How about a product that is already out there, M-Systems 2.5" Ram Disks in the 1 to 128Gb area as used but the army with a US military standard (MIL-STD-810F)
1k production of Iram seems pretty meager. I wonder what GByte is thinking. Is it really a usefull hardware addition or does it need a revision or two?
This thing is not intended for use as a primary drive.
It's not for gamers.
I will be buying this as soon as it comes out, for one reason: SCRATCH DISK.
For anyone who does desktop publishing, DV editing, photography, animation or modeling, this is a godsend.
I cannot tell you how many hundreds of times working in Premiere or After Effects, Photoshop or Illustrator, I have craved having a 2-4gig ramdrive. This thing will make lives soo much easier for those of us who do any creative work.
I really can't believe there werent any comparisons using any productivity software, where this product is really intended and would be useful
1. Not everyone runs the newest 74g Raptor, which is what the comparison was to. Granted if you were willing to spend on this, you might be more likely to have a raptor, but nonetheless, you have more of a speed increase coming from slower.
2. I believe the majority of us that do have raptors hear them. If they coudl give me more space, I would consider swapping out the raptor for one of these. Although hard drives are not that loud, Raptors are not that quiet, especially when you are a noise freak.
Lastly, In regards to file transfers, yes, I-ram to I-ram seems rather stupid, unless you had 1 I-ram on one computer, and another I-ram on another computer and transferred them over the network. All of the sudden you might get to use that network cable for more than just bragging rights. Going at 100mb/s seems like a toasty offer over the network cable in comparison to what happens now.
I think the i-RAM needs software improvements, not hardware improvements.
In its present state, it's an intriguing idea, and all of us techy guys are thinking, "hey, it's kind of cool". But there's not much genuinely useful that can be done with it, without a lot of faffing about or copying, IMO.
But I think a change of perspective is needed.
Stop thinking about i-RAM as a small, fast hard drive.
Now, start to think about some of the details proposed about the new "hybrid hard drives" Microsoft talked about in regards to Windows Longhorn/Vista.
Memory on the drive, transparent to the user, but used to speed Windows up by, e.g. copying Windows startup files to it when Windows is shutdown.
Now, what if you consider the i-RAM as a bridging step between these hybrid hard drives and today's hard drives.
When you hibernate your PC, instead of writing that hibernation data to the HD, it could transparently be written to the i-RAM, allowing faster boots.
Now take a leap from there, and start to think about what else it could do to transparently optimise your system.
Windows XP already stores some limited prefetch data to boost application startup speed. Let's think about moving that prefetch data onto the i-RAM and taking the concept further.
When I first read about the optimisations in the Windows XP boot, I thought they sounded really clever and sensible. Microsoft found that there are sections of the boot where the disk didn't do anything, so in XP, XP does stuff in the background while the disk stuff is read into memory in preparation of when it is needed.
Now let's look at Battlefield 2, or almost any other current game. Aren't all those splash screens and animations that happen when you run the game annoying? What if, instead of just sitting there, while all those logos are being displayed, the i-RAM started copying files likely to be used onto itself, from the hard drive. That way as soon as you start to load a level, you could get the i-RAM speed boost... Without having to install the entire application onto the i-RAM.
Well I thought it was a good idea, anyway ;)
It certainly set my mind off wondering how it could be done, and whether much of it would make much sense even to computers without an i-RAM.
Yeah but if you are gonna have some kind of smart software that predicts what you data you are about to access, that data may as well be loaded to RAM straight away where it is destined instead of the i-ram. The i-ram aint all that much quicker than a HDD anyway!
I am having a hard time being enthusiastic about this thing.
Pros
approx 25% faster load times? basically negligiable difference over a hard drive.
Copying from i-disk to i-disk 4 times faster - what use is that though? not going to sit around all day duplicating files on the same drive just to 'experience' the speed. When duplicating files, or zipping or rendering etc to avoid 10meg/s speeds it is best to have 2 hard drives and copy from one drive to the other at ~60meg/s.
silent - hdds are not very loud either
cons
VERY expensive
Lose all data if power is lost for a day
lose a pci slot - this is important to consider if building a media PC with only 1 or 2 slots (since the thing is silent)
Very Very small capacity, lets not kid ourselves here, even if it was 8gb or 16gig (OMG THE COST!) and even if it was worth the 5 second saving in level loading, I don't think it would be long before a game is too large for it, I may be wrong but with HD-dvd and blueray comming to consoles PC games are likely to also get larger still and some are over 4 gig already. Forget multiple games on it.
in summary, interesting in theory but flawed in practise.
I don't think the cheap 1GB sticks are that far off. I was browsing eBay and it seems that you can get DDR266/DDR333 1GB sticks for about $60 shipped; very reasonable IMO. If you combine shipping, you can have 4GBs worth of storage for just over $200. And since the RAM only runs at 200mhz anyway, it doesn't really matter if you get DDR266 or anything above that. I think I might get one of these bad boys if they run a special or a promotion for <$100.
THIS COMMENT SYSTEM SUCKS! Please Anand, change it back. It takes me twice as long to read the comments now. Can't afford to read on the job. Please do something!!!!!!!
I have to say I think it is very good.
From what I have seen thus far the commments seem to be much better and more constructive than with the old system.
Back to the main topic.
I think this sort of technology is very promising. Not in a general user system at the moment but for enthusiast systems where users demand the very best performance I think it will do very well.
I myself would almost certainly be in the market for one of these if it supported 300MB/s transfer and had an 8GB capacity. If Gigabyte or indeed any other manufacturer made a device that could acheive this then I reckon the performance on tap would be even more impressive.
you do have a point, i haven't seen any First Posts or Russian jokes, but still, I would like to see the comments on the same page as the news. I usually open all the news subjects on a separate tab, so if there are 10 news articles i would have 10 tabs open and now 10 popups for the comments. that is ridiculous.
I wonder how much faster this would have been helped by
1) Supporting memory that's actually running at DDR400 or DDR500
2) Having Sata II support.
3) Putting a 128 bit interface on there (like current motherboards do)
I get the feeling that most of the performance bottleneck comes from SATA's implementation hardly ever reaching theoretical, but I bet the other two would help performance, too.
Finally, this is one of those devices that's <b>begging</b> to be put on 1 lane PCI express.
I also wonder how faster memory might help, but then there's the possibility its SATA interface would bottleneck it making the faster memory no better than if its running 200MHz DDR. SATA II might help there, but we'd need both. Again we also run into problems with the dual channel 128bit interface if SATA is too slow for it. Obviously you'd think it'd be a lot easier to get the memory running @ 400MHz oppposed to 200MHz instead of 128bit interface, which generally provides much less performance boost than doubling your memory speed.
One thing that really makes me wonder is RAIDing some of these bad boys together. Not only do you solve some of the capacity problems, you also have a chance to increase speed in certain cases. Although doubling the i-RAM is also going to double the already steep entry price. I guess unless you have a ton of 512MB sticks lying around (as 1GB aren't exactly as common), it might be more worth it but with two i-RAMs you're already pushing a grand.
It is very intriguing.
Future revs and prehaps competition from other companies could make things even more interesting and far more realistic even (ie $50 per card if not less). Good news is the DDR1 prices seem to be steadily falling, especially for 1GB modules - they're actually a reasonable option now (not outrageously expensive as they were only months ago) for an upgrade, and I'm definately considering getting a pair to upgrade my system. It might not be long before they're only $50 a pop for cheapie stuff that you'd definately not mind having run @ a mere 200MHz.
Although even still, in such an "ideal" situation, the total cost would be $250 or so for a mere 4GB SSD. $500 for 8GB in RAID-0. Raptors seem like a bargain in comparison.
The niche might be its biggest role. Completely silent systems such as HTPC setups would definately be and ideal situation for the i-RAM.
Even if nothing really big comes directly from this, it is definately a step in the right direction in pushing solutions for faster storage devices that aren't insanely expensive.
Although it does make you wonder how cheap memory manufactures could produce super high capacity DDR200 for specific roles such as this. Usually the slower memory fades away because it becomes very poor for modern system memory performance. You definatey do not want DDR200 when you can have DDR500, and memory manufacturers aren't going to want to produce it with no demand. However if there was that demand they could probably produce it for cheap.
I know its just dream talk, but it'd be great if a future version supported more than 1GB sticks and we had memory manufacturers pumping out cheap DDR200 at high capacity (2GB or even more)
I see a lot of room for improvement on this little device. There's a lot to be gained here.
DDR 200 or DDR 400 will perform precisely the same. Check the bandwidth on this table (http://www.acme.com/build_a_pc/bandwidth.html)">http://www.acme.com/build_a_pc/bandwidth.html) to see that BOTH DDR 200 and DDR 400 saturate the SATA bus... (1.6Gbps and 3.2Gbps). Both standards run on the usual 64bit bus width. SATA (I) width is 1.5Gbps. SATA II is 3.0Gbps. So if you actually moved onto SATA II you'd gain a lot from i-RAM and DDR 400, and lose a lot using DDR 200.
Where are the REAL applications for this? Like someone said: scratch disk! BUT also if people worry about the power issue... think servers. all sorts of servers would benefit from iRAM. I can imagine that server farms would also benefit a lot. I can see the workstation market coming up with special iRAM-totting solutions.
A 2nd generation of these things could effectively benefit from using not 1 but 2 SATA controllers onboard (there are tons of add-in cards available on the market that already sport 2 SATA controllers onboard). This would effectively create a dual iRAM card with a 3.0Gbps threshold (but still no point in using DDR400 there, as you'd actually be doing "dual channel DDR200") - RAID Zeroed of course.
SATA-II chips are here but the hard-drives suck big time, you're not getting a proportional increase in performance (double the bandwidth ISN'T double the data transfer) BUT if you actually use iRAM with SATA-II controllers you could actually double the iRAM performance and take it waaayyy beyond the ability of any harddrive today. 300MB/s plus the possibility of RAIDing, which would take it to 600MB/s... SCSI could finally Rest in Peace.
Unfortunately GIGABYTE shot itself in the foot: for 150 bucks a pop, plus the cost of RAM you can go with current top of the range RAPTOR's, raided together and get almost identical performance for half the price. Gigabyte screwed up big time on pricing, imho. Greedy buggers.
Actually, coming back to that Microsoft thingie... why the heck aren't we reintroducing things like L3 Cache in the middle of the SATA interfaces? That way we could actually do those hybrid drives.
I think this product is very interesting, but it really needs to have 8 or 16 slots, and slow DDR has to cost <$50 GB . Whenever that happens, (2 years?) a huge market for this type of device will exist. Perhaps they could put it in a double size HD enclosure and use 4 pin molex connectors instead? Space starts becoming the big problem I suppose when you're talking about eight or more slots.
1). install windows xp on the i-ram drive
2). boot windows xp off another HDD on same comp (d:\windows, for instance)
3). copy c:\Program Files\ (from i-ram) onto a temporary folder on the other HDD
4). copy c:\documents and settings\ onto a temporary folder on the other HDD
5). with a third hard, unpartitioned HDD, create 2 NTFS partitions (let's say 10 gigs each)
6). mount one partition as C:\Program Files\ using Disk Management
7). copy program files back to c:\Program Files
8). repeat steps 6-7 for Documents and Settings
So, what would this do? In theory (untested of course because I do not have a i-ram drive) you should be able to boot winXP fast while still being able to install programs in C:\Program Files\ and use My Documents...
It's overly complex. When Program Files and similar directories are accessed, the program doing it is basically just following a symlink/shortcut provided by Windows; all you'd need to do is use PowerTools/RegEdit to change where Windows thinks these directories are, and you'd have accomplished the same thing without so much work or partition madness. It's a reasonable idea though as long as you're sure your Windows i-Ram card won't be powerless for too long.
It definitely has some potential, but I'm thinking this version is too limited. If it makes it to ver2 or ver3 we'll have something worth tinkering with, but I will pass on this go 'round.
One point that would be interesting in knowing if something like the I-Ram could be programmed for, so that all paging out always uses the I-Ram (effectively increasing your ram) or commonally used files could be placed on it, something like a 8GB cache for a hard-drive.
8GB is too small to put every file on (even though I would really like to)... but if the 80% most used files are automatically installed on that 8GB I think that most programs would see an increase in productivity.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
27 Comments
Back to Article
supermicroman - Thursday, November 10, 2005 - link
Does anyone know where I can get an I-Ram?I have been looking for one for days and it is nowhere to be found.
Did Gigabyte decide not to produce the 1,000 units that were slated for the first production run?
This would be a perfect for I/0 intensive applications, or for productivity software that can use a scratch drive.
hegars - Thursday, August 11, 2005 - link
How about a product that is already out there, M-Systems 2.5" Ram Disks in the 1 to 128Gb area as used but the army with a US military standard (MIL-STD-810F)I know its flash so its a little different
http://www.m-sys.com/site/en-US/Products/IDESCSIFF...">http://www.m-sys.com/site/en-US/Product...s_/SATA_...
exdeath - Monday, August 8, 2005 - link
100 MB/s? It can't be that slow.I'm pushing 140 MB/sec with two WD740GD Raptor's in nForce4 RAID0.
http://members.cox.net/exdeath/pc/tach.jpg">http://members.cox.net/exdeath/pc/tach.jpg
http://members.cox.net/exdeath/pc/raid.jpg">http://members.cox.net/exdeath/pc/raid.jpg
ksherman - Friday, August 5, 2005 - link
I dont suppose we have any new information as to how close we are to x86 Macs yet... Kinda interested in the design and features coming up...CrystalBay - Tuesday, August 2, 2005 - link
1k production of Iram seems pretty meager. I wonder what GByte is thinking. Is it really a usefull hardware addition or does it need a revision or two?noen - Sunday, July 31, 2005 - link
This thing is not intended for use as a primary drive.It's not for gamers.
I will be buying this as soon as it comes out, for one reason: SCRATCH DISK.
For anyone who does desktop publishing, DV editing, photography, animation or modeling, this is a godsend.
I cannot tell you how many hundreds of times working in Premiere or After Effects, Photoshop or Illustrator, I have craved having a 2-4gig ramdrive. This thing will make lives soo much easier for those of us who do any creative work.
I really can't believe there werent any comparisons using any productivity software, where this product is really intended and would be useful
waldo - Thursday, July 28, 2005 - link
I think there has been a bit of forgettfulness.1. Not everyone runs the newest 74g Raptor, which is what the comparison was to. Granted if you were willing to spend on this, you might be more likely to have a raptor, but nonetheless, you have more of a speed increase coming from slower.
2. I believe the majority of us that do have raptors hear them. If they coudl give me more space, I would consider swapping out the raptor for one of these. Although hard drives are not that loud, Raptors are not that quiet, especially when you are a noise freak.
Lastly, In regards to file transfers, yes, I-ram to I-ram seems rather stupid, unless you had 1 I-ram on one computer, and another I-ram on another computer and transferred them over the network. All of the sudden you might get to use that network cable for more than just bragging rights. Going at 100mb/s seems like a toasty offer over the network cable in comparison to what happens now.
Turnip - Wednesday, July 27, 2005 - link
I think the i-RAM needs software improvements, not hardware improvements.In its present state, it's an intriguing idea, and all of us techy guys are thinking, "hey, it's kind of cool". But there's not much genuinely useful that can be done with it, without a lot of faffing about or copying, IMO.
But I think a change of perspective is needed.
Stop thinking about i-RAM as a small, fast hard drive.
Now, start to think about some of the details proposed about the new "hybrid hard drives" Microsoft talked about in regards to Windows Longhorn/Vista.
Memory on the drive, transparent to the user, but used to speed Windows up by, e.g. copying Windows startup files to it when Windows is shutdown.
Now, what if you consider the i-RAM as a bridging step between these hybrid hard drives and today's hard drives.
When you hibernate your PC, instead of writing that hibernation data to the HD, it could transparently be written to the i-RAM, allowing faster boots.
Now take a leap from there, and start to think about what else it could do to transparently optimise your system.
Windows XP already stores some limited prefetch data to boost application startup speed. Let's think about moving that prefetch data onto the i-RAM and taking the concept further.
When I first read about the optimisations in the Windows XP boot, I thought they sounded really clever and sensible. Microsoft found that there are sections of the boot where the disk didn't do anything, so in XP, XP does stuff in the background while the disk stuff is read into memory in preparation of when it is needed.
Now let's look at Battlefield 2, or almost any other current game. Aren't all those splash screens and animations that happen when you run the game annoying? What if, instead of just sitting there, while all those logos are being displayed, the i-RAM started copying files likely to be used onto itself, from the hard drive. That way as soon as you start to load a level, you could get the i-RAM speed boost... Without having to install the entire application onto the i-RAM.
Well I thought it was a good idea, anyway ;)
It certainly set my mind off wondering how it could be done, and whether much of it would make much sense even to computers without an i-RAM.
Hmmmmm.
What does everyone think?
MrFantastic - Wednesday, July 27, 2005 - link
Yeah but if you are gonna have some kind of smart software that predicts what you data you are about to access, that data may as well be loaded to RAM straight away where it is destined instead of the i-ram. The i-ram aint all that much quicker than a HDD anyway!MrFantastic - Wednesday, July 27, 2005 - link
I am having a hard time being enthusiastic about this thing.Pros
approx 25% faster load times? basically negligiable difference over a hard drive.
Copying from i-disk to i-disk 4 times faster - what use is that though? not going to sit around all day duplicating files on the same drive just to 'experience' the speed. When duplicating files, or zipping or rendering etc to avoid 10meg/s speeds it is best to have 2 hard drives and copy from one drive to the other at ~60meg/s.
silent - hdds are not very loud either
cons
VERY expensive
Lose all data if power is lost for a day
lose a pci slot - this is important to consider if building a media PC with only 1 or 2 slots (since the thing is silent)
Very Very small capacity, lets not kid ourselves here, even if it was 8gb or 16gig (OMG THE COST!) and even if it was worth the 5 second saving in level loading, I don't think it would be long before a game is too large for it, I may be wrong but with HD-dvd and blueray comming to consoles PC games are likely to also get larger still and some are over 4 gig already. Forget multiple games on it.
in summary, interesting in theory but flawed in practise.
notposting - Tuesday, July 26, 2005 - link
is broken.Gage8 - Wednesday, July 27, 2005 - link
not broken, just the wrong link. This takes you to the weblog page instead of the articleBoboSama - Tuesday, July 26, 2005 - link
I don't think the cheap 1GB sticks are that far off. I was browsing eBay and it seems that you can get DDR266/DDR333 1GB sticks for about $60 shipped; very reasonable IMO. If you combine shipping, you can have 4GBs worth of storage for just over $200. And since the RAM only runs at 200mhz anyway, it doesn't really matter if you get DDR266 or anything above that. I think I might get one of these bad boys if they run a special or a promotion for <$100.aggie02 - Tuesday, July 26, 2005 - link
THIS COMMENT SYSTEM SUCKS! Please Anand, change it back. It takes me twice as long to read the comments now. Can't afford to read on the job. Please do something!!!!!!!George Powell - Tuesday, July 26, 2005 - link
I have to say I think it is very good.From what I have seen thus far the commments seem to be much better and more constructive than with the old system.
Back to the main topic.
I think this sort of technology is very promising. Not in a general user system at the moment but for enthusiast systems where users demand the very best performance I think it will do very well.
I myself would almost certainly be in the market for one of these if it supported 300MB/s transfer and had an 8GB capacity. If Gigabyte or indeed any other manufacturer made a device that could acheive this then I reckon the performance on tap would be even more impressive.
aggie02 - Tuesday, July 26, 2005 - link
you do have a point, i haven't seen any First Posts or Russian jokes, but still, I would like to see the comments on the same page as the news. I usually open all the news subjects on a separate tab, so if there are 10 news articles i would have 10 tabs open and now 10 popups for the comments. that is ridiculous.SeventhCycle - Tuesday, July 26, 2005 - link
I wonder how much faster this would have been helped by1) Supporting memory that's actually running at DDR400 or DDR500
2) Having Sata II support.
3) Putting a 128 bit interface on there (like current motherboards do)
I get the feeling that most of the performance bottleneck comes from SATA's implementation hardly ever reaching theoretical, but I bet the other two would help performance, too.
Finally, this is one of those devices that's <b>begging</b> to be put on 1 lane PCI express.
bunnyfubbles - Tuesday, July 26, 2005 - link
I also wonder how faster memory might help, but then there's the possibility its SATA interface would bottleneck it making the faster memory no better than if its running 200MHz DDR. SATA II might help there, but we'd need both. Again we also run into problems with the dual channel 128bit interface if SATA is too slow for it. Obviously you'd think it'd be a lot easier to get the memory running @ 400MHz oppposed to 200MHz instead of 128bit interface, which generally provides much less performance boost than doubling your memory speed.One thing that really makes me wonder is RAIDing some of these bad boys together. Not only do you solve some of the capacity problems, you also have a chance to increase speed in certain cases. Although doubling the i-RAM is also going to double the already steep entry price. I guess unless you have a ton of 512MB sticks lying around (as 1GB aren't exactly as common), it might be more worth it but with two i-RAMs you're already pushing a grand.
It is very intriguing.
Future revs and prehaps competition from other companies could make things even more interesting and far more realistic even (ie $50 per card if not less). Good news is the DDR1 prices seem to be steadily falling, especially for 1GB modules - they're actually a reasonable option now (not outrageously expensive as they were only months ago) for an upgrade, and I'm definately considering getting a pair to upgrade my system. It might not be long before they're only $50 a pop for cheapie stuff that you'd definately not mind having run @ a mere 200MHz.
Although even still, in such an "ideal" situation, the total cost would be $250 or so for a mere 4GB SSD. $500 for 8GB in RAID-0. Raptors seem like a bargain in comparison.
The niche might be its biggest role. Completely silent systems such as HTPC setups would definately be and ideal situation for the i-RAM.
Even if nothing really big comes directly from this, it is definately a step in the right direction in pushing solutions for faster storage devices that aren't insanely expensive.
Although it does make you wonder how cheap memory manufactures could produce super high capacity DDR200 for specific roles such as this. Usually the slower memory fades away because it becomes very poor for modern system memory performance. You definatey do not want DDR200 when you can have DDR500, and memory manufacturers aren't going to want to produce it with no demand. However if there was that demand they could probably produce it for cheap.
I know its just dream talk, but it'd be great if a future version supported more than 1GB sticks and we had memory manufacturers pumping out cheap DDR200 at high capacity (2GB or even more)
Malikot - Tuesday, August 16, 2005 - link
I see a lot of room for improvement on this little device. There's a lot to be gained here.DDR 200 or DDR 400 will perform precisely the same. Check the bandwidth on this table (http://www.acme.com/build_a_pc/bandwidth.html)">http://www.acme.com/build_a_pc/bandwidth.html) to see that BOTH DDR 200 and DDR 400 saturate the SATA bus... (1.6Gbps and 3.2Gbps). Both standards run on the usual 64bit bus width. SATA (I) width is 1.5Gbps. SATA II is 3.0Gbps. So if you actually moved onto SATA II you'd gain a lot from i-RAM and DDR 400, and lose a lot using DDR 200.
Where are the REAL applications for this? Like someone said: scratch disk! BUT also if people worry about the power issue... think servers. all sorts of servers would benefit from iRAM. I can imagine that server farms would also benefit a lot. I can see the workstation market coming up with special iRAM-totting solutions.
A 2nd generation of these things could effectively benefit from using not 1 but 2 SATA controllers onboard (there are tons of add-in cards available on the market that already sport 2 SATA controllers onboard). This would effectively create a dual iRAM card with a 3.0Gbps threshold (but still no point in using DDR400 there, as you'd actually be doing "dual channel DDR200") - RAID Zeroed of course.
SATA-II chips are here but the hard-drives suck big time, you're not getting a proportional increase in performance (double the bandwidth ISN'T double the data transfer) BUT if you actually use iRAM with SATA-II controllers you could actually double the iRAM performance and take it waaayyy beyond the ability of any harddrive today. 300MB/s plus the possibility of RAIDing, which would take it to 600MB/s... SCSI could finally Rest in Peace.
Unfortunately GIGABYTE shot itself in the foot: for 150 bucks a pop, plus the cost of RAM you can go with current top of the range RAPTOR's, raided together and get almost identical performance for half the price. Gigabyte screwed up big time on pricing, imho. Greedy buggers.
Actually, coming back to that Microsoft thingie... why the heck aren't we reintroducing things like L3 Cache in the middle of the SATA interfaces? That way we could actually do those hybrid drives.
robaroncape - Friday, December 14, 2007 - link
BAD DEALS FOR BAD MEMORY ON EBAYPlease read my guide about the I-ram, high density ram and Bzboys. If you are looking for high density ram for your I-ram you must read this. It will save you alot of money,time and headaches.
This is the fight to hve them removed from ebay.
Thanks, Rob
http://reviews.ebay.com/BAD-DEALS-FOR-BAD-MEMORY-O...">http://reviews.ebay.com/BAD-DEALS-FOR-B...MORY-ON-...
StanleyBuchanan - Monday, July 25, 2005 - link
I would love to have 12GB to work with...That's Windows XP, a productivity suite, and 1 modern game.
Anything beyond that... NAS
lolerton - Monday, July 25, 2005 - link
I think this product is very interesting, but it really needs to have 8 or 16 slots, and slow DDR has to cost <$50 GB . Whenever that happens, (2 years?) a huge market for this type of device will exist. Perhaps they could put it in a double size HD enclosure and use 4 pin molex connectors instead? Space starts becoming the big problem I suppose when you're talking about eight or more slots.houst - Monday, July 25, 2005 - link
I was wondering if the following is possible:1). install windows xp on the i-ram drive
2). boot windows xp off another HDD on same comp (d:\windows, for instance)
3). copy c:\Program Files\ (from i-ram) onto a temporary folder on the other HDD
4). copy c:\documents and settings\ onto a temporary folder on the other HDD
5). with a third hard, unpartitioned HDD, create 2 NTFS partitions (let's say 10 gigs each)
6). mount one partition as C:\Program Files\ using Disk Management
7). copy program files back to c:\Program Files
8). repeat steps 6-7 for Documents and Settings
So, what would this do? In theory (untested of course because I do not have a i-ram drive) you should be able to boot winXP fast while still being able to install programs in C:\Program Files\ and use My Documents...
Thoughts?
ViRGE - Monday, July 25, 2005 - link
It's overly complex. When Program Files and similar directories are accessed, the program doing it is basically just following a symlink/shortcut provided by Windows; all you'd need to do is use PowerTools/RegEdit to change where Windows thinks these directories are, and you'd have accomplished the same thing without so much work or partition madness. It's a reasonable idea though as long as you're sure your Windows i-Ram card won't be powerless for too long.Houdani - Monday, July 25, 2005 - link
It definitely has some potential, but I'm thinking this version is too limited. If it makes it to ver2 or ver3 we'll have something worth tinkering with, but I will pass on this go 'round.Captante - Wednesday, July 27, 2005 - link
For $50 I would have ordered one of these right away since I have 3 1gb DIMMS doing nothing in a box right now...for $150 they can keep it. :/kleinwl - Monday, July 25, 2005 - link
One point that would be interesting in knowing if something like the I-Ram could be programmed for, so that all paging out always uses the I-Ram (effectively increasing your ram) or commonally used files could be placed on it, something like a 8GB cache for a hard-drive.8GB is too small to put every file on (even though I would really like to)... but if the 80% most used files are automatically installed on that 8GB I think that most programs would see an increase in productivity.