The 7 minutes is clearly a bug in the client, or a configuration problem on the computer which makes it abort the processing right away. They guy figuring high up on the list with 7 minutes processing time said he had just one single 200 MHz Pentium running NT. In other words, it should have taken 60 hours not 7 minutes (I have essentially the same setup right here). Presumably the others in the stats with around the same times suffer from the same bug. The setiathome folks should fix the clients and rip out the bad results from the stats. I assume it will happen at some time in the future:-) TA
Sure you can write your own proxy even though you don't have the source. A netstat shows that the client connects to sagan.ssl.berkeley.edu, and a 'strings' on the binary shows 'shserver.ssl.berkeley.edu' which turns out to be the same as 'sagan' right now. So just make something that can take the connects from setiathome on port 80, and forward it to shserver.ssl.berkeley.edu port 80 (and the other way). Put this 'something' (which also understands your local proxy system of course) and put it on a computer that looks like 'shserver.ssl.berkeley.edu' for the client, you can do that just by putting a fake entry in the/etc/hosts file (and if you have a/etc/nsswitch.conf then set it to check local files before DNS of course). You can probably do it in Perl. TA
Hold your horses, the database server is off-line and has been for many hours. Before there it had been acting funny for many more hours (not recognizing user names, so impossible to set up teams). Try again tomorrow. TA
I think it's a great idea. Now let's hope that Intel won't interfere with the 'inside' part, I don't think they liked that 'Linux Inside' logo that was popular for a while. But that one tried to look like the Intel logo and that's something else. And Maddog can talk to Intel through LI can't he? TA
Re:what does the gcc command mean then?
on
GCC-2.95 in July
·
· Score: 1
Try 'gcc source.f' or any other . supported by the compiler collection. It compiles all of them, with the right front end. So 'gcc' doesn't mean 'C compiler' and hasn't for a long long time. TA
What you're describing is almost word-by-word the "self-test" you should try before applying to air traffic control school, according to an interview with some air traffic official in a newspaper the other day. If you can't do it then just forget about applying. Very few people can do it:-) TA
Thanks a lot, Steffen. It's really appreciated. The binaries are not available yet, but I keep checking.. although I suspect it will be Monday before they're there, nothing has moved since Friday (it's so strange to observe that people are not working on their pet projects in the weekends too, I'm not used to that:-) TA
If you want to post on/. every time there's a new development kernel you'll just swamp yourself. There could be a new development kernel maybe several times a day, now just after the new development tree has been opened. Linus always starts with revamping things completely, and then he sends out new versions at a frenetic rate so that the other developers can fix their parts (in other words, those kernels might not even compile).
"Branching is a particularly human activity. Other species cannot keep a goal in mind and at the same time switch to another task and then return to what they were originally doing without skipping a beat." I believe the above to be utter BS. I have never heard that claim before, nor seen anything to that effect in animals. And humans are not particularly good at multitasking either, the only things we can do at the same time are things that we have automated (and that means hard-wiring the neuron connections in the brain). Try concentrating on two or more things at the same time.. it's almost impossible. The very few individuals who can do that ends up as air traffic controllers or in the stock markets (those guys with all those telephones:-) TA
The point of the Drake equation isn't to validate any number you get out of it, it's the equation itself. You look at the equation and then you either agree or you don't that if you had numbers to put in then what comes out could be about right. Then there's the additional fun.. if you do put in numbers you'll find out that the result is a very low number even if you vary the input between whatever extremes you are willing to accept as maybe possible.. TA
Arecibo's main function is as receiver, so you don't "hear" it at all. But occasionally they use it as a Really Big radar, and then you could "hear" it too, however no way you could hear it across the galaxy -- too much noise in between (and inside the galaxy is where you want to listen anyway). The FAQ at the web site says it could be detected 10000 light years away. But: There's the big problem -- the beam is very narrow (a millionth of the total sky, according to the FAQ) and "hitting" something useful (like a civilization) is unlikely.
Now that's the right guy to talk to.. can't you make a statically linked version available? It should run on most if not all versions of Linux (the versions I have run so far have all been static). I'm sitting here with a bunch of now useless Linux boxes:-( TA
I agree it wasn't smart of them to go to glibc2.1, but the reason you see slowdown isn't because you went from i686 to i386, it's because the 0.42 version did only 1/8 as much work as the new version you tried. All the newer versions are slow, but don't run the older versions -- I'm sure they throw away any results from those versions, because they don't analyze the data the way the new versions do.
Of course it grabs 98% CPU! What do you think spare cycles actually are? If you start setiathome on a "fresh" machine with only you hacking at the keyboard then your CPU would be 99.9% idle. That's the cycles setiathome will use. That been said, even though setiathome runs with a lower "priority" (nice value) you will notice it, at least if you try to compile or something. Because as soon as it has the time slice it will spend it, even though it is low down on the priority list of tasks to get timeslices. It's quite a bit different on SGI boxes, you just start it with e.g. npri -h 200 (or npri -w on the newest versions of IRIX) and it will be truly low-priority, i.e. you will still see it using 98-100% CPU most of the time but it will yield instantly (preempting the timeslice) as soon as something else want to run. You don't notice anything. Great! I got this great tip from a nice guy on a mailing list. Thanks, if you read this:-) TA
Who's talking out of his ass here? If you haven't got better comments than that please shut up. Here's the.sig Dave Miller used last year: Yow! 11.26 MB/s remote host TCP bandwidth &//// 199 usec remote TCP latency over 100Mb/s//// ethernet. Beat that!////
I've been wondering about how well a Linux PC would actually be doing on a 100Mbit Ethernet.. I'm a bit worried that it wouldn't be too good, as there seem to be real work for the CPU to do. I became aware of this when I found that an R5000 SGI O2 couldn't do more than max. 5 MB/sec, memory-to-memory TCP no disk involved! And this used more than 60% of the CPU, the system was completely kneeling and with all the other work going on the TCP became a real bottleneck:-( IRIX 6.3 isn't the worst operating system in the world so this got me thinking. That on-board TCP stack seems interesting, but only if it supports something else than NT of course. But then there's the problem of embedded TCP stacks, I've yet to see one withouth strange bugs here and there. TCP stacks are notoriously difficult to get right, in practice it's only a real, open-source preferably Unix box that can be trusted to (eventually) get it Right. TA
The 'flat mode' hasn't worked well since the last big upgrade. In the past it was possible to look through the comment list, select a comment, select a reply comment, and then flattening that single subthread. Now I can't do that, if I want to flatten a subthread I have to select 'flat' mode for the whole thing and that's absolutely not what I want. That reloads it all and I have to start from the top, and then I have to go back etc. In short, reading threads easily is now a pain. Thanks for any improvements that can be done to this.. TA
A satellite groundstation I know in nothern Sweden (I don't have a map here but let's say it's around 67-68 deg. northern latitude) heats the building with the heat that the computer room coolers generate. However, many years ago when they were installing the thing only one of the two huge coolers (each is as big as a ship's engine) was working.. it got so hot in the computer room we had to keep all the 3-meter tall windows wide open, with -35 C (-31 F I think) outside and the temperature brakers still tripped:-) TA
You can't just say that the network would be a bottleneck, that's totally dependent on the application. Look at setiathome, they will be processiong enormous amounts of data in the end but the CPU/data rate is still very big so the slow Internet is more than fast enough. 340 KB of data takes half an hour on a fast MIPS CPU, one hour twenty on a 400MHz Pentium-II. They're doing something like the equivalent of 15 *years* of Pentium CPU time *each day* now, and growing.. TA
First, when Crimson came out it had a R4000, not R4400. And about the 200MHz+ Intel beating R4400 I'm not so sure.. I've been running `setiathome' the last week and a 400MHz Pentium uses 80 minutes to do one workload, that's about equal to a 215-220MHz R4400 if such a thing existed, in other words slightly faster than one CPU in my 200MHz old Challenge running the same program. A 250MHz R10000 runs circles around the 400MHz Pentium II. A bit disappointing actually, particularly taking into account that `setiathome' isn't even compiled for an R10000, my other applications go 30% faster when I compile for it.
Why should anybody think that Microsoft has endorsed the "where do you want to go tomorrow?" slogan? Of course MS hasn't endorsed it! Why should anybody imagine they had??? As I already said, "where do you want to go tomorrow" ridicules the MS slogan, and rightly so. Nobody wants nor needs Microsoft to "endorse" that! This isn't Microsoft's slogan, it's not what they trademarked, it's none of Microsoft's business. They don't like the ridicule, but there's no law against that in any democratic country I know about. TA
Why should SGI, or the stockholders themselves, care about what the stockholders think? It's the customers opinion that matters! I was the guy who advocated we should start looking into SGI servers for our technical computing a bunch of years ago, to replace what we were using by then. What got me interested in the first place was the name: "Silicon Graphics"! It was, and still is, hi-tech sounding, and it got me interested enough to check them up and realize it meant they had high bandwidth, high capacity, good I/O, fast CPUs. "Silicon Graphics", hey this means they need to be fast overall, not just have peak performance in one place (like HP) or another place (like DEC Alpha boxes which were introduced back then), or any other. Well we still buy a lot of them because the benchmark went well:-) The name started it all.. never would I have been interested if they had this new name back then (which is so boring I just can't remember it five seconds).
That's a pretty bad move of SGI, the original name was, and still is very special, something that makes you think a bit about what it implies. The new name is even more boring than "International Business Machines". Much worse. And the new logo.. what a disappointment. Funky 'g', but it just disappears, it's mainstream, it's tragic. Now what incredibly bad consultancy company managed to get rich from such a bad job? And who was the CEO at SGI who thought this was a good idea? And the board? Definitively the worst name/logo change I have seen in at least ten years. TA
Slashdot Mirror
The 7 minutes is clearly a bug in the client, or a configuration problem on the computer which makes it abort the processing right away. They guy figuring high up on the list with 7 minutes processing time said he had just one single 200 MHz Pentium running NT. In other words, it should have taken 60 hours not 7 minutes (I have essentially the same setup right here). :-)
Presumably the others in the stats with around the same times suffer from the same bug. The setiathome folks should fix the clients and rip out the bad results from the stats. I assume it will happen at some time in the future
TA
Sure you can write your own proxy even though you don't have the source. /etc/hosts file (and if you have a /etc/nsswitch.conf then set it to check local files before DNS of course).
A netstat shows that the client connects to sagan.ssl.berkeley.edu, and a 'strings' on the binary shows 'shserver.ssl.berkeley.edu' which turns out to be the same as 'sagan' right now.
So just make something that can take the connects from setiathome on port 80, and forward it to shserver.ssl.berkeley.edu port 80 (and the other way). Put this 'something' (which also understands your local proxy system of course) and put it on a computer that looks like 'shserver.ssl.berkeley.edu' for the client, you can do that just by putting a fake entry in the
You can probably do it in Perl.
TA
Hold your horses, the database server is off-line and has been for many hours. Before there it had been acting funny for many more hours (not recognizing user names, so impossible to set up teams).
Try again tomorrow.
TA
I think it's a great idea. Now let's hope that Intel won't interfere with the 'inside' part, I don't think they liked that 'Linux Inside' logo that was popular for a while. But that one tried to look like the Intel logo and that's something else. And Maddog can talk to Intel through LI can't he?
TA
Try 'gcc source.f' or any other . supported by the compiler collection.
It compiles all of them, with the right front end. So 'gcc' doesn't mean 'C compiler' and hasn't for a long long time.
TA
What you're describing is almost word-by-word the "self-test" you should try before applying to air traffic control school, according to an interview with some air traffic official in a newspaper the other day. :-)
If you can't do it then just forget about applying. Very few people can do it
TA
Thanks a lot, Steffen. It's really appreciated. :-)
The binaries are not available yet, but I keep checking.. although I suspect it will be Monday before they're there, nothing has moved since Friday (it's so strange to observe that people are not working on their pet projects in the weekends too, I'm not used to that
TA
If you want to post on /. every time there's a new development kernel you'll just swamp yourself. There could be a new development kernel maybe several times a day, now just after the new development tree has been opened. Linus always starts with revamping things completely, and then he sends out new versions at a frenetic rate so that the other developers can fix their parts (in other words, those kernels might not even compile).
"Branching is a particularly human activity. Other species cannot keep :-)
a goal in mind and at the same time switch to another task and then
return to what they were originally doing without skipping a beat."
I believe the above to be utter BS. I have never heard that claim before, nor seen anything to that effect in animals. And humans are not particularly good at multitasking either, the only things we can do at the same time are things that we have automated (and that means hard-wiring the neuron connections in the brain). Try concentrating on two or more things at the same time.. it's almost impossible. The very few individuals who can do that ends up as air traffic controllers or in the stock markets (those guys with all those telephones
TA
The point of the Drake equation isn't to validate any number you get out of it, it's the equation itself. You look at the equation and then you either agree or you don't that if you had numbers to put in then what comes out could be about right.
Then there's the additional fun.. if you do put in numbers you'll find out that the result is a very low number even if you vary the input between whatever extremes you are willing to accept as maybe possible..
TA
Arecibo's main function is as receiver, so you don't "hear" it at all. But occasionally they use it as a Really Big radar, and then you could "hear" it too, however no way you could hear it across the galaxy -- too much noise in between (and inside the galaxy is where you want to listen anyway). The FAQ at the web site says it could be detected 10000 light years away. But:
There's the big problem -- the beam is very narrow (a millionth of the total sky, according to the FAQ) and "hitting" something useful (like a civilization) is unlikely.
Now that's the right guy to talk to.. can't you make a statically linked version available? It should run on most if not all versions of Linux (the versions I have run so far have all been static). :-(
I'm sitting here with a bunch of now useless Linux boxes
TA
I agree it wasn't smart of them to go to glibc2.1, but the reason you see slowdown isn't because you went from i686 to i386, it's because the 0.42 version did only 1/8 as much work as the new version you tried. All the newer versions are slow, but don't run the older versions -- I'm sure they throw away any results from those versions, because they don't analyze the data the way the new versions do.
Of course it grabs 98% CPU! What do you think spare cycles actually are? If you start setiathome on a "fresh" machine with only you hacking at the keyboard then your CPU would be 99.9% idle. That's the cycles setiathome will use. That been said, even though setiathome runs with a lower "priority" (nice value) you will notice it, at least if you try to compile or something. Because as soon as it has the time slice it will spend it, even though it is low down on the priority list of tasks to get timeslices. :-)
It's quite a bit different on SGI boxes, you just start it with e.g. npri -h 200 (or npri -w on the newest versions of IRIX) and it will be truly low-priority, i.e. you will still see it using 98-100% CPU most of the time but it will yield instantly (preempting the timeslice) as soon as something else want to run. You don't notice anything. Great! I got this great tip from a nice guy on a mailing list. Thanks, if you read this
TA
Thanks,
TA
>I always patch this away when I compile a kernel) :-)
Would you mind posting a patch for us lazy people?
TA
Who's talking out of his ass here? If you haven't got better comments than that please shut up. .sig Dave Miller used last year: //// //// ////
Here's the
Yow! 11.26 MB/s remote host TCP bandwidth &
199 usec remote TCP latency over 100Mb/s
ethernet. Beat that!
I've been wondering about how well a Linux PC would actually be doing on a 100Mbit Ethernet.. I'm a bit worried that it wouldn't be too good, as there seem to be real work for the CPU to do. I became aware of this when I found that an R5000 SGI O2 couldn't do more than max. 5 MB/sec, memory-to-memory TCP no disk involved! And this used more than 60% of the CPU, the system was completely kneeling and with all the other work going on the TCP became a real bottleneck :-(
IRIX 6.3 isn't the worst operating system in the world so this got me thinking.
That on-board TCP stack seems interesting, but only if it supports something else than NT of course.
But then there's the problem of embedded TCP stacks, I've yet to see one withouth strange bugs here and there. TCP stacks are notoriously difficult to get right, in practice it's only a real, open-source preferably Unix box that can be trusted to (eventually) get it Right.
TA
The 'flat mode' hasn't worked well since the last big upgrade. In the past it was possible to look through the comment list, select a comment, select a reply comment, and then flattening that single subthread. Now I can't do that, if I want to flatten a subthread I have to select 'flat' mode for the whole thing and that's absolutely not what I want. That reloads it all and I have to start from the top, and then I have to go back etc. In short, reading threads easily is now a pain.
Thanks for any improvements that can be done to this..
TA
A satellite groundstation I know in nothern Sweden (I don't have a map here but let's say it's around 67-68 deg. northern latitude) heats the building with the heat that the computer room coolers generate. However, many years ago when they were installing the thing only one of the two huge coolers (each is as big as a ship's engine) was working.. it got so hot in the computer room we had to keep all the 3-meter tall windows wide open, with -35 C (-31 F I think) outside and the temperature brakers still tripped :-)
TA
You can't just say that the network would be a bottleneck, that's totally dependent on the application. Look at setiathome, they will be processiong enormous amounts of data in the end but the CPU/data rate is still very big so the slow Internet is more than fast enough. 340 KB of data takes half an hour on a fast MIPS CPU, one hour twenty on a 400MHz Pentium-II.
They're doing something like the equivalent of 15 *years* of Pentium CPU time *each day* now, and growing..
TA
First, when Crimson came out it had a R4000, not R4400.
And about the 200MHz+ Intel beating R4400 I'm not
so sure.. I've been running `setiathome' the last week
and a 400MHz Pentium uses 80 minutes to do one workload,
that's about equal to a 215-220MHz R4400 if such a thing
existed, in other words slightly faster than one CPU in my
200MHz old Challenge running the same program.
A 250MHz R10000 runs circles around the 400MHz Pentium II.
A bit disappointing actually, particularly taking into account
that `setiathome' isn't even compiled for an R10000, my other applications go 30% faster when I compile for it.
Why should anybody think that Microsoft has endorsed the "where do you want to go tomorrow?" slogan? Of course MS hasn't endorsed it! Why should anybody imagine they had??? As I already said, "where do you want to go tomorrow" ridicules the MS slogan, and rightly so. Nobody wants nor needs Microsoft to "endorse" that!
This isn't Microsoft's slogan, it's not what they trademarked, it's none of Microsoft's business. They don't like the ridicule, but there's no law against that in any democratic country I know about.
TA
Why should SGI, or the stockholders themselves, care about what the stockholders think? :-)
It's the customers opinion that matters! I was the guy who advocated we should start looking into SGI servers for our technical computing a bunch of years ago, to replace what we were using by then. What got me interested in the first place was the name: "Silicon Graphics"! It was, and still is, hi-tech sounding, and it got me interested enough to check them up and realize it meant they had high bandwidth, high capacity, good I/O, fast CPUs. "Silicon Graphics", hey this means they need to be fast overall, not just have peak performance in one place (like HP) or another place (like DEC Alpha boxes which were introduced back then), or any other. Well we still buy a lot of them because the benchmark went well
The name started it all.. never would I have been interested if they had this new name back then (which is so boring I just can't remember it five seconds).
That's a pretty bad move of SGI, the original name was, and still is very special, something that makes you think a bit about what it implies. The new name is even more boring than "International Business Machines". Much worse.
And the new logo.. what a disappointment. Funky 'g', but it just disappears, it's mainstream, it's tragic.
Now what incredibly bad consultancy company managed to get rich from such a bad job? And who was the CEO at SGI who thought this was a good idea? And the board?
Definitively the worst name/logo change I have seen in at least ten years.
TA