Do you know why they are 20 times more expensive than normal bulbs? Because they take approximately 20 times more resources to make (the lighting field is highly competitive, so the cost to you basically reflects the cost to manufacture them). So if you don't save more than the $10 extra to manufacture one energy efficient bulb, over its lifetime, in saved electricity, then you have done more harm than good.
I bought a whole bunch of energy efficient bulbs. Most of them died within a year because they don't like dirty electricity and being cycled rapidly -- shorter than the average lifetime I get out of normal bulbs (despite the marketing blurb explaining that they last 8 times longer). The only energy efficient bulbs I have retained are the ones in the living/lounge room -- ie, the ones that are on for a substantial part of the day and are kept on for hours at a time without being cycled, and if they last 1 year, then I have likely saved >$10 in electricity, hence they have acted as a net energy saving.
Unfortunately, so much of the stuff that uneducated environmentalists (it/is/ actually possible to be an educated environmentalist, you know, and I attempt to be one) and politicians come up with are really bad for the environment. Obtaining alcohol from corn/cane sugar (never understood why Americans love getting their sugar from corn, blech!) costs far more in energy to run the harvesting/transport/refining equipment than you get out of the alcohol in the end.
If only people that had an actual influence on these things (politicians, businesses) performed triple-bottom-line analysis; unfortunately, the only people that do this currently are people who don't have to answer to share holders.
I recycle.
Again, a lot of energy. Is it really so hard these days to simply limit your consumption in the first place? Heck, I'm regretting that I may have to soon replace my 5 year old laptop, because even fvwm is getting too bloated.
Anyway, I really ought to go ride home. I feel guilty now for getting a bike with carbon forks.
I'm not actually sure what SPARC hardware is good for these days. Every time I benchmark something, it loses. Granted, our best SPARC machine is an 8-way UltraSPARC-III 1.2 GHz. So maybe a faster SPARC chip might keep up with PowerPC and Intel a little better.
I completed some benchmarking of some MPI parallised Tree-SPH code recently, and found that our Linux cluster with 2.6GHz machines was dead slow. 3 or 4 year old Alpha ES40's with 1GHz clocks and 8 times the cache per processor were 4 times faster. Hardly suprising (well, the scale of the difference was), because SPH codes need to lots of scattered memory accesses.
The only thing that came close to the alphas were the new 2Ghz opterons I can use. They basically performed the same. They only have twice the cache of the P4s, but on the other hand, they don't have a brain-dead memory architecture (for instance, I found our dual CPUs were useless, the code was completely memory bandwidth saturated, and the second CPU added *nothing* to the performance - ie, for my code, our cluster has 90 entirely wasted CPUs. Naturally, this has bad implications for our scheduler - other peoples unrelated jobs are going to affect mine, and vice versa. The opterons, of course, have a memory bus coming out of each CPU.)
Pretty simple at Swinburne (down right now for a major machine room upgrade of the electricity supply and UPS). The switch automatically detects the presence of a virus infected computer, and shuts off the port. The luser then has to go and inform ITS, and when they can prove that the machine is clean, the port is reconnected. Easy peasy.
The true solution is the simplest. Go back to paper ballots. What is your vote? Whatever you marked on the paper with a pen. No chance of anything changing your vote. Count the ballots by hand, with an observer from each party watching every vote get counted. No chance to drop any votes to swing a close result. Isn't this the true geek way? Ultimate transparency? Many eyeballs making the problem shallow?
Every time I see the way Americans vote, and how they choose the most complicated system, I laugh.
In Australia, we hand count the results. It takes about 4 hours to get the final result (although a couple of days, if it a very close election, because you have to wait for postal votes to come in, if they could add up to change the results), sometimes less if the election is not close (you only have to count until you find a winner; the rest are only needed to find out exactly how much they won by).
Why don't Americans just hire 10 times the people to count everything, since you have 10 times the population? You ought to have 10 times the money to hold elections, so this ought not be a problem for wages. Voting is incredibly simple - just mark your ballot and put it in the box.
Why do you have to take the most complcated broken system? No punch cards, no lining up holes with names, just a pen and paper. Really *really* amazingly simple!
(don't get me started on your broken "first past the post" voting method where if you have a few closely aligned but different parties, they will always lose relative to the larger more united party, even if the many smaller parties add up to even larger)
supercomputing department huh? Yeah, that alpha 500Au was the same type of machine that was under my desk 2 years ago. It was half as powerful as one of the (of the order of) 30 nodes that we also used 2-3 years ago (our cluster back then consisted of 1000AU servers -- now it is 120 dual P4's)
Anyway, look on the bright side. It's all good publicity. Paul hates advertising banners, etc, but we were debating putting up a link to the astronomy online course, given that we are just starting to take enrolments for this year:)
I'm a little surprised that this has never happened before, as they often have featured logos. I guess those fractals must have just looked too alluring, and people had to see them.
Another problem is that Paul has 2GB of graphics and movies on his page. Those fractal pages are the tip of the iceberg - Paul's pages being one of the authorative sources for a lot of graphics related things. We have banned people by IP before, because they run a wget session and try to slurp the whole fricking 2GB down the line.
So when google directs a large amount of people to a page that turns out to be interesting, and so a lot of those people end up clicking their way around a site, then the problem starts. I doubt this would have happened to anyone who's page consisted entirely of a few pretty fractals, and that's it.
There's not alot of connections to Australia but they're reasonably fast. The Southern Cross cable, for example, has three pairs each capable of 160Gbps.
The Uni might have had 1Gbps
I think we have even more than that. It wasn't the link that was the trouble. The poor server is a lowly alpha 500AU (IIRC, my ssh session is tailing the logs, but it seems I can't get bash swapped back in to run a uname -a). Our connections are fine from the department to the outside world and vice-versa.
The trouble is, this is a webserver for an astronomy and supercomputing department. Paul Bourke's page is from his research - graphics and visualisation. He was intending google to index it - it is our most popular set of pages, and is one of the authorative sources on a lot of things to do with graphics. We just weren't expecting google to link to it from their little graphic thingy. We don't want to ban google from indexing the pages, because that renders the research pointless if no-one can use it.
FYI, the server transferred 30GB in the first day of being hit by the google effect. I transferred 30GB in one day from one of the other supercomputing centres in Victoria, but this was just two processes. The poor webserver is struggling under the load of 150 apache instances right now.
I have six virtual desktops on my current single-monitor setup. I normally have at least four in use, and frequently use all six. If I had to group these on to physical monitors I would need to move to a bigger room, and would probably end up with whiplash.
I have 2 virtual desktops, (one "theory", and one "play"), and each of those is divided into a 6x4 virtual pages. These are usually full of junk, but I have sort of worked out that a particular program always goes on a particular page, (and xemacs takes up 8 different pages at a time, with position in the desktop dependant on which program is in the buffer).
I have used multiple monitor systems before (and have a laptop next to me currently, and linked by using debian's x2x - which is just the r0x0rs), and love them to death. The control computer at ATNF is a 4 headed beast, unfortunately on win NT (that crashes once a week) - the left monitor has a browser, usually open to the BOM weather page, the other 3 have about 5 different control panels and diagnostic plots.
Now, here comes the question. How does one use both a dual headed setup with a pager? I use fvwm2, and would never ever ever get rid of FvwmPager. But can it integrate with a xinerama setup? Is each of the virtual pages twice the normal size, so one virtual page spans across both monitors? Or do you have one pager per monitor?
I read something in the docs that seemed only relevant where you are not running xinerma, and instead have two separate displays -:0.0 and:0.1. I also thought I read somewhere that fvwm2 was actually xinerama aware, or being worked on.
Anyone know?
Of course, this is a moot point, because I have a 19" LCD, and there is no way in heck the department are going to buy me another flat panel, no matter what size it is (and there is not enough room on my desk for a CRT)
>>with any luck there will be another story today about the SEC suspending trading of SCOX and the FBI carting Canopy Group's board and execs off to jail >no... that wouldn't just be "any" luck... that would be poetic justice at its finest.
Mozilla's answer to this patent is that it only mentions interprocess communication. Mozilla loads the plugin into the same memory space as the browser
Goddam, no wonder mozilla with any plugins is so damn unreliable.
Did they not learn the lesson Operating Systems learnt 30 years ago? Just like they kinda threw out the unix philosophy by including everything and the kitchen sink into a browser.
Trying to propogate a virus without root is quite a bit harder than with root, so I cannot see a virus spreading so quickly.
I honestly fail to see this. Most windows virii these days are of the email variety, aren't they?
Sure, maybe the RPC exploit would be contained by the program not being able to send malicious packets unless authorized, but for simple email exploits...
You missed the point. While wiping/home would be 'unfortunate' for you, it reduces the virus' spread.
Since this article is about the spread of virii on popular systems, let's concider for the moment how most people use computers. Most people have one computer to themselves. They will set up an account for themselves, and probably their entire family uses that one account. They store a year's worth of data on it, and then a virus comes along. Now, you are saying, well, it's only limited to the one account. For most people, this is everything. The OS can be reinstalled. Everything is reproducable, *except* for the data in the user's home directory. And this is precisely the stuff the virii will delete.
Now, concider the action of spreading. What about being an unpriveleged user stops the spreading of the virii? Blocking of ports below 1024? Doesn't affect sending an email to everyone on the address book.
The guy also talks about how the lack of a dominant monoculture means virii will never spread under linux (despite the argument being that when Linux is dominant, virii still won't spread). Intel vs AMD vs alpha vs MIPS, whether the user uses mozilla or kmail. Well, condider that when Linux is popular, most people will settle on the program that gets set up by default on the default desktop, using the most popular distribution. We don't see a monoculture *today*, because most Linux users use what they prefer, not what comes by default. Oh, and of course, on an Intel box.
My ISP (internet express in regional NSW, australia) receently entered into a contract with MSN to supply search services and with altavista and google search pages only the MSN one comes up instead
Which ISP, pray tell?
If this is true, then given its illegaility, I would be contacting my friends at the ACCC over this.
no fortran compiler? did you do 0 research at all? in about two minutes i found g77 and Computation Tools 4.0, so there are probably more out there.
Oh yes, and g77 compiles fortran 90/95 code, does it?
couldnt find any speed difference in 3 days? it sounds like you gave each group a day a piece. given that your developers didnt have any experience with the platform, im not surprised you couldnt come up with anything.
You'll be surprised by the number of mac zealots in the house - particularly the pulsar group. If anyone would be able to get the best performance out of the machine, it would be them.
and as for your preference for FVWM...well...im sorry. its not 1981 anymore.
Nice troll. Show me something as configurable as fvwm? Just because it is 10 years old (not 22), doesn't mean it is any less functional than something more "modern". Just because it doesn't hog CPU and memory doing useless eye-candy shit, doesn't mean it is not as good a WM. To the contrary - it never crashes, it does exactly what I expect a WM to do, it doesn't make work harder than what it ought to be (ie, it suites my habits very well - the virtual pager is the best virtual pager any wm has come up with, most others are seriously lacking).
I expect my computer to do work, not look pretty, or like a windows box on steroids.
We tested a dual G5 for a few days (perhaps not enough) and were extrememly disappointed with it. We have 3 groups within the centre, and none of them wanted the machine after 3 days.
Except for the galaxies group, we all produce our own software.
The pulsar group (mostly using C code) found that there was no speedup using 2 threads, because of memory bandwidth, and the thing was about half the speed of our dual 2.4GHz Dell machines. The galaxies group found their imaging software was not as quick (I don't know by how much) than on a dual 2.8GHz workstation.
And the cosmology group found there was no fortran compiler. Good way to win over the scientific community. Sure, altivec would (could?) be nice, but not if we can't compile code for it.
Maybe the gcc G95 project might work eventually, but given that we can use ifc from the Intel compiler right now, we aint going to be making the move to apple for our next cluster upgrade.
Then of course, as a workstation, I used it for 20 minutes, then gave up. Give me back my FVWM window manager please - that apple OSX interface sucks (well, for a hardcore linux user, anyway).
My cat has managed to press ctrl-alt-backspace on my laptop before, killing X.
He has also managed to press alt-sysrq-t or something that dumps tasks to the console (no harm done that time - imagine if he pressed b instead, without s then u first?). And the other cat kept on pressing the power button, until I put a "molly gaurd" over the button (I had to do a similar thing to the heater, because I got home one day and discovered the heater had been on all day, because one of them had stepped on the touch sensetive switch).
Yes it is. In fact, X is liked by so many because of its network transparency.
However, the amount of data that a typical "rich" X client sends (e.g., mozilla) is huge.
Rich? I don't really understand why. The mozilla screen I'm looking at is not as complex as say the screen that has my simulations on it.
Many X clients are not optimized in terms of the amount of display information they output (that is, they output a lot of stuff that could probably be optimized away).
I would more say most applications these days are pessimised. Have you seen gtk2 lately? How fscking hard is it to open up a bloody menu? Why does it take 20 round trips?
For many developers, this is within reason since they figure that most of the time the xserver and xclient will be on the same machine (e.g., running mozilla on my box to display on my monitor). For a lot of people, this has never and will never be the case. That is why we use and like X. I would say the problem comes down more to sheer lazyness of programmers of modern software.
I mean, gtk2 just blows goats balls. I wish programmers would spend more time thinking, instead of waving their dicks around in the air and getting their latest 3GHz dual P4 with 4GB or RAM.
Lazy programmers have always bugged the heck out of me - just think of all those fools who never bother to make sure their arrays are big enough, and keep using strcpy().
The lower bumpers are to protect other cars... SUVs have a nasty tendency to ride over other cars in a collision and squish them. (A friend of mine had half his Japanese import squashed in a collision with an SUV.) Rollovers are typically single-vehicle accidents, caused by cornering too fast. SUVs are notorious for having their center of gravity too high.
And notorious for having drivers too stupid to slow down to an appropriate speed too.
Slashdot Mirror
I use energy efficient lightbulbs.
/is/ actually possible to be an educated environmentalist, you know, and I attempt to be one) and politicians come up with are really bad for the environment. Obtaining alcohol from corn/cane sugar (never understood why Americans love getting their sugar from corn, blech!) costs far more in energy to run the harvesting/transport/refining equipment than you get out of the alcohol in the end.
Do you know why they are 20 times more expensive than normal bulbs? Because they take approximately 20 times more resources to make (the lighting field is highly competitive, so the cost to you basically reflects the cost to manufacture them). So if you don't save more than the $10 extra to manufacture one energy efficient bulb, over its lifetime, in saved electricity, then you have done more harm than good.
I bought a whole bunch of energy efficient bulbs. Most of them died within a year because they don't like dirty electricity and being cycled rapidly -- shorter than the average lifetime I get out of normal bulbs (despite the marketing blurb explaining that they last 8 times longer). The only energy efficient bulbs I have retained are the ones in the living/lounge room -- ie, the ones that are on for a substantial part of the day and are kept on for hours at a time without being cycled, and if they last 1 year, then I have likely saved >$10 in electricity, hence they have acted as a net energy saving.
Unfortunately, so much of the stuff that uneducated environmentalists (it
If only people that had an actual influence on these things (politicians, businesses) performed triple-bottom-line analysis; unfortunately, the only people that do this currently are people who don't have to answer to share holders.
I recycle.
Again, a lot of energy. Is it really so hard these days to simply limit your consumption in the first place? Heck, I'm regretting that I may have to soon replace my 5 year old laptop, because even fvwm is getting too bloated.
Anyway, I really ought to go ride home. I feel guilty now for getting a bike with carbon forks.
I'm not actually sure what SPARC hardware is good for these days. Every time I benchmark something, it loses. Granted, our best SPARC machine is an 8-way UltraSPARC-III 1.2 GHz. So maybe a faster SPARC chip might keep up with PowerPC and Intel a little better.
I completed some benchmarking of some MPI parallised Tree-SPH code recently, and found that our Linux cluster with 2.6GHz machines was dead slow. 3 or 4 year old Alpha ES40's with 1GHz clocks and 8 times the cache per processor were 4 times faster. Hardly suprising (well, the scale of the difference was), because SPH codes need to lots of scattered memory accesses.
The only thing that came close to the alphas were the new 2Ghz opterons I can use. They basically performed the same. They only have twice the cache of the P4s, but on the other hand, they don't have a brain-dead memory architecture (for instance, I found our dual CPUs were useless, the code was completely memory bandwidth saturated, and the second CPU added *nothing* to the performance - ie, for my code, our cluster has 90 entirely wasted CPUs. Naturally, this has bad implications for our scheduler - other peoples unrelated jobs are going to affect mine, and vice versa. The opterons, of course, have a memory bus coming out of each CPU.)
Pretty simple at Swinburne (down right now for a major machine room upgrade of the electricity supply and UPS). The switch automatically detects the presence of a virus infected computer, and shuts off the port. The luser then has to go and inform ITS, and when they can prove that the machine is clean, the port is reconnected. Easy peasy.
The true solution is the simplest. Go back to paper ballots. What is your vote? Whatever you marked on the paper with a pen. No chance of anything changing your vote. Count the ballots by hand, with an observer from each party watching every vote get counted. No chance to drop any votes to swing a close result. Isn't this the true geek way? Ultimate transparency? Many eyeballs making the problem shallow?
Every time I see the way Americans vote, and how they choose the most complicated system, I laugh.
In Australia, we hand count the results. It takes about 4 hours to get the final result (although a couple of days, if it a very close election, because you have to wait for postal votes to come in, if they could add up to change the results), sometimes less if the election is not close (you only have to count until you find a winner; the rest are only needed to find out exactly how much they won by).
Why don't Americans just hire 10 times the people to count everything, since you have 10 times the population? You ought to have 10 times the money to hold elections, so this ought not be a problem for wages. Voting is incredibly simple - just mark your ballot and put it in the box.
Why do you have to take the most complcated broken system? No punch cards, no lining up holes with names, just a pen and paper. Really *really* amazingly simple!
(don't get me started on your broken "first past the post" voting method where if you have a few closely aligned but different parties, they will always lose relative to the larger more united party, even if the many smaller parties add up to even larger)
supercomputing department huh?
:)
Yeah, that alpha 500Au was the same type of machine that was under my desk 2 years ago. It was half as powerful as one of the (of the order of) 30 nodes that we also used 2-3 years ago (our cluster back then consisted of 1000AU servers -- now it is 120 dual P4's)
Anyway, look on the bright side. It's all good publicity.
Paul hates advertising banners, etc, but we were debating putting up a link to the astronomy online course, given that we are just starting to take enrolments for this year
I'm a little surprised that this has never happened before, as they often have featured logos. I guess those fractals must have just looked too alluring, and people had to see them.
Another problem is that Paul has 2GB of graphics and movies on his page. Those fractal pages are the tip of the iceberg - Paul's pages being one of the authorative sources for a lot of graphics related things. We have banned people by IP before, because they run a wget session and try to slurp the whole fricking 2GB down the line.
So when google directs a large amount of people to a page that turns out to be interesting, and so a lot of those people end up clicking their way around a site, then the problem starts. I doubt this would have happened to anyone who's page consisted entirely of a few pretty fractals, and that's it.
There's not alot of connections to Australia but they're reasonably fast. The Southern Cross cable, for example, has three pairs each capable of 160Gbps.
The Uni might have had 1Gbps
I think we have even more than that. It wasn't the link that was the trouble. The poor server is a lowly alpha 500AU (IIRC, my ssh session is tailing the logs, but it seems I can't get bash swapped back in to run a uname -a). Our connections are fine from the department to the outside world and vice-versa.
The trouble is, this is a webserver for an astronomy and supercomputing department. Paul Bourke's page is from his research - graphics and visualisation. He was intending google to index it - it is our most popular set of pages, and is one of the authorative sources on a lot of things to do with graphics. We just weren't expecting google to link to it from their little graphic thingy. We don't want to ban google from indexing the pages, because that renders the research pointless if no-one can use it.
FYI, the server transferred 30GB in the first day of being hit by the google effect. I transferred 30GB in one day from one of the other supercomputing centres in Victoria, but this was just two processes. The poor webserver is struggling under the load of 150 apache instances right now.
I have six virtual desktops on my current single-monitor setup. I normally have at least four in use, and frequently use all six. If I had to group these on to physical monitors I would need to move to a bigger room, and would probably end up with whiplash.
:0.0 and :0.1. I also thought I read somewhere that fvwm2 was actually xinerama aware, or being worked on.
I have 2 virtual desktops, (one "theory", and one "play"), and each of those is divided into a 6x4 virtual pages. These are usually full of junk, but I have sort of worked out that a particular program always goes on a particular page, (and xemacs takes up 8 different pages at a time, with position in the desktop dependant on which program is in the buffer).
I have used multiple monitor systems before (and have a laptop next to me currently, and linked by using debian's x2x - which is just the r0x0rs), and love them to death. The control computer at ATNF is a 4 headed beast, unfortunately on win NT (that crashes once a week) - the left monitor has a browser, usually open to the BOM weather page, the other 3 have about 5 different control panels and diagnostic plots.
Now, here comes the question. How does one use both a dual headed setup with a pager? I use fvwm2, and would never ever ever get rid of FvwmPager. But can it integrate with a xinerama setup? Is each of the virtual pages twice the normal size, so one virtual page spans across both monitors? Or do you have one pager per monitor?
I read something in the docs that seemed only relevant where you are not running xinerma, and instead have two separate displays -
Anyone know?
Of course, this is a moot point, because I have a 19" LCD, and there is no way in heck the department are going to buy me another flat panel, no matter what size it is (and there is not enough room on my desk for a CRT)
>>with any luck there will be another story today about the SEC suspending trading of SCOX and the FBI carting Canopy Group's board and execs off to jail
>no... that wouldn't just be "any" luck... that would be poetic justice at its finest.
Not "poetic justice" -- just "justice"
Mozilla's answer to this patent is that it only mentions interprocess communication. Mozilla loads the plugin into the same memory space as the browser
Goddam, no wonder mozilla with any plugins is so damn unreliable.
Did they not learn the lesson Operating Systems learnt 30 years ago? Just like they kinda threw out the unix philosophy by including everything and the kitchen sink into a browser.
Trying to propogate a virus without root is quite a bit harder than with root, so I cannot see a virus spreading so quickly.
I honestly fail to see this. Most windows virii these days are of the email variety, aren't they?
Sure, maybe the RPC exploit would be contained by the program not being able to send malicious packets unless authorized, but for simple email exploits...
You missed the point. While wiping /home would be 'unfortunate' for you, it reduces the virus' spread.
Since this article is about the spread of virii on popular systems, let's concider for the moment how most people use computers. Most people have one computer to themselves. They will set up an account for themselves, and probably their entire family uses that one account. They store a year's worth of data on it, and then a virus comes along. Now, you are saying, well, it's only limited to the one account. For most people, this is everything. The OS can be reinstalled. Everything is reproducable, *except* for the data in the user's home directory. And this is precisely the stuff the virii will delete.
Now, concider the action of spreading. What about being an unpriveleged user stops the spreading of the virii? Blocking of ports below 1024? Doesn't affect sending an email to everyone on the address book.
The guy also talks about how the lack of a dominant monoculture means virii will never spread under linux (despite the argument being that when Linux is dominant, virii still won't spread). Intel vs AMD vs alpha vs MIPS, whether the user uses mozilla or kmail. Well, condider that when Linux is popular, most people will settle on the program that gets set up by default on the default desktop, using the most popular distribution. We don't see a monoculture *today*, because most Linux users use what they prefer, not what comes by default. Oh, and of course, on an Intel box.
>>> My ISP (internet express in regional NSW, australia) ...
>> Which ISP, pray tell?
>Not too quick on the uptake, are you?
Yup, pretty slow. Wups.
Skipped right over the name.
My ISP (internet express in regional NSW, australia) receently entered into a contract with MSN to supply search services and with altavista and google search pages only the MSN one comes up instead
Which ISP, pray tell?
If this is true, then given its illegaility, I would be contacting my friends at the ACCC over this.
no fortran compiler? did you do 0 research at all? in about two minutes i found g77 and Computation Tools 4.0, so there are probably more out there.
Oh yes, and g77 compiles fortran 90/95 code, does it?
couldnt find any speed difference in 3 days? it sounds like you gave each group a day a piece. given that your developers didnt have any experience with the platform, im not surprised you couldnt come up with anything.
You'll be surprised by the number of mac zealots in the house - particularly the pulsar group. If anyone would be able to get the best performance out of the machine, it would be them.
and as for your preference for FVWM...well...im sorry. its not 1981 anymore.
Nice troll.
Show me something as configurable as fvwm? Just because it is 10 years old (not 22), doesn't mean it is any less functional than something more "modern". Just because it doesn't hog CPU and memory doing useless eye-candy shit, doesn't mean it is not as good a WM. To the contrary - it never crashes, it does exactly what I expect a WM to do, it doesn't make work harder than what it ought to be (ie, it suites my habits very well - the virtual pager is the best virtual pager any wm has come up with, most others are seriously lacking).
I expect my computer to do work, not look pretty, or like a windows box on steroids.
We tested a dual G5 for a few days (perhaps not enough) and were extrememly disappointed with it. We have 3 groups within the centre, and none of them wanted the machine after 3 days.
Except for the galaxies group, we all produce our own software.
The pulsar group (mostly using C code) found that there was no speedup using 2 threads, because of memory bandwidth, and the thing was about half the speed of our dual 2.4GHz Dell machines.
The galaxies group found their imaging software was not as quick (I don't know by how much) than on a dual 2.8GHz workstation.
And the cosmology group found there was no fortran compiler. Good way to win over the scientific community. Sure, altivec would (could?) be nice, but not if we can't compile code for it.
Maybe the gcc G95 project might work eventually, but given that we can use ifc from the Intel compiler right now, we aint going to be making the move to apple for our next cluster upgrade.
Then of course, as a workstation, I used it for 20 minutes, then gave up. Give me back my FVWM window manager please - that apple OSX interface sucks (well, for a hardcore linux user, anyway).
Oh yes they can be struck by accident!
My cat has managed to press ctrl-alt-backspace on my laptop before, killing X.
He has also managed to press alt-sysrq-t or something that dumps tasks to the console (no harm done that time - imagine if he pressed b instead, without s then u first?). And the other cat kept on pressing the power button, until I put a "molly gaurd" over the button (I had to do a similar thing to the heater, because I got home one day and discovered the heater had been on all day, because one of them had stepped on the touch sensetive switch).
Hmmm, searching for help on LaTeX can sometimes be... distracting.
Yes it is. In fact, X is liked by so many because of its network transparency.
However, the amount of data that a typical "rich" X client sends (e.g., mozilla) is huge.
Rich? I don't really understand why. The mozilla screen I'm looking at is not as complex as say the screen that has my simulations on it.
Many X clients are not optimized in terms of the amount of display information they output (that is, they output a lot of stuff that could probably be optimized away).
I would more say most applications these days are pessimised. Have you seen gtk2 lately? How fscking hard is it to open up a bloody menu? Why does it take 20 round trips?
For many developers, this is within reason since they figure that most of the time the xserver and xclient will be on the same machine (e.g., running mozilla on my box to display on my monitor).
For a lot of people, this has never and will never be the case. That is why we use and like X. I would say the problem comes down more to sheer lazyness of programmers of modern software.
I mean, gtk2 just blows goats balls. I wish programmers would spend more time thinking, instead of waving their dicks around in the air and getting their latest 3GHz dual P4 with 4GB or RAM.
Lazy programmers have always bugged the heck out of me - just think of all those fools who never bother to make sure their arrays are big enough, and keep using strcpy().
Yeah, I know, I get what I pay for.
So, it looks like his job was @stake?
Sigh.
The lower bumpers are to protect other cars ... SUVs have a nasty tendency to ride over other cars in a collision and squish them. (A friend of mine had half his Japanese import squashed in a collision with an SUV.) Rollovers are typically single-vehicle accidents, caused by cornering too fast. SUVs are notorious for having their center of gravity too high.
And notorious for having drivers too stupid to slow down to an appropriate speed too.
Why else would they get one?
Of course, being human, my opinion is suspect.
Haw haw! Because, I am not human, and hence my opinion is inherently superior to yours!
What next? An LED that displays hard drive activity?
We have 16TB of the apple X-raids. The best feature about them? The blinkenlights. Blue ones. Very hipnotising.
2 bar graph displays for activity (bi-colour?), plus a bi colour LED per drive (blue for something/green for something else)
Very *drool*worthy.
Or perhaps one for network activity!
Sigh, I miss those old Sun keyboards.
I did install tleds on my laptop, but whenever the keyboard LED light up, it interferes with DPMS on the display.
My desktop doesn't have LEDs, because it is a wireless (why can't wireless keyboards have an LCD display - they don't drain the battery much).
Why are all Melbourne scientists dj's?
Alex deLarge is an astronomer here, and would be DJing tonight, if he didn't get hit by a car on his scooter last week.