If your current Sun servers are handling the load properly, don't mess with them. They're probably upgradable to bigger disks & more RAM, if they're getting worked, but for mail/nis/etc you don't really need a lot of horsepower for only a few dozen people and don't forget the immeasurable bonus of having it actually working.
As for the desktops, if you're careful, you can -stay- with Solaris _AND_ switch to fast, cheap, x86 hardware for the workstations. You might be stuck with Linux on the Itanium compute server (which is only really going to be useful if you get >4GB RAM...), but you can keep the desktops virtually the same (assuming your software has Solarix86 support).
If you're not really a 'qualified' admin, I'd try to change as little as possible. Doing LFS for a handful of compute servers is pointless; Take slackware or debian, do a custom kernel compile, remove some unneeded packages and services and then recompile a few key apps with excessive optimizations. You'll save yourself a load of time, have a system that actually works right and the engineers won't notice the difference. It might be different if you were building a cluster or something, but it's not worth your (or the company's) time in this situation.
A large percentage of the masters students in the CS department where I'm at are from India. Many of them are very bright, but they don't really know much about computers, for many of them, computer time was limited and they didn't get much real screen time until they were well into the program. This is in distinct contrast to USians who have grown up with computers and often have an innate understanding of the -feel- of computing.
While many of the Indians have a great grasp of the more theoretical aspects, they lack the 'street smarts' of how to actually -use- a computer.
I can't believe that anyone responsible for ghosting 30 computers didn't do an install from scratch. There's too much crap that gets included with an install from Dell to even bother cleaning out; a fresh install with your non-OEM-altered copies of the OS is an obvious thing to do.
Yes, Itanic sucked, but the second generation Itanium 2 chips, even at the slower clock rates, turn out some of the best floating-point benchmarks of ANY other single chip on the market, by a significant margin.
Don't forget that OSX ships with a a vector-capable version of gcc and VecLib, a collection of Altivec optimized routines (all kinds of juicy number-crunching things like FFTs & linear-algebra bits)
One thing you overlook is the performance improvement that vector-ops can provide. To start with, Intel eviscerated the vector-processing (MMX/SSE/SSE2) capabilities of the P4 compared to the P3; limiting the number of execution units it can run on and various other things. The G4 (and presumable the PPC970) has the Altivec unit which provides significantly more vector registers, operates on vectors double the size, and is capable of running vector operatons on multiple execution units.
The end result is that the PPC chip is capable of doing heavy number crunching (think photoshop & scientific apps, rather than updating spreadsheets & getting higher framerates in games) more effectively when it counts. How much the increased efficiency would help is highly dependant on the specific application (and the effort the developer puts into optimization).
Combine this with the lower power consumption and you could see blade systems packing the new PPC become a viable choice for compute clusters.
IE -
Imagine a Beowulf of these things...
Re:Huge waste of money
on
RAMdisk RAID?
·
· Score: 1
Start out w/ the fastest CPU you can afford, then get the motherboard that supports the fastest RAM speed for that CPU, then max it out with as much of that fast RAM as you can put on the board. Get yourself some large, fast HDDs and put them in a RAID-0 (don't forget to do backups!). Don't forget the 21" monitor.
Wham. Find a case & a midrange ($50-75) video-card and you've got a good photoshop machine. Remember, Photoshop is essentially is essentially a number-crunching program, and there's not much difference in the 2D performance of the graphics cards available on the market. Picking an optimal system for gaming is more dificult for 2 reasons: First, gaming performance is the result of the combination of CPU power and GPU power; a balance must be kept between the extremes of having a fast-ass video-card that's starved for data by a slow CPU and having a CPU that can compute infinite loops in 10 clock cycles that's not allowed to show things at more than 10FPS 'cuz you've got a crap video-card. The other thing is that for a majority of serious Photoshop users the system upgrade is going to be a business expense that they can write-off come tax time, where gamers are doing this for personal entertainment.
Competitive pricing, great selection & fast, reliable service. What more could you ask for?
That system the parent posted (gf4mx440, k7s5a) came to $206 w/ an XP2100+, $188 w/ an XP2000+ and $174 for an XP1800+; all with free shipping.
Definately a good start for a nice, quick budget gaming machine. Being the owner of a K7S5a myself, the only qualms I have with the board is the onboard sound, but that's a minor issue and a cheap upgrade (newegg's got SB-Live5.1 digitals for $30 or so).
This is great assuming that you're told about these things before taking the job. By the time you've already agreed to take the job and probably informed other potential employers that you were off the market, you're kinda SOL. By dropping these things on you at the last minute, they take away your bargaining leverage.
However, on the 'net it is _FAR_ easier to obtain power than it is elsewhere. Raising an army of thousands of zombies to stop all trafic in a major city isn't terribly likely for the average pimply-faced highschooler skript-kiddie still living in their parents' basement, but on the 'net, it's not too difficult.
OTOH, a number of ATX motherboards I've worked on recently don't use the far-right (towards the CPU) slot, putting the AGP in the 2nd slot. In cases like this, one could not only have a 2-slot cooling system, but have a convenient exhaust vent attached to the card-cage, if they were to use the extra space behind the card.
Considering that even good motherboards barely break the $150 mark, while high-end GPUs can be $400+, it doesn't make much sense to make the GPU fit the mobo, when you can find a mobo to work with your GPU of choice.
Actually, BSD -was- encumbered by AT&T patents, hence the stripped-down, non-functional Net/2 release of BSD (the last one from Berkeley). The unencumbered Net/2 was where the free *BSDs started from, replacing the missing pieces with Free code.
yes, it most definately is a special case. You obviously have never seen actual popup spam, as it's unlikely that anyone would bother spaming from inside of the school. Commercial popup spamming involves sending messages from OUTSIDE of your local network to you which contain adverts. To make matters worse, they're not easy to trace, as the sender is only identified by the WINS name, not IP, which you, not being on the same local network as them, can't look up.
Reminds me of an old machine I put together for my sister so she could learn Linux. We called it Gimpy.
Gimpy started life as a 486DX-25, one of Dell's better models at the time; He lived in a FULL-tower case, with a large, expandable cache memory card and expandable onboard VGA controller (2MB). Gimpy had all the available bells & whistles that were available at the time of his birth, including a comprehensive self-diagnostic tool, built into the BIOS.
I got Gimpy as scrap from school. Primarily he was given away 'cuz the IDE controller on his motherboard was fried; the self-diagnostics would paste a big, red FAILED and lock up immediately after starting to test them. Second, the not-entirely-standard sized AT power-supply was toast....not to mention that this was some time in 2000, and a 486 was considered beyond slow.
Fortunately, being friends w/ the tech guy, I got all the goodies I could find that went along w/ Gimpy; enough RAM to bump him up to 24MB (!!!), a 486DX2-66, and a power-supply from an dead 386. The power supply was one that looked like it came from the original PC, you know... with the Big Red Switch on the back corner of the machine.
The first step in bringing gimpy back to life was to get the juice flowing. With the sheetmetal panel enclosing the case removed, the PSU fit without a problem. Unfortunately, trying ot close up the case ended up covering the big red switch that turned the machine on (the orignal PSU had a cable going from a switch on the front-panel to the PSU; remember, this was pre-ATX soft-power). The obvious solution was to take a pair of tin snips & cut a hole in the case, resulting in a tower with a big red switch sticking out of it; classic ghetto-tech.
Step 2 in bringing the beast back to life was getting some HDDs attached. Being a collector of old PCs & components, I dug through my pile of spare parts, and found a pair of 386 'servers' that had disc controllers & decently sized HDDs. The first was an ESDI controller with a 500MB-ish drive attached and this cool bank of 8 diagnostic LEDs that did this Night Rider pulsing thing during normal operation. The other option was a full-lenght, 16-bit ISA slow-narrow SCSI-1 card with the various SCSI drives I'd collected for it (a 300MB Quantum, a 500 from a microVAX-II, and an 80MB Quantum stripped out of a Mac).
Purely for capacity reasons, I went with the SCSI.
After this point, things were fairly simple. Toss a pair of floppies (3.5" and 5.25") on it & give it a NIC (a real NE2000, which got hooked up to my coax ethernet network), give it a monitor (IBM 8514; those things are tanks, I've had 2 and they NEVER die or go bad, you just get annoyed at the tiny screen). I then threw slackware onto it, and told my sister how to login & use man....eventually Gimpy got retired when my sister got a 'real' PC with enough power to run win2k, but she happily got through a year of APCS using it. Now he sits dormant as a reminder that you -can- get by with less than 100MHZ & use free hardware for useful jobs. All it takes is a little knowlege & a little creativity.
...not to mention that most mobo chipsets, these days, have various integrated stuff already included in them. Asking to get a board w/o integrated sound & NIC is almost like trying to buy a car without a radio.
Slashdot Mirror
What happens when there's nowhere 'safe' left on the planet?
If you've got dual 2.2G Xeons for your 'play machine', what's your main box look like?
If your current Sun servers are handling the load properly, don't mess with them. They're probably upgradable to bigger disks & more RAM, if they're getting worked, but for mail/nis/etc you don't really need a lot of horsepower for only a few dozen people and don't forget the immeasurable bonus of having it actually working.
As for the desktops, if you're careful, you can -stay- with Solaris _AND_ switch to fast, cheap, x86 hardware for the workstations. You might be stuck with Linux on the Itanium compute server (which is only really going to be useful if you get >4GB RAM...), but you can keep the desktops virtually the same (assuming your software has Solarix86 support).
If you're not really a 'qualified' admin, I'd try to change as little as possible. Doing LFS for a handful of compute servers is pointless; Take slackware or debian, do a custom kernel compile, remove some unneeded packages and services and then recompile a few key apps with excessive optimizations. You'll save yourself a load of time, have a system that actually works right and the engineers won't notice the difference. It might be different if you were building a cluster or something, but it's not worth your (or the company's) time in this situation.
A large percentage of the masters students in the CS department where I'm at are from India. Many of them are very bright, but they don't really know much about computers, for many of them, computer time was limited and they didn't get much real screen time until they were well into the program. This is in distinct contrast to USians who have grown up with computers and often have an innate understanding of the -feel- of computing.
While many of the Indians have a great grasp of the more theoretical aspects, they lack the 'street smarts' of how to actually -use- a computer.
Perhaps the situation of gov't coders in Snow Crash (such as YT's mom) would be a more appropriate alusion to make.
I can't believe that anyone responsible for ghosting 30 computers didn't do an install from scratch. There's too much crap that gets included with an install from Dell to even bother cleaning out; a fresh install with your non-OEM-altered copies of the OS is an obvious thing to do.
Yes, Itanic sucked, but the second generation Itanium 2 chips, even at the slower clock rates, turn out some of the best floating-point benchmarks of ANY other single chip on the market, by a significant margin.
Don't forget that OSX ships with a a vector-capable version of gcc and VecLib, a collection of Altivec optimized routines (all kinds of juicy number-crunching things like FFTs & linear-algebra bits)
One thing you overlook is the performance improvement that vector-ops can provide. To start with, Intel eviscerated the vector-processing (MMX/SSE/SSE2) capabilities of the P4 compared to the P3; limiting the number of execution units it can run on and various other things. The G4 (and presumable the PPC970) has the Altivec unit which provides significantly more vector registers, operates on vectors double the size, and is capable of running vector operatons on multiple execution units.
The end result is that the PPC chip is capable of doing heavy number crunching (think photoshop & scientific apps, rather than updating spreadsheets & getting higher framerates in games) more effectively when it counts. How much the increased efficiency would help is highly dependant on the specific application (and the effort the developer puts into optimization).
Combine this with the lower power consumption and you could see blade systems packing the new PPC become a viable choice for compute clusters.
IE -
Imagine a Beowulf of these things...
time to upgrade to 64b 66MHz pci gear.
Two problems with that.
A) try calling it Theta.
B) your trig assumes that you've got a square screen. Most computer displays use non-square pixels.
It's simple...
Start out w/ the fastest CPU you can afford, then get the motherboard that supports the fastest RAM speed for that CPU, then max it out with as much of that fast RAM as you can put on the board. Get yourself some large, fast HDDs and put them in a RAID-0 (don't forget to do backups!). Don't forget the 21" monitor.
Wham. Find a case & a midrange ($50-75) video-card and you've got a good photoshop machine. Remember, Photoshop is essentially is essentially a number-crunching program, and there's not much difference in the 2D performance of the graphics cards available on the market. Picking an optimal system for gaming is more dificult for 2 reasons: First, gaming performance is the result of the combination of CPU power and GPU power; a balance must be kept between the extremes of having a fast-ass video-card that's starved for data by a slow CPU and having a CPU that can compute infinite loops in 10 clock cycles that's not allowed to show things at more than 10FPS 'cuz you've got a crap video-card. The other thing is that for a majority of serious Photoshop users the system upgrade is going to be a business expense that they can write-off come tax time, where gamers are doing this for personal entertainment.
new egg
Competitive pricing, great selection & fast, reliable service. What more could you ask for?
That system the parent posted (gf4mx440, k7s5a) came to $206 w/ an XP2100+, $188 w/ an XP2000+ and $174 for an XP1800+; all with free shipping.
Definately a good start for a nice, quick budget gaming machine. Being the owner of a K7S5a myself, the only qualms I have with the board is the onboard sound, but that's a minor issue and a cheap upgrade (newegg's got SB-Live5.1 digitals for $30 or so).
This is great assuming that you're told about these things before taking the job. By the time you've already agreed to take the job and probably informed other potential employers that you were off the market, you're kinda SOL. By dropping these things on you at the last minute, they take away your bargaining leverage.
If you can move all of the core developers to Canda, you could call it GNU/X
IE :
The plural of anecdote is not data.
However, on the 'net it is _FAR_ easier to obtain power than it is elsewhere. Raising an army of thousands of zombies to stop all trafic in a major city isn't terribly likely for the average pimply-faced highschooler skript-kiddie still living in their parents' basement, but on the 'net, it's not too difficult.
according to http://irc.netsplit.de/networks/, the most populated warez/mp3/vcd channels are all on EFNet. Why would they mess with DALnet?
OTOH, a number of ATX motherboards I've worked on recently don't use the far-right (towards the CPU) slot, putting the AGP in the 2nd slot. In cases like this, one could not only have a 2-slot cooling system, but have a convenient exhaust vent attached to the card-cage, if they were to use the extra space behind the card.
Considering that even good motherboards barely break the $150 mark, while high-end GPUs can be $400+, it doesn't make much sense to make the GPU fit the mobo, when you can find a mobo to work with your GPU of choice.
I just hope it doesn't have access times measured in fortnights.
50 karma IS a license to troll ...as is evidenced by the countless bootleg posts you've been making on this story.
Actually, BSD -was- encumbered by AT&T patents, hence the stripped-down, non-functional Net/2 release of BSD (the last one from Berkeley). The unencumbered Net/2 was where the free *BSDs started from, replacing the missing pieces with Free code.
yes, it most definately is a special case. You obviously have never seen actual popup spam, as it's unlikely that anyone would bother spaming from inside of the school. Commercial popup spamming involves sending messages from OUTSIDE of your local network to you which contain adverts. To make matters worse, they're not easy to trace, as the sender is only identified by the WINS name, not IP, which you, not being on the same local network as them, can't look up.
Reminds me of an old machine I put together for my sister so she could learn Linux. We called it Gimpy.
...not to mention that this was some time in 2000, and a 486 was considered beyond slow.
...eventually Gimpy got retired when my sister got a 'real' PC with enough power to run win2k, but she happily got through a year of APCS using it. Now he sits dormant as a reminder that you -can- get by with less than 100MHZ & use free hardware for useful jobs. All it takes is a little knowlege & a little creativity.
Gimpy started life as a 486DX-25, one of Dell's better models at the time; He lived in a FULL-tower case, with a large, expandable cache memory card and expandable onboard VGA controller (2MB). Gimpy had all the available bells & whistles that were available at the time of his birth, including a comprehensive self-diagnostic tool, built into the BIOS.
I got Gimpy as scrap from school. Primarily he was given away 'cuz the IDE controller on his motherboard was fried; the self-diagnostics would paste a big, red FAILED and lock up immediately after starting to test them. Second, the not-entirely-standard sized AT power-supply was toast.
Fortunately, being friends w/ the tech guy, I got all the goodies I could find that went along w/ Gimpy; enough RAM to bump him up to 24MB (!!!), a 486DX2-66, and a power-supply from an dead 386. The power supply was one that looked like it came from the original PC, you know... with the Big Red Switch on the back corner of the machine.
The first step in bringing gimpy back to life was to get the juice flowing. With the sheetmetal panel enclosing the case removed, the PSU fit without a problem. Unfortunately, trying ot close up the case ended up covering the big red switch that turned the machine on (the orignal PSU had a cable going from a switch on the front-panel to the PSU; remember, this was pre-ATX soft-power). The obvious solution was to take a pair of tin snips & cut a hole in the case, resulting in a tower with a big red switch sticking out of it; classic ghetto-tech.
Step 2 in bringing the beast back to life was getting some HDDs attached. Being a collector of old PCs & components, I dug through my pile of spare parts, and found a pair of 386 'servers' that had disc controllers & decently sized HDDs. The first was an ESDI controller with a 500MB-ish drive attached and this cool bank of 8 diagnostic LEDs that did this Night Rider pulsing thing during normal operation. The other option was a full-lenght, 16-bit ISA slow-narrow SCSI-1 card with the various SCSI drives I'd collected for it (a 300MB Quantum, a 500 from a microVAX-II, and an 80MB Quantum stripped out of a Mac).
Purely for capacity reasons, I went with the SCSI.
After this point, things were fairly simple. Toss a pair of floppies (3.5" and 5.25") on it & give it a NIC (a real NE2000, which got hooked up to my coax ethernet network), give it a monitor (IBM 8514; those things are tanks, I've had 2 and they NEVER die or go bad, you just get annoyed at the tiny screen). I then threw slackware onto it, and told my sister how to login & use man.
...not to mention that most mobo chipsets, these days, have various integrated stuff already included in them. Asking to get a board w/o integrated sound & NIC is almost like trying to buy a car without a radio.