Perpendicular
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Perpendicular
Hitachi's said to increase platter density with a factor of 10
http://www.newscientist.com/article.ns?id=dn7226
and with less text:
http://www.hitachigst.com/hdd/research/ ... ation.html
Sorry if old - tried to search
http://www.newscientist.com/article.ns?id=dn7226
and with less text:
http://www.hitachigst.com/hdd/research/ ... ation.html
Sorry if old - tried to search
All the big HDD companies are working on perpendicular drive technology, so I wouldn't get too exicted about a press release with no hardware attached Within a year or so all the big players; Seagate, WD, etc, will all be releasing drives with similar technology. No one gets too far ahead in this game for very long. (actually other reports suggest the Seagate may be first in market with these.)ONEshot wrote:Hitachi is now my favorite company.
-
- *Lifetime Patron*
- Posts: 477
- Joined: Thu Feb 17, 2005 6:41 am
- Location: Nashville, TN
And as usual, there's always another hard drive technology just around the corner....
(yes, I know it's a THG link. but it's just a re-post of a press release.)
(yes, I know it's a THG link. but it's just a re-post of a press release.)
-
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
Interesting. I didn't know they were doing this to hard drives. I've seen reports of a similar technology being developed for use with optical technology, where they were looking at using the angle of reflection to transmit more than simply binary 0 or 1 in a single "bit" on the disc. I believe the article I read said they had a working prototype using a base-32 storage system. Sounds like these perpendicular storage drives will be using a base-10 data scheme. I bet it doesn't stay at 10 for long...
Yes Seal, the theory is that they will be both denser and faster. Since the data is stored in a smaller area, given the same rotational speed, the head will pass over a lot more bits per second. It's with advancements like this in mind that the future SATA specs were developed...they knew they'd never get a current technology drive to those kinds of speeds. (heck, current drives can't even make full use of SATA I's potential speed)
Not to me. That isn't going to make my future computer able to do anything my current computer can't do.
OTOH, an affordable 20gig microdrive makes it really easy to make a compact silent computer.
I'm looking forward to when an entire useful silent computer fits inside a wireless mouse which happens to also be a cellphone. (With a wireless keyboard and hires projection display at home.)
OTOH, an affordable 20gig microdrive makes it really easy to make a compact silent computer.
I'm looking forward to when an entire useful silent computer fits inside a wireless mouse which happens to also be a cellphone. (With a wireless keyboard and hires projection display at home.)
Rotation speed can't be reduced. Seek times are directly proportional to the time it takes for the hard drive platters to do one full rotation.Seal wrote:so... does that also mean that for the same RPM, drivespeeds will *theoretically* be increased 10 fold too? Either that or drivespeed can be greatly reduced and ensure the same throughput as current 7200rpm drives... and you know what that means guys! less heat + less noise! woowoo!
However, think about this: Currently there are 1" drives with 5GB capacity. 10x storage density would mean 1" drives with 50GB capacity (I think that could even be interesting for IsaacKuo). And I think there are 100GB 2.5" drives, so these could be up to a TB. Who'd need 3.5" anymore? Smaller platter diameter -> less vibration and also less power needed to spin it.
-
- Posts: 968
- Joined: Fri Jan 07, 2005 7:13 pm
- Location: Bristol, UK
- Contact:
Umm...that's exactly what I'm most excited about (i.e. an affordable roomy microdrive). Of course, today's 5gig microdrives are rather expensive.lm wrote:However, think about this: Currently there are 1" drives with 5GB capacity. 10x storage density would mean 1" drives with 50GB capacity (I think that could even be interesting for IsaacKuo).
I think there's a potential for reducing spin rates to favor reduced power consumption over fast seek times. As system RAM increases, more and more of the OS and applications and frequently accessed files will fit within RAM. Fast seek times would thus be less important, as the disk is usually only accessed for streaming of bulky data files (like HD videos).
And how exactly does your data get duplicated onto each 180 degrees? With twice the writes that have to go at least 180 degrees around (4ms).StarfishChris wrote:I don't know - you could set up a disk of half the normal capacity where there is only 180° of data - the other half is exactly the same, so seek times are doubled (or you can halve rotational speed).
Whether that's actually feasible or not, I don't know. If it was it would already be possible.
-
- Posts: 968
- Joined: Fri Jan 07, 2005 7:13 pm
- Location: Bristol, UK
- Contact:
Yep, that's the only problem with it. I don't know how the hard drives stores 'runs' of data (whether it writes as many sectors/clusters/whatever as possible), if it continues then write times will be doubled, if for some reason it waits to complete a full revolution then the overhead is marginal.
That kind of behaviour is much like a fragmented drive, except unlike fragmented drives your seek time is halved. That's probably better for applications where you do a lot more reading than writing (web servers perhaps?), but no good for media editing/PVR.
Now, if the drive had two heads for simultaneous writing (so you could write what was written to the first half, and write new data at the same time) then I think that problem is solved. Write times would be at worst the same as ordinary drives. My brain is a little fuzzy right now so that might be wrong... pick apart as much as you can (not of my brain, of the idea).
That kind of behaviour is much like a fragmented drive, except unlike fragmented drives your seek time is halved. That's probably better for applications where you do a lot more reading than writing (web servers perhaps?), but no good for media editing/PVR.
Now, if the drive had two heads for simultaneous writing (so you could write what was written to the first half, and write new data at the same time) then I think that problem is solved. Write times would be at worst the same as ordinary drives. My brain is a little fuzzy right now so that might be wrong... pick apart as much as you can (not of my brain, of the idea).
Seek times are less important if the data that you need to read (or write) is placed contiguously on the disc. Command queing and large caches also help. i.e. if you need to read four sectors place randomly on the disc, then you have to perform four seeks, and your fast seeking drive has an average seek of 4ms, and your slow seek drive has a seek of 12ms, then you lose 8ms per seek to the fast drive, or 32ms overall. However if your data was in sequence, you would still lose 8ms per seek, but you would only have to seek once; so the total time you lost to the fast drive would be 8ms rather than 32. If the slow seek drive took 2ms to read the data, and the fast seek drive 10ms, both drives have taken 14ms to deliver the data you requested.
And what does this matter? The data won't be any more contiquous with the new perpendicular drives than what it has been with the current drives. Why 7200rpm drives seem to be more popular than 5400rpm drives if the seek times don't matter? And the quietest 7200rpm drives seem to be just as quiet as those 5400rpm drives.|Romeo| wrote:Seek times are less important if the data that you need to read (or write) is placed contiguously on the disc. Command queing and large caches also help. i.e. if you need to read four sectors place randomly on the disc, then you have to perform four seeks, and your fast seeking drive has an average seek of 4ms, and your slow seek drive has a seek of 12ms, then you lose 8ms per seek to the fast drive, or 32ms overall. However if your data was in sequence, you would still lose 8ms per seek, but you would only have to seek once; so the total time you lost to the fast drive would be 8ms rather than 32. If the slow seek drive took 2ms to read the data, and the fast seek drive 10ms, both drives have taken 14ms to deliver the data you requested.
Imo physically smaller hard drives would be the way to go. There was this "silent future is 2.5" wide" thread on the forum which talked about it. Now, why not make the drives physically smaller with same rotation speed instead of same size but slower rotation?
that flash animation is the coolest sad thing i've ever seen..
2.5" (and smaller) drives and 5,400rpm drives will exist on this technology but i dont see any chance of them becoming standard
i dont see this 'data written twice' (each 180*) ever happening.. they aren't going to sell a drive capable of 4TB as a 2TB drive
2.5" (and smaller) drives and 5,400rpm drives will exist on this technology but i dont see any chance of them becoming standard
i dont see this 'data written twice' (each 180*) ever happening.. they aren't going to sell a drive capable of 4TB as a 2TB drive
Not according to Seagate. They see the 3.5" drive going the way of the 5.25" HDD within a couple of generations. Their Savvio drives are the tip of the iceberg, so to speak. Once the data density gets high enough, there's no reason not to go to a 2.5" drive. They're smaller and consume less power. In the space required to mount 3 3.5" drives you could mount a dozen 2.5", and still produce less heat. Plus the reduced platter size has the potential to increase performance over 3.5" platters.mb2 wrote:2.5" (and smaller) drives and 5,400rpm drives will exist on this technology but i dont see any chance of them becoming standard
Even if all these would have a serious impact on the importance of seek time, I don't think hard drive evolution would ever go towards slower seek times.|Romeo| wrote:I'm going to suggest WinFS, improved defragmentation algorithms, more sophisticated command queing and larger caches will lead to data being more both contiguous, and drives being able to read the data in a more contiguous manner.
As already said, smaller diameter will give you all the benefits you wanted to achieve with slower spinning.
it's also worth mentioning that it goes both ways - for a 7200rpm drive, the time spent (on average) waiting for the desired sector to come around only accounts for about half of the seek time. the rest comes from the actual head seek, and will take about the same time on any drive, so going up or down one step in RPM won't have a huge effect on seeks, just latency.
also, does this development mean 10x the linear density alone? it doesn't look like it entails any decrease in track width, so that'd mean 10x the STR of current drives = SATA II could be fairly easily saturated with one 5400rpm drive and no port multipliers, heh.
also, does this development mean 10x the linear density alone? it doesn't look like it entails any decrease in track width, so that'd mean 10x the STR of current drives = SATA II could be fairly easily saturated with one 5400rpm drive and no port multipliers, heh.
Probably I'm being exceptionally stupid here; but I fail to see how a drive spinning at 7.2K will have less (or the same) power consumption for the same platter size than one spinning at 4.2K, or the same spin up/spin down performance.lm wrote:Even if all these would have a serious impact on the importance of seek time, I don't think hard drive evolution would ever go towards slower seek times.|Romeo| wrote:I'm going to suggest WinFS, improved defragmentation algorithms, more sophisticated command queing and larger caches will lead to data being more both contiguous, and drives being able to read the data in a more contiguous manner.
As already said, smaller diameter will give you all the benefits you wanted to achieve with slower spinning.
Seagate confirms they are working on this also and hopes to be among the first introducing it to the market.
http://www.pcworld.com/resource/article ... RSS,00.asp
http://www.pcworld.com/resource/article ... RSS,00.asp
-
- *Lifetime Patron*
- Posts: 477
- Joined: Thu Feb 17, 2005 6:41 am
- Location: Nashville, TN
Here's another press release on the holographic technology I mentioned earlier (minus flash animation goodness ) Looks like they're going to beat perpendicular to the market, if only by a little bit.
I'm not sure if you are talking about the same thing I am. I never said a 7200rpm 3.5" drive would use less power than 4200rpm 3.5" drive.|Romeo| wrote:Probably I'm being exceptionally stupid here; but I fail to see how a drive spinning at 7.2K will have less (or the same) power consumption for the same platter size than one spinning at 4.2K, or the same spin up/spin down performance.lm wrote:Even if all these would have a serious impact on the importance of seek time, I don't think hard drive evolution would ever go towards slower seek times.|Romeo| wrote:I'm going to suggest WinFS, improved defragmentation algorithms, more sophisticated command queing and larger caches will lead to data being more both contiguous, and drives being able to read the data in a more contiguous manner.
As already said, smaller diameter will give you all the benefits you wanted to achieve with slower spinning.
If we compare a 3.5" platter and a 2.5" platter, both 7200rpm... The smaller one can do with smaller bearings, which give less friction and need less power to spin. Just compare the power usage of any current 7200rpm drives of sizes 3.5" and 2.5", with equal number of platters per drive. You will see that the smaller one needs less power.