercials on the Simpsons must really suck, but if I rarely fastforward through the commercials on Friends, then those must be really good, well targeted commercials. In reality it is completely the opposite. If I am actually paying attention to a show (like the Simpsons) then I will always fastforward through the commercials, but if the show is nearly meaningless to me (like Friends) then I don't fastforward through the commercials because I am probably in the kitchen washing dishes or something where I can't even see the TV.
I wish the best to Tivo, and I hope they can pull huge amounts of money from the networks and advertisers for all of my "data", but I reserve the right to laugh at their interpretation.
The whole school system should be setup to push each student to the maximum of their potential, regardless of how high or low it is. Sucks if my kid isn't as smart as that other one, but don't stop teaching calculus in high school, just because some of the kids won't do well in it.
Actually it is quite high. I don't have the statistics handy, but the SAT is basically an intelligence test.
The SAT is not even a particularly good predictor of good grades in college
This may be the case, but there is nothing better.
good grades in college is a pretty bad predictor of success in life
Success in life is a bit harder to define than good grades in college, but if "success" is measured as salary and responsibility, then there is not much better than grades in college to predict future success.
(re: all of this good and better business, just because something is the best, doesn't mean it is good, there is simply nothing that is closer to perfection. i.e. all operating systems suck, some just suck less than others.)
If we had a good test for intelligence
There are good tests for intelligence, people just don't like what they measure. Intelligence is one of the most reliable psychological characteristics that can be measured about a person. The correlation between a test given to an individual at one time point, and at a later time point is about 0.80, which is very high for psychological characteristics.
This is getting back to the point of the original post, it isn't that intelligence tests are bad, it is that people don't like what they say, which is that not all people are the same, some are smarter, some are much smarter, and there isn't much that anybody can do about it.
This doesn't matter. The test is designed to be a valid indicator of how well (as defined by grades) a person will do in college. If it asked about yak farming, because that was the best indicator of future performance, then that would be a valid question.
This all might seem silly, and no test can capture all of the complexities of an individual, but that is not the point. All of these personal, anecdotal exceptions to the relevancy of SATs don't matter.
You can disagree with the philosophy behind the following statements, but right now they reflect the situation in education as it exists: Currently colleges must do something to limit who can come in and who cannot. Colleges consider successful students to be those who go on to graduate from the college with high marks. The best indicator of how well somebody will do in college is a combination of their high school GPA and SAT or ACT score.
Now, as for the anecdotes, of course these statements don't apply exactly to you. This is where college admissions boards have an extremely difficult and possibly impossible job. They are trying to apply broad statistical principles to individuals, and many of them get it horribly wrong. I have seen lists of applicants with a (literal) cutoff line drawn at 1200 (or wherever). This is completely the wrong way to go about it. There is no valid reason that an 1190 with a 3.8 gpa is worse than a 1210 with a 3.7 gpa. The problem with the lines, is that somebody is always on the border. It is easy to see that a 900/2.1 might not be suitable for a particular institution, but what about the people between 1200 and 900?
Yes, this is exactly it. These tests, and standards based admissions, and GPAs, and grades themselves all suggest that people are different. Because so many people take literally the statement "all men are created equal" they think that these tests are somehow evil. Of course what "all men are created equal" means, is that all people should be given equal opportunities. Everybody is allowed to go to high school, but some people blow it there (or, through lack of ability, fail to do well), so they are then not given equal access to college. This is how a merit based system (such as our educational process) is supposed to work.
I am not saying this is a perfect world, of course people are not given equal opportunities, which is the whole point of the civil rights movement. The SATs and other standardized tests do not judge equally across different minority groups, and the reasons for this are not understood. Is it possible to quantify hundreds of years of keeping a group down into a few bonus points on a test?
Many Americans seem to have a problem with the suggestion that some people are smarter than others, and that some people have a better shot at doing well in society than others. The logical conclusion to this thinking (because we can only make people smarter to a limited degree) is to make everybody dumber (which is much easier than making people smarter).
The people who study and work hard to perform better on the SATs will tend to get better scores on the SATs than those who slacked off. Guess which two skills are extremely important in relation to getting good grades in college? Studying and working hard. In general (always exceptions) people who study and work hard in college make better grades than those who slack off.
I have a list of words that have appeared on SAT tests in the past, and you have a different list. We take the same test, and just by chance 10% of my words were on the test, but 20% of yours, so you do a little better (because otherwise we are pretty equal in our abilities). This is called error, and anybody using the results of the test should take this into account. If we both took the test again (another version) maybe I would do a little better because this time my words are all over the place. The admission committees knows this, and won't (unless they are stupid) think that somebody who scored an 1150 is clearly a better student than somebody who scored an 1130, though they both will probably do better in school than somebody who scored a 1000.
Of course the article is old (10/97?) and states that service between Philedelphia and Europe should start in 2000. I guess they are a bit behind their earlier estimates. The computerized photo on ZZZ is has more detail than the computerized photo at SciAm, so I guess they have done something in 3.5 years.
<sarcasm>
Hello tech support, I tried not-installing the card stick non-driver both before and after I installed the special Sony hardware detector, but it still doesn't work.
<\sarcasm>
Also, (this is a serious question) how do I install Windows 2000 on a device that does not have a cd-rom or any pre-existing OS on the hard disk?