One thing to keep in mind is that, last I checked, Beowulf clusters can still only be used on 2.0.x kernels, which implies reduced performance, security, and stability compared to a system such as MOSIX which will run on 2.2.x (or future 2.4.x) kernels. If this has changed, please let me know, but it was accurate as of September, and I haven't seen any new announcements on freshmeat about it, so I'm assuming it hasn't changed.
I admin a small (8 node) cluster, and so far my options are stay at Redhat 5.2/2.0.36 until Beowulf is ported to 2.2 (with all the problems of finding glibc 2.0.x RPMs now, etc), or install Redhat 6.1 and then reinstall 2.0.36. Neither of these seems particularly optimal.
I still get that feeling, however, setting and programing, and even more so, setting and playing something like Quake2 or Unreal Tournament.
Freespace (btw, one of the best games I've ever played) is like that for me. I can sit down a quite literaly play for hours without thinking once, it's all just instinct and reaction. You use one hand to control the keypad (you need all the number keys for controlling the ship decently well), and the other for things like weapons and speed control. Since it's all configurable (good thing, since the defaults kind of suck, IIRC).
I've only been in the zone when I'm programming once or twice. Sadly, I have a hard time getting into it with vi or emacs; Borland C++ 5 is the only editor I've found that can do it. Sadly, I hate Windows.:(
She really believed (and still believes, as far as I know) in government serving the public interest, and told me it's a common Canadian attitude.
I hadn't heard of this before. I mean, OpenBSD is based in Canada; how much more paranoid can a project get? [Note: I like paranoia:)]. Also, my mom is from Canada, and she trusts the govt probably about as much as I do (ie, none).
Last I checked, pretty good (much better than the US). The SSLeay (now OpenSSL) and Cryptix projects were both started there. Kind of an odd contradiction, really... unless, of course, they're planning on making some, uh, alterations (read: reversals) to their encryption laws soon (which would not suprise me much).
True, but sandworms don't need to fly. They just wait until Smaug goes to sleep (or just sits down to count his treasure) and then they eat his stupid ass.:) And don't try to tell me that Smaug was smart. Why didn't he just kill Bilbo when Smaug heard him the first time? Did he really need to play stupid riddle games? (And double that with Gollum!) And how about The God Emperor? Not only was he fscking huge and strong, but could also see the future and was extremely intelligent.
For must stuff, I've really got to say it just doesn't matter. For instance, anyone who tries to optimize GUI code for speed should be repeatedly hit on the head. OTOH, some areas still require good optimiztation: graphics (I mean the low-level libs here), real time stuff, crypto, numerical work, things like that. However, most of that is not getting fast code, but fast algorithms, and finding "tricks" for doing those algorithms even faster. That is to say, worry about high level, first order effeciency, and let the compiler handle the hard stuff (like pipelining).
But for the most part, it's best to just write clean, readable code. It doesn't matter that the code is fast if it can't be read, since someone will just have to redo it when it needs changes. Also it usually helps the compiler optimize more if it can figure out what's going on. For instance, the KAI C++ compiler has a very very good optimizer (my code runs 2-3 times as fast over gcc 2.95.2 on the same machine), and the docs recommend you write plain, readable code so the optimizer can see what's going on. One time I replaced some code with inline asm, and it got significantly slower. I took it back out.:)
I've always defined science fiction as writing that used science/technology to make "what if?" questions possible, but not emphasizing the tech just for the sake of the tech. Frank Herbert being a good example. Most of the time, he avoids tech as much as possible, just adding it in where he needs it.
Fantasy, OTOH, tends to involve elements that are just plain impossible in the "real world", such as dragons, magic, gods, etc., though the line between fantasy and sci-fi can get quite thin at times. For instance, Paul Atriedies (sp?) in Dune is condsidered as a god throughout the universe, and those sandworms would kick Smaug's ass anyday, I'd bet.
Other authors, like Clarke, tend to steer clear of such things for the most part. Been a while seen I've read much sci-fi, so I'm being deliberatly vauge here.
but what that cracker did was illegal - so surely he should be punished?
But the punishment should fit the crime. 1.5 years in jail for defacing a web page is like executing someone for having cigarettes underage (OK, those are necessarily equivalent, but you get my drift). This isn't even a money-making organization... if it were someone like ebay or e*trade I could understand that there were monetary damages, etc. and it would seem at least a little reasonable (but not really). There is no reason for the govt to be doing this except that he made them look like idiots, and they don't like it, though I don't see why they don't just get used to it, seeing how often it happens.:)
That said, breaking into the White House's web page is pretty fscking stupid. If he was smart, he would have posted the vulnerability to Bugtraq (along with a working 'sploit), got a lot of publicity, and someone else could get busted for actually using it. Oh, well...
One of the FAQs (check the website), says that both Linux and BeOS will be supported... so I'd imagine that eventually drivers for both will be released, though probably not as early as the Windows drivers. Also in the FAQ, there are no plans for Mac support at this time.
Basically, I choose a phrase or common theme (like a musical group I like, etc) and then take the first letter or two of each word, then 37337-1z3 it. This can generate nice long passwords if you need them, for instance, my PGP key is encrypted with an 18 character long phrase based on a musical group, using such obscure things that it would be rather hard for someone to guess.
Also, using psuedo-perl code generates instant line noise passwords, and as long as you're up on your perl, everything is easy to remember. For instance (this one is easy, but you get the idea):
I think RMS's demands are not only reasonable, but good ideas as well. I know I'd support them: they can only help OSS. And since software patents, in general, totally suck, demand #2 would be so, so awesome. I wish more companies would do this voluntarily. (Does anyone know of any that do?)
Interestingly, one of Sun's demands (a link is posted elsewhere) matches demand #1. Probably very popular (and the WINE and SAMBA people would love it).
And the suggestions given by RMS would benefit not only (as he says) GNU/Linux, but also *BSD (*nix in general, really), BeOS, MacOS, (and others!) and the developers of apps for all of those OSes. This means better sw for all involved, since less effort to port => more sw ported => more options => less likely to be stuck with crap b/c it's the only thing for the platform.
OK, hypothetical question. You rub a magic lamp, and a genie comes out. Specifically, a cryptographic protocol genie. He can come up with an effecient, secure protocol for any activity you want (assuming a protocol is possible, of course). What would you pick, and more importantly, why?
I mean, do I really care if my computional analysis experiments, or neural nets finish 20% faster?
Yeah, generally pretty true. I do a lot of development work, though, and spending 4-6 minutes for every build is not my idea of a fun way to spend time. Course I'm doing it on a POS P-II 350, so that really shouldn't be a suprise.
what if we computed FPS/dollar for AMD/Intel and Alpha? hmm? hehe
Well, yeah, a fast Alpha is mad $$$. But for speed, it's where it's at. [flamebait]And anyway, the intel architechture is crap[end flamebait].
Since when? Maybe very very VERY suspect boards, but I really don't think a decent board would die like that. I've never heard of something like this. Is this for real, or are you just fucking with my head?
And as for running hot, have you used a P-II lately?:)
Since intel code names are based on geographic features, i'd assume kass-kades, as that's how you pronounce the matching river if you're from the area.
I'm from Oregon, and have never heard of the Cascade river... just the Cascade mountains. At least not that I can remember. Am I just having a bad mental day? The only major rivers in oregon are the willamette (despite living in oregon up to age 17, I still have trouble spelling that freakin' river's name!) and the columbia. Unless, of course, that 6 pack of dew I just drank fried my brain, which really isn't that unlikely.
"This release will allow Intel to say, once again, that it offers the fastest chip on the market."
Fastest chips, my ass. Alpha, anyone? If you need raw computational power, Intel is not the way to go, IMO. Unless its in a 8-32 node cluster, which is pretty nice.:)
The best book for learning crypto is, IMHO, the Handbook of Applied Cryptography, by Alfred J. Menezes, Paul C. van Oorschot and Scott A. Vanstone. You can find more about it at http://www.cacr.math.uwaterloo.ca/hac/.
BTW, I didn't really feel AC was really that great... he ignores of a lot of important details, which is fine for a simple overview, but it makes it hard to do anything useful based soley on the information contained within AC.
Really, the best referece is the conference papers from the Fast Software Encryption, Crypto, EuroCrypt, AusCrypt (now defunct?), and AsiaCrypt conferences (also the AES conferences, but there will only be 3 of those). Most (probably all, actually) are published in the LCNS series by Springer (lots of good titles, check them out).
Also, a lot of good stuff is available on the web, since most of the academic people who write papers put them up on the web, and commercial companies often publish things in RFCs, etc. Here are a few to start you out:
The RC ciphers, RC2, RC4 and RC5, are copyrighted.
WRONG. RC5 (and RC6) are patented by RSADSI, as well as trademarked. The actual patent is on data-dependant rotations [great, now they're patenting machine instructions...I wonder if I can get xor:)]. But anyway, you don't want to use RC5 unless a) You have a license from RSADSI. b) You live in a country where it's not patented. c) You like to break the law.
The Weizmann Institute is one of the best research institutes in the world (think MIT-level). Shamir (inventor of Twinkle, co-inventor of RSA, co-inventor of differential cryptoanalysis, etc) worked (still works?) there, as does Eli Biham (differential analysis, related-key analysis, impossible differentials, and Serpent), Goldwasser, sereral others who are very presitious in the crypto community. If this were done anywhere in the world, it would probably be there.
OTOH, I haven't been keeping up on the state of quantum computing research, so I don't know if something like this could be built. (I remember that several stories about Twinkle made it sound like it actually existed, while it's just a design).
Given the general public's clue-lessness about crypto, I'd wait for some confirmation, preferably by the Weizmann Institue itself.
Slashdot Mirror
One thing to keep in mind is that, last I checked, Beowulf clusters can still only be used on 2.0.x kernels, which implies reduced performance, security, and stability compared to a system such as MOSIX which will run on 2.2.x (or future 2.4.x) kernels. If this has changed, please let me know, but it was accurate as of September, and I haven't seen any new announcements on freshmeat about it, so I'm assuming it hasn't changed.
I admin a small (8 node) cluster, and so far my options are stay at Redhat 5.2/2.0.36 until Beowulf is ported to 2.2 (with all the problems of finding glibc 2.0.x RPMs now, etc), or install Redhat 6.1 and then reinstall 2.0.36. Neither of these seems particularly optimal.
I still get that feeling, however, setting and programing, and even more so, setting and playing something like Quake2 or Unreal Tournament.
:(
Freespace (btw, one of the best games I've ever played) is like that for me. I can sit down a quite literaly play for hours without thinking once, it's all just instinct and reaction. You use one hand to control the keypad (you need all the number keys for controlling the ship decently well), and the other for things like weapons and speed control. Since it's all configurable (good thing, since the defaults kind of suck, IIRC).
I've only been in the zone when I'm programming once or twice. Sadly, I have a hard time getting into it with vi or emacs; Borland C++ 5 is the only editor I've found that can do it. Sadly, I hate Windows.
AAAAHHHH!!!! Late for class!!
She really believed (and still believes, as far as I know) in government serving the public interest, and told me it's a common Canadian attitude.
:)]. Also, my mom is from Canada, and she trusts the govt probably about as much as I do (ie, none).
I hadn't heard of this before. I mean, OpenBSD is based in Canada; how much more paranoid can a project get? [Note: I like paranoia
What is the state of encryption laws Down Under?
Last I checked, pretty good (much better than the US). The SSLeay (now OpenSSL) and Cryptix projects were both started there. Kind of an odd contradiction, really... unless, of course, they're planning on making some, uh, alterations (read: reversals) to their encryption laws soon (which would not suprise me much).
True, but sandworms don't need to fly. They just wait until Smaug goes to sleep (or just sits down to count his treasure) and then they eat his stupid ass. :) And don't try to tell me that Smaug was smart. Why didn't he just kill Bilbo when Smaug heard him the first time? Did he really need to play stupid riddle games? (And double that with Gollum!) And how about The God Emperor? Not only was he fscking huge and strong, but could also see the future and was extremely intelligent.
For must stuff, I've really got to say it just doesn't matter. For instance, anyone who tries to optimize GUI code for speed should be repeatedly hit on the head. OTOH, some areas still require good optimiztation: graphics (I mean the low-level libs here), real time stuff, crypto, numerical work, things like that. However, most of that is not getting fast code, but fast algorithms, and finding "tricks" for doing those algorithms even faster. That is to say, worry about high level, first order effeciency, and let the compiler handle the hard stuff (like pipelining).
:)
But for the most part, it's best to just write clean, readable code. It doesn't matter that the code is fast if it can't be read, since someone will just have to redo it when it needs changes. Also it usually helps the compiler optimize more if it can figure out what's going on. For instance, the KAI C++ compiler has a very very good optimizer (my code runs 2-3 times as fast over gcc 2.95.2 on the same machine), and the docs recommend you write plain, readable code so the optimizer can see what's going on. One time I replaced some code with inline asm, and it got significantly slower. I took it back out.
Just my 2% of a $.
I've always defined science fiction as writing that used science/technology to make "what if?" questions possible, but not emphasizing the tech just for the sake of the tech. Frank Herbert being a good example. Most of the time, he avoids tech as much as possible, just adding it in where he needs it.
Fantasy, OTOH, tends to involve elements that are just plain impossible in the "real world", such as dragons, magic, gods, etc., though the line between fantasy and sci-fi can get quite thin at times. For instance, Paul Atriedies (sp?) in Dune is condsidered as a god throughout the universe, and those sandworms would kick Smaug's ass anyday, I'd bet.
Other authors, like Clarke, tend to steer clear of such things for the most part. Been a while seen I've read much sci-fi, so I'm being deliberatly vauge here.
but what that cracker did was illegal - so surely he should be punished?
:)
But the punishment should fit the crime. 1.5 years in jail for defacing a web page is like executing someone for having cigarettes underage (OK, those are necessarily equivalent, but you get my drift). This isn't even a money-making organization... if it were someone like ebay or e*trade I could understand that there were monetary damages, etc. and it would seem at least a little reasonable (but not really). There is no reason for the govt to be doing this except that he made them look like idiots, and they don't like it, though I don't see why they don't just get used to it, seeing how often it happens.
That said, breaking into the White House's web page is pretty fscking stupid. If he was smart, he would have posted the vulnerability to Bugtraq (along with a working 'sploit), got a lot of publicity, and someone else could get busted for actually using it. Oh, well...
One of the FAQs (check the website), says that both Linux and BeOS will be supported... so I'd imagine that eventually drivers for both will be released, though probably not as early as the Windows drivers. Also in the FAQ, there are no plans for Mac support at this time.
Basically, I choose a phrase or common theme (like a musical group I like, etc) and then take the first letter or two of each word, then 37337-1z3 it. This can generate nice long passwords if you need them, for instance, my PGP key is encrypted with an 18 character long phrase based on a musical group, using such obscure things that it would be rather hard for someone to guess.
Also, using psuedo-perl code generates instant line noise passwords, and as long as you're up on your perl, everything is easy to remember. For instance (this one is easy, but you get the idea):
my=~s/$p4ss/@w0rd/g;
It doesn't make sense, but that's ok.
It's a knife in Wolfenstein, fist and chainsaw in Doom, axe in Quake I, don't know in Quake II (never played it)
Quake II had this little blaster thing. It never runs out of ammo, but it's pretty wussy. I'd take a chainsaw or an axe any day.
I think RMS's demands are not only reasonable, but good ideas as well. I know I'd support them: they can only help OSS. And since software patents, in general, totally suck, demand #2 would be so, so awesome. I wish more companies would do this voluntarily. (Does anyone know of any that do?)
Interestingly, one of Sun's demands (a link is posted elsewhere) matches demand #1. Probably very popular (and the WINE and SAMBA people would love it).
And the suggestions given by RMS would benefit not only (as he says) GNU/Linux, but also *BSD (*nix in general, really), BeOS, MacOS, (and others!) and the developers of apps for all of those OSes. This means better sw for all involved, since less effort to port => more sw ported => more options => less likely to be stuck with crap b/c it's the only thing for the platform.
So, good all around, I'd say
They don't have a braille version of every book they publish. It's blatant discrimination.
I say the blind should go after the National Gallery... after all, they are restricted from seeing the paintings!
Oh, this wonderful (not) PC world we live in.
OK, hypothetical question. You rub a magic lamp, and a genie comes out. Specifically, a cryptographic protocol genie. He can come up with an effecient, secure protocol for any activity you want (assuming a protocol is possible, of course). What would you pick, and more importantly, why?
maybe if they port Quake3 to an alpha :)
Oh, god, I wish.
I mean, do I really care if my computional analysis experiments, or neural nets finish 20% faster?
Yeah, generally pretty true. I do a lot of development work, though, and spending 4-6 minutes for every build is not my idea of a fun way to spend time. Course I'm doing it on a POS P-II 350, so that really shouldn't be a suprise.
what if we computed FPS/dollar for AMD/Intel and Alpha? hmm? hehe
Well, yeah, a fast Alpha is mad $$$. But for speed, it's where it's at. [flamebait]And anyway, the intel architechture is crap[end flamebait].
Since when? Maybe very very VERY suspect boards, but I really don't think a decent board would die like that. I've never heard of something like this. Is this for real, or are you just fucking with my head?
:)
And as for running hot, have you used a P-II lately?
Since intel code names are based on geographic features, i'd assume kass-kades, as that's how you pronounce the matching river if you're from the area.
I'm from Oregon, and have never heard of the Cascade river... just the Cascade mountains. At least not that I can remember. Am I just having a bad mental day? The only major rivers in oregon are the willamette (despite living in oregon up to age 17, I still have trouble spelling that freakin' river's name!) and the columbia. Unless, of course, that 6 pack of dew I just drank fried my brain, which really isn't that unlikely.
"This release will allow Intel to say, once again, that it offers the fastest chip on the market."
:)
Fastest chips, my ass. Alpha, anyone? If you need raw computational power, Intel is not the way to go, IMO. Unless its in a 8-32 node cluster, which is pretty nice.
The best book for learning crypto is, IMHO, the Handbook of Applied Cryptography, by Alfred J. Menezes, Paul C. van Oorschot and Scott A. Vanstone. You can find more about it at http://www.cacr.math.uwaterloo.ca/hac/.
/
BTW, I didn't really feel AC was really that great... he ignores of a lot of important details, which is fine for a simple overview, but it makes it hard to do anything useful based soley on the information contained within AC.
Really, the best referece is the conference papers from the Fast Software Encryption, Crypto, EuroCrypt, AusCrypt (now defunct?), and AsiaCrypt conferences (also the AES conferences, but there will only be 3 of those). Most (probably all, actually) are published in the LCNS series by Springer (lots of good titles, check them out).
Also, a lot of good stuff is available on the web, since most of the academic people who write papers put them up on the web, and commercial companies often publish things in RFCs, etc. Here are a few to start you out:
Blowfish: http://www.counterpane.com/bfsverlag.html
CAST5: http://www.ietf.org/rfc/rfc2144.txt
Tiger: http://www.cs.technion.ac.il/~biham/Reports/Tiger
MD2, MD4, MD5, RC5, and others are avaialable as RFCs as well (use the search, Luke!).
Ok, one more, all of the AES Canadates are
at http://www.nist.gov/aes/
Also, looking at crypto code often is helpful for understanding what's going on (since you can relate your programming knowledge to the crypto).
Good luck
WRONG. RC5 (and RC6) are patented by RSADSI, as well as trademarked. The actual patent is on data-dependant rotations [great, now they're patenting machine instructions...I wonder if I can get xor :)]. But anyway, you don't want to use RC5 unless
a) You have a license from RSADSI.
b) You live in a country where it's not patented.
c) You like to break the law.
Sorry, but using RC5 would be a very bad idea.
The Weizmann Institute is one of the best research institutes in the world (think MIT-level). Shamir (inventor of Twinkle, co-inventor of RSA, co-inventor of differential cryptoanalysis, etc) worked (still works?) there, as does Eli Biham (differential analysis, related-key analysis, impossible differentials, and Serpent), Goldwasser, sereral others who are very presitious in the crypto community. If this were done anywhere in the world, it would probably be there.
OTOH, I haven't been keeping up on the state of quantum computing research, so I don't know if something like this could be built. (I remember that several stories about Twinkle made it sound like it actually existed, while it's just a design).
Given the general public's clue-lessness about crypto, I'd wait for some confirmation, preferably by the Weizmann Institue itself.