I have done some driver work, build and debuging and linux is a driver writers dream compared to windows. I would estimate that a linux driver will consume about 50% as much time as a windows drive writing either from scratch. I reality, you may have a lot or reusable code from the windows drivers so you could be looking at less. Also do take the advice of other posts and get the basic driver out ASAP, and allow the OS community help out on the rest.
Americans are disgusted with the idea of a foreign nation marching on them moreso than ANY OTHER COUNTRY ANYWHERE. If Canada looked at us funny, we would are and be ready! This ideology requires citizens to have guns, to feel safe. Also, the "only criminals have guns" idea is very sound here. Being the melting pot of the world, the worlds largest populous nation and by far the most imigrated to nation.
Now imagine keeping all of those people in check. Can you see why Americans like their guns soooo much?
It would also be impossible to take guns from Americans, our constitution guarantees our ability to own guns for personal protection. Americans would see that as the government breaking some holy vow or something and you'd have armies of people fighting the government.
---
now, jumping the fence a little, i despise guns. I think its rediculous that any civilized society NEEDS GUNS! killing other people should never be an issue. A civilized country should have vitually no crime that a civilian needs a gun to protect themselves from. People should both fear and respect the law as if it were ordered by God/a god/The One god/Buddah said it/Mighty Thor demands it, whatever....
I have spent enough time thinking about all the wrongs that "civilized" nations do everyday and know that Guns have no positive benefit to the masses.
how about the money that you spend on your gun, instead gets spent on a starving child somewhere. and the money governments spend to keep guns in check is also spent on that. How about Europeans that have nothing better to do that say Americans are gun toteing yanks spend a little less on other indulgences like inconceivably high consumption of alcohol and cigarettes and other various drugs that promote crime and the need for guns, and do a little charity work instead of "talking about policing the world, then letting America do the work, and the bitch about America trying to police the world"(where did that come from!) Guns rate right up their with Sadam Hussien, Nukes, Bio Weapons, and slavery. Almost as high as Capitol(hypocritical) punishment.(punish someone for killing someone, and then go and kill someone to prove how wrong it is)
yeah, i think my rant is over now. hell, i didn't like karma anyway, just slavey by a different name!
of course you could by one copy of photoshop and use it anywhere in you company.
You could only have it running on a SINGLE machine at any one time and you would problably have to make some sort of effort to enforce this. Adobe may have a clause in the licensing aggreement that states the software may only be installed on one machine, then in fact you would have license to a single copy of the software and could use it on as many as you wished as long as it was installed on just one.
This could be done via mapped drives or something like that, assuming photoshop would run without 10,000 registry additions and dll in system32/....
i think that it has a good chance of happening to some degree. Machines will most likely make the products cutting a good amount of hours from human workers, but humans wil problably still work the counter. People will rather talk to a human that a machine in public places.
I also dont think that this will be as devastating to the economy as some do. I see that by 2050, as many people(by %) will be employed but hours will change. a full work day will likely be just 3 or less hours. with people working just 1/3 the amount of time, 3 times as many people are needed to fill a work day, keeping the unemployment rates down.
also keep in mind that machines are extremely cheap to have do things, and this trend will likely continue. some examples: in a fast food place like Wendy's, human labor cosists of about 17% of the gross profits. This could be reduced to 2-3% and maybee another 3% for maintainance and energy costs. shipping gets cheaper having machines do the work also. All this means that cost is less so price can be less makeing the situation even better for people that only work 3 hours a day.
of course america will be at the front of the line in implementing this so that we can get more fat and lazy. This in turn will counteract the fact that people should live longer because of less work related abuse to their bodies.
i keep most images in TIFF. Its big but easy to work with, i dont spend cpu cycles on decompressing bits to work with them. Lossless compression is nice, but tar.gz and rar are really great lossless compressors:) . I can't wait til a stable filesystem that can mount compressed formats:)
I know a bit about X, and I know a bit about DirectFB(just research, not code knowledge)
i know that i would like to see KDE and QT compatible with both directfb AND X windows. X is extremely usefull and will evolve very nicely as a lot of legacy support is removed, but it just can't be as fast as directfb can be and supporting both would be very nice. esspecially supporting both at the same time so that the local workstation can run the Desktop Environments on directfb and remote users can run them on X.
use mplayer as an example. It performs better on directfb than xv or sdl(on my systems) but it is so nice having xv and sdl as an option.
AFAIK Linus has stated that most if not all of the code that SCO is B1tchin' about has be rewritten in 2.6 . The offending code was already planned to be replaced as is was a little bit to "old school" unixy and didn't play nice with some of the new scheduling AND the new SMP code is better than the old SMP code that was in question.
So why not just speed up developement of 2.6 kernels in Distros and just skip this whole issue? I personally think that IBM will win based on my limited legal knowledge and what I have read so IBM should be fine and SCO will fade away.
solid state sounds great, but its not currently feasable to get 200+GB of Memory in a machine. the cost is just too high. look at current pricing on flash memory and RAM. 64MB cost from 8$-30$. even at just 8$, thats 8$ per MB, consider that HardDrive storage is at 1$/GB you can see that its just not feasable right now. your hard drive costs would skyrocket to 1000$ for a 100-120GB drive. Do you want to pay that?
assume that a 3.5" drives spindles are 3.3" to account for casing and 2.5" is 2.3"
Also consider that their are 5280ft in a mile.
3.5 7200rpm = 6220 FEET PER MINUTE travel on the outside edge of the platter. 10000rpm = 8639 FEET PER MINUTE 15000rpm = 12958 2.5 7200rpm = 4335 10000rpm = 6021 15000rpm = 9031 20000rpm = 12042***********20kRPM same heat and reliability as 15kRPM
being able to knock off 30% of the edge velocity on the platters will also avoid a very large amount of the heat generated by friction.
though this will reduce transfer speeds on the outer edge, it will not reduce the average read/write rates by 30%, but only about 10%-15% because 70% of the drive is still exactly the same.
This would also potentially reduce costs because of less platter and casing matterials. Also giving faster seek times when the heads have to seek from edge to center of the platters.
Now with improved read speed based on higher data density and improved read heads, along with faster interfaces, the lack of a substantial gain in transfer speeds from current generations to the next will be overshadowed by increased reliablility and potentially faster spindle speeds. with less heat issues. also, having the outer edge of the spindle much closer to the center reduces vibration.
I have a 6MP digital camera and am a photoholic(is that a word?) i keep many images fully uncompressed to preserve quality and can easily use up a 120GB hard drive in a month. I currently have a 2TB array of SATA drives to store images and video footage. and will have that filled in a few months.
I also have a DVD+/-R and archive things that i haven't used in a while to disk.
I am a perfect example of someone who needs a sub 200$ terabyte in a few short years.
dont mean to be rude here, but your an idiot. what part about utilizing the depth of the plater to improve storage density instead of just the surface does not = 3d?
we have xyz - x=radius,y=circumfrence,z=plater depth
one of my biggest issues with linux is that X is slow and bulky. You can compare it to any other major OS is terms of Memory footprint(which is bad) The Ability to run Qt and eventually all of KDE on DirectFB is great. Should also push other toolkits to this, or maybee to evas or something.
this is a "cracker" , this is willfull distruction of anothers property and is against "hacker" ethics. "hackers" are all about freedom of information and will gladly break into a system and take data, but not destroy it.
why not use some native windows program than can be run as a terminal service or maybee even make one that makes an image of the windows system, compresses it, and saves it on a network share. would this not serve the same purpose and also keep windows file permissions with NTFS?
I administer a number of small schools networks and one in particular is a good example to compare to your needs. The school has about 275 users on a daily basis. This server handles e-mail, webserving, and is also a firewall/router for a 512k DSL line. It is a dual p2 450 with 1Gb ram and 2 10GB scsi in RAID1. This machine servers a static page for each student, and about 15 pages of misc school information. This machine also servers dynamic content from a database on the schools greenhouse class and also has a internet available copy of the schools library inventory and availability as the school library functions as the town's library.
as you can see, this machine is not very powerful but performs numerous takes easily. I have zero problems with this machine and I would consider keeping the current setup up to about 350 users. This machine is a Dell Poweredge in case you are wondering.
I run a miniITX with a C3-900Mhz. I got a nice little CF->IDE adapter on ebay for $10 and use a 128MB CF card for permanent storage. The machine has 512MB of RAM and i have a custom gentoo setup.
i run gentoo on my main box and export/use/portage via NFS so its available on the server. The server runs a basic system without X/sound/etc.
Re:BitTorrent = SHIT and still not legal anyway
on
ClusterKnoppix
·
· Score: 2, Informative
i must second the thought of the AC. i am using bittorrent to grab this file in a %100 legal manner and im getting well over 200K.
if everybody submits their $.02, then maybee Penny-Arcade can get the legal help they need. I have submitted a "story" for just this, to set up a paypal donation fun so that those interested can give a little to help a lot. Not only do i really love these guy's comic, but I very much wish to retain free speech and fighting this crap is the only way to retain it.
DS9 didn't suck(in my opinion) i liked it a lot. I also liked voyager. my ranking of trek: ST:TNG ST:DS9 ST:Voyager ST:Enterprise ST:TOS
i like them all, and though enterprise is no TNG, its still descent Sci-Fi, berman on the other hand almost seems determined to destroy it. Where is Frakes? He should com eto the rescue!
UT2003's OpenGL renderer performs within 5% of the D3D renderer on Windows.
Radeon9700 pro, P4 2.4@3.06
DirectX performs more than 30% better than OpenGL on my system, with the most up-2-date drivers. My other system runs a GF4 4600 and the difference between OpenGL and DirectX 8.1 are similar.
I Have seen complaints about this around the web quite a bit by people who are known to be knowledgable about gameing systems and software.
It seems to me that if the computer industry is going to change from 32bit x86, then this should be the point that they should move to a better IS. Is x86 something that need extending? It is old and inefficient for modern software, most modern processors spend a great deal of time working around the limitations of x86.
I think(personal oppinion) that this is a good point to move up to a modern IS, not necessarilt RISC but most likely.
From what i have read, im not a big fan of VLIW but it may be a good idea, other than the fact that the only VLIW processor out is a mamoth Itanic that is more efficient as a space heater than a CPU(maybe crusoe is VLIW?)
And if Backwards compatability is an issue, a hardware translator could be implemented that would allow x86 instructions to be translated(not emulated) into native instructions, similar to what the crusoe does.
linux would be great for this kind of processor, if the translator were a rom/ram chip, programmable on the fly, then the kernel could boot up in native IS and then load modules for different IS from disk. Essentially allowing the computer to be any computer out.
The crusoe does this VERY VERY well. And if performance is an arguement for you consider than a 1Ghz crusoe is ~athlon600 or so, and also consider that it has 1/3 the transistor count and great power features. take some time, up the transistor count, implement some good hardware Floating Point/SimD instructions that can easily handle translated sse/mmx/altivec/etc..OR native SimD.
--
of coarse the disadvantage is that the translated instructions would not be as fast as native instructions, but with good caching algorithyms and prediction units, and a good compiler that can to a considerable amount of translating on compile time(quasi-native executables, very simple "ports" to the new IS) this should be minimal. Most major production houses could offer simple patches with these "compiler ports" to allow a quick performance improvement while porting their software over to native IS.
--
emulators(really Virtual Processor Environments) would also become a "usable" means of running software, as running PPC code would be as simple as loading a module and an environmrent interface for memory and disk systems, something than current emulators due well but fall short on processor emulation.
-a boot loader to pick up a PPC module and then load OSX, or load x86/sparc/whatever.
-give us a chance to get away from x86 and also pc bios and into a better system.
1) upgrade drives to WD new 10,000RPM SATA drives, get a number of them and a TRUE hardware SATA RAID card with at least the ability to have 8MB cache/drive via at least pc100 sdr. Eight of these drives with a true hardware SATA RAID card could give you a near fully PCI load(133Mb/s, which is as good as your going to get without moving up to 64/66 PCI system(528Mb/s). The Hardware SATA RAID controller keeps you CPU usage to a minimum(SCSI levels or BETTER) and also gives you the thouroughput needed at 150Mb/s/drive max.
2)look into one of the SCSI LAN systems, using a SCSI160 or 320 controller to link computers together, and look at building some inexpensive machines like duron800/k7s5a combos for 100$ + 1.5Gb of ram per. The SCSI 160 connection easily fills up the PCI bus on the RAM machines and gives pretty good thouroughput because SCSI is very low latency. On the HOST machine, you can use pci64/66 so your BUS isn't much of a bottleneck. then you can use Linux Ramdisk on the RAM machines and export the drive over the SCSI interfact and mount it on the host. You can then create a RAID array on these if you like, but i think its unnecessary and just piling the RAM drive mounts together in a lateral array would be faster considering that RAID on the mounted disks will be in software, and lateral spanning is less processor intesive than striping
Slashdot Mirror
I have done some driver work, build and debuging and linux is a driver writers dream compared to windows. I would estimate that a linux driver will consume about 50% as much time as a windows drive writing either from scratch. I reality, you may have a lot or reusable code from the windows drivers so you could be looking at less. Also do take the advice of other posts and get the basic driver out ASAP, and allow the OS community help out on the rest.
Americans are disgusted with the idea of a foreign nation marching on them moreso than ANY OTHER COUNTRY ANYWHERE. If Canada looked at us funny, we would are and be ready! This ideology requires citizens to have guns, to feel safe. Also, the "only criminals have guns" idea is very sound here. Being the melting pot of the world, the worlds largest populous nation and by far the most imigrated to nation.
Now imagine keeping all of those people in check. Can you see why Americans like their guns soooo much?
It would also be impossible to take guns from Americans, our constitution guarantees our ability to own guns for personal protection. Americans would see that as the government breaking some holy vow or something and you'd have armies of people fighting the government.
---
now, jumping the fence a little, i despise guns. I think its rediculous that any civilized society NEEDS GUNS! killing other people should never be an issue. A civilized country should have vitually no crime that a civilian needs a gun to protect themselves from. People should both fear and respect the law as if it were ordered by God/a god/The One god/Buddah said it/Mighty Thor demands it, whatever....
I have spent enough time thinking about all the wrongs that "civilized" nations do everyday and know that Guns have no positive benefit to the masses.
how about the money that you spend on your gun, instead gets spent on a starving child somewhere. and the money governments spend to keep guns in check is also spent on that. How about Europeans that have nothing better to do that say Americans are gun toteing yanks spend a little less on other indulgences like inconceivably high consumption of alcohol and cigarettes and other various drugs that promote crime and the need for guns, and do a little charity work instead of "talking about policing the world, then letting America do the work, and the bitch about America trying to police the world"(where did that come from!) Guns rate right up their with Sadam Hussien, Nukes, Bio Weapons, and slavery. Almost as high as Capitol(hypocritical) punishment.(punish someone for killing someone, and then go and kill someone to prove how wrong it is)
yeah, i think my rant is over now. hell, i didn't like karma anyway, just slavey by a different name!
of course you could by one copy of photoshop and use it anywhere in you company.
....
You could only have it running on a SINGLE machine at any one time and you would problably have to make some sort of effort to enforce this. Adobe may have a clause in the licensing aggreement that states the software may only be installed on one machine, then in fact you would have license to a single copy of the software and could use it on as many as you wished as long as it was installed on just one.
This could be done via mapped drives or something like that, assuming photoshop would run without 10,000 registry additions and dll in system32/
i think that it has a good chance of happening to some degree. Machines will most likely make the products cutting a good amount of hours from human workers, but humans wil problably still work the counter. People will rather talk to a human that a machine in public places.
I also dont think that this will be as devastating to the economy as some do. I see that by 2050, as many people(by %) will be employed but hours will change. a full work day will likely be just 3 or less hours. with people working just 1/3 the amount of time, 3 times as many people are needed to fill a work day, keeping the unemployment rates down.
also keep in mind that machines are extremely cheap to have do things, and this trend will likely continue. some examples:
in a fast food place like Wendy's, human labor cosists of about 17% of the gross profits. This could be reduced to 2-3% and maybee another 3% for maintainance and energy costs. shipping gets cheaper having machines do the work also. All this means that cost is less so price can be less makeing the situation even better for people that only work 3 hours a day.
of course america will be at the front of the line in implementing this so that we can get more fat and lazy. This in turn will counteract the fact that people should live longer because of less work related abuse to their bodies.
anyone agree?
i keep most images in TIFF. Its big but easy to work with, i dont spend cpu cycles on decompressing bits to work with them. Lossless compression is nice, but tar.gz and rar are really great lossless compressors :) . I can't wait til a stable filesystem that can mount compressed formats :)
I know a bit about X, and I know a bit about DirectFB(just research, not code knowledge)
i know that i would like to see KDE and QT compatible with both directfb AND X windows. X is extremely usefull and will evolve very nicely as a lot of legacy support is removed, but it just can't be as fast as directfb can be and supporting both would be very nice. esspecially supporting both at the same time so that the local workstation can run the Desktop Environments on directfb and remote users can run them on X.
use mplayer as an example. It performs better on directfb than xv or sdl(on my systems) but it is so nice having xv and sdl as an option.
AFAIK Linus has stated that most if not all of the code that SCO is B1tchin' about has be rewritten in 2.6 . The offending code was already planned to be replaced as is was a little bit to "old school" unixy and didn't play nice with some of the new scheduling AND the new SMP code is better than the old SMP code that was in question.
So why not just speed up developement of 2.6 kernels in Distros and just skip this whole issue? I personally think that IBM will win based on my limited legal knowledge and what I have read so IBM should be fine and SCO will fade away.
solid state sounds great, but its not currently feasable to get 200+GB of Memory in a machine. the cost is just too high. look at current pricing on flash memory and RAM. 64MB cost from 8$-30$. even at just 8$, thats 8$ per MB, consider that HardDrive storage is at 1$/GB you can see that its just not feasable right now. your hard drive costs would skyrocket to 1000$ for a 100-120GB drive. Do you want to pay that?
assume that a 3.5" drives spindles are 3.3" to account for casing and 2.5" is 2.3"
Also consider that their are 5280ft in a mile.
3.5
7200rpm = 6220 FEET PER MINUTE travel on the outside edge of the platter.
10000rpm = 8639 FEET PER MINUTE
15000rpm = 12958
2.5
7200rpm = 4335
10000rpm = 6021
15000rpm = 9031
20000rpm = 12042***********20kRPM same heat and reliability as 15kRPM
being able to knock off 30% of the edge velocity on the platters will also avoid a very large amount of the heat generated by friction.
though this will reduce transfer speeds on the outer edge, it will not reduce the average read/write rates by 30%, but only about 10%-15% because 70% of the drive is still exactly the same.
This would also potentially reduce costs because of less platter and casing matterials. Also giving faster seek times when the heads have to seek from edge to center of the platters.
Now with improved read speed based on higher data density and improved read heads, along with faster interfaces, the lack of a substantial gain in transfer speeds from current generations to the next will be overshadowed by increased reliablility and potentially faster spindle speeds. with less heat issues. also, having the outer edge of the spindle much closer to the center reduces vibration.
all made feasable by higher data density/inch
agreed with above post.
I have a 6MP digital camera and am a photoholic(is that a word?) i keep many images fully uncompressed to preserve quality and can easily use up a 120GB hard drive in a month. I currently have a 2TB array of SATA drives to store images and video footage. and will have that filled in a few months.
I also have a DVD+/-R and archive things that i haven't used in a while to disk.
I am a perfect example of someone who needs a sub 200$ terabyte in a few short years.
dont mean to be rude here, but your an idiot. what part about utilizing the depth of the plater to improve storage density instead of just the surface does not = 3d?
we have xyz - x=radius,y=circumfrence,z=plater depth
thats 3D
one of my biggest issues with linux is that X is slow and bulky. You can compare it to any other major OS is terms of Memory footprint(which is bad) The Ability to run Qt and eventually all of KDE on DirectFB is great. Should also push other toolkits to this, or maybee to evas or something.
this is a "cracker" , this is willfull distruction of anothers property and is against "hacker" ethics. "hackers" are all about freedom of information and will gladly break into a system and take data, but not destroy it.
why not use some native windows program than can be run as a terminal service or maybee even make one that makes an image of the windows system, compresses it, and saves it on a network share. would this not serve the same purpose and also keep windows file permissions with NTFS?
I agree that this setup is quite overkill.
I administer a number of small schools networks and one in particular is a good example to compare to your needs. The school has about 275 users on a daily basis. This server handles e-mail, webserving, and is also a firewall/router for a 512k DSL line. It is a dual p2 450 with 1Gb ram and 2 10GB scsi in RAID1. This machine servers a static page for each student, and about 15 pages of misc school information. This machine also servers dynamic content from a database on the schools greenhouse class and also has a internet available copy of the schools library inventory and availability as the school library functions as the town's library.
as you can see, this machine is not very powerful but performs numerous takes easily. I have zero problems with this machine and I would consider keeping the current setup up to about 350 users. This machine is a Dell Poweredge in case you are wondering.
I run a miniITX with a C3-900Mhz. I got a nice little CF->IDE adapter on ebay for $10 and use a 128MB CF card for permanent storage. The machine has 512MB of RAM and i have a custom gentoo setup.
/use/portage via NFS so its available on the server. The server runs a basic system without X/sound/etc.
i run gentoo on my main box and export
i must second the thought of the AC. i am using bittorrent to grab this file in a %100 legal manner and im getting well over 200K.
a bittorent site specifically for slashdot victims.
go to
BitTorrent Files for Slashdot Effect Victims
why is their an overwhelming number of posts claiming that changing from QWERTY is stupid because its non-conformist!
The majority of computer users use windows, does that make us Linux guys stupid bacause we are non-conformist!?
if everybody submits their $.02, then maybee Penny-Arcade can get the legal help they need. I have submitted a "story" for just this, to set up a paypal donation fun so that those interested can give a little to help a lot. Not only do i really love these guy's comic, but I very much wish to retain free speech and fighting this crap is the only way to retain it.
this is just evil to do, April fools sucks.
i read the title and about flew off the handle uintil about 3 seconds later when i realized it was april fools. PURE EVIL!!
DS9 didn't suck(in my opinion) i liked it a lot. I also liked voyager. my ranking of trek:
ST:TOS
ST:TNG
ST:DS9
ST:Voyager
ST:Enterprise
i like them all, and though enterprise is no TNG, its still descent Sci-Fi, berman on the other hand almost seems determined to destroy it. Where is Frakes? He should com eto the rescue!
UT2003's OpenGL renderer performs within 5% of the D3D renderer on Windows.
Radeon9700 pro, P4 2.4@3.06 DirectX performs more than 30% better than OpenGL on my system, with the most up-2-date drivers. My other system runs a GF4 4600 and the difference between OpenGL and DirectX 8.1 are similar.
I Have seen complaints about this around the web quite a bit by people who are known to be knowledgable about gameing systems and software.
It seems to me that if the computer industry is going to change from 32bit x86, then this should be the point that they should move to a better IS. Is x86 something that need extending? It is old and inefficient for modern software, most modern processors spend a great deal of time working around the limitations of x86.
I think(personal oppinion) that this is a good point to move up to a modern IS, not necessarilt RISC but most likely.
From what i have read, im not a big fan of VLIW but it may be a good idea, other than the fact that the only VLIW processor out is a mamoth Itanic that is more efficient as a space heater than a CPU(maybe crusoe is VLIW?)
And if Backwards compatability is an issue, a hardware translator could be implemented that would allow x86 instructions to be translated(not emulated) into native instructions, similar to what the crusoe does.
linux would be great for this kind of processor, if the translator were a rom/ram chip, programmable on the fly, then the kernel could boot up in native IS and then load modules for different IS from disk. Essentially allowing the computer to be any computer out.
The crusoe does this VERY VERY well. And if performance is an arguement for you consider than a 1Ghz crusoe is ~athlon600 or so, and also consider that it has 1/3 the transistor count and great power features. take some time, up the transistor count, implement some good hardware Floating Point/SimD instructions that can easily handle translated sse/mmx/altivec/etc..OR native SimD.
--
of coarse the disadvantage is that the translated instructions would not be as fast as native instructions, but with good caching algorithyms and prediction units, and a good compiler that can to a considerable amount of translating on compile time(quasi-native executables, very simple "ports" to the new IS) this should be minimal. Most major production houses could offer simple patches with these "compiler ports" to allow a quick performance improvement while porting their software over to native IS.
--
emulators(really Virtual Processor Environments) would also become a "usable" means of running software, as running PPC code would be as simple as loading a module and an environmrent interface for memory and disk systems, something than current emulators due well but fall short on processor emulation.
-a boot loader to pick up a PPC module and then load OSX, or load x86/sparc/whatever.
-give us a chance to get away from x86 and also pc bios and into a better system.
==
excuse the long post, and have a good morning.
1) upgrade drives to WD new 10,000RPM SATA drives, get a number of them and a TRUE hardware SATA RAID card with at least the ability to have 8MB cache/drive via at least pc100 sdr. Eight of these drives with a true hardware SATA RAID card could give you a near fully PCI load(133Mb/s, which is as good as your going to get without moving up to 64/66 PCI system(528Mb/s). The Hardware SATA RAID controller keeps you CPU usage to a minimum(SCSI levels or BETTER) and also gives you the thouroughput needed at 150Mb/s/drive max.
2)look into one of the SCSI LAN systems, using a SCSI160 or 320 controller to link computers together, and look at building some inexpensive machines like duron800/k7s5a combos for 100$ + 1.5Gb of ram per. The SCSI 160 connection easily fills up the PCI bus on the RAM machines and gives pretty good thouroughput because SCSI is very low latency. On the HOST machine, you can use pci64/66 so your BUS isn't much of a bottleneck. then you can use Linux Ramdisk on the RAM machines and export the drive over the SCSI interfact and mount it on the host. You can then create a RAID array on these if you like, but i think its unnecessary and just piling the RAM drive mounts together in a lateral array would be faster considering that RAID on the mounted disks will be in software, and lateral spanning is less processor intesive than striping