I don't really know much about hard drives, but I do know that it uses sectors (or something like that) and I think that I can improve on that. My idea is to have a seperate, smaller hard drive (in the same HDD case), or a small amount (like 3 gigs) of flash memory that will have all of the information about the locations of all programs and other things are (like where all the bits of the program are). This MAY create faster access time, and searches would be a lot faster. This could also hold info for programs like Launchy that need a catalog to access programs and/or files. Like I said before, I don't know much about hard drivesand they may already use flash memory to hold a catalog of stuff. I DO know, though, that they don't use a seperate disk to store info.
I am thinking that if a seperate disk is used, it would be 10,000 rpm's +. I wouldn't really know what makes flash memory fast, but I think that faster access to the catalog would make it a bit better.
NOTE: If anybody has any questions about my idea, I would be happy to answer them to the best of my abilities, and if I don't know, I don't know. Hopefully, somebody will have an answer for you.
NEW HARD DRIVE IDEA!faster HDD access???
16 Replies - 16708 Views - Last Post: 01 June 2010 - 07:22 AM
Replies To: NEW HARD DRIVE IDEA!
#2
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 10:58 AM
Drives were just announced with 4 GB of flash as a kind of cache/swap space for frequently used files.
The problem with HDDs is that because of how data is written, information is not always contiguous and creates slow random read/writes. I don't see having a 0ms access time on the MFT (talking only about windows here...) doing much if the head still has to jump all over the place to access the data.
In the best case scenario, seek times may decrease a little bit. But most drives already have an average read seek time of ~4ms.
The problem with HDDs is that because of how data is written, information is not always contiguous and creates slow random read/writes. I don't see having a 0ms access time on the MFT (talking only about windows here...) doing much if the head still has to jump all over the place to access the data.
In the best case scenario, seek times may decrease a little bit. But most drives already have an average read seek time of ~4ms.
This post has been edited by Antiokus: 24 May 2010 - 12:48 PM
#3
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 11:38 AM
wait... so somebody already made my idea then?
#4
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 11:52 AM
http://www.engadget....-inch-momentus/
Isn't exactly what you were describing, but it incorporates both flash and platter tech.
And again, the flash on these drives don't hold the location data. That information is maintained (for windows) in the MFT. The flash is used for oft-accessed files. Which, IMO, is a better way since a really fast catalog wouldn't do much for drive performance or transfer speed.
Isn't exactly what you were describing, but it incorporates both flash and platter tech.
And again, the flash on these drives don't hold the location data. That information is maintained (for windows) in the MFT. The flash is used for oft-accessed files. Which, IMO, is a better way since a really fast catalog wouldn't do much for drive performance or transfer speed.
This post has been edited by Antiokus: 24 May 2010 - 12:28 PM
#5
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 12:24 PM
Why don't drive have multiple moving armatures instead of one.
You could cut latency in half, you had two.
You could cut latency in half, you had two.
#6
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 12:39 PM
Not sure that multiple heads would improve performance either.
As far as I know, HDD controllers have to read back the bits in a contiguous manner, so it's not really something that could be done in parallel. And if it could, it would probably require two controllers, a cache for each, and then a third cache/controller that would then have to "assemble" the data stream to send to the SATA controller/CPU.
But it seems to me that the overhead and additional cost to do that would make it unfeasible.
As far as I know, HDD controllers have to read back the bits in a contiguous manner, so it's not really something that could be done in parallel. And if it could, it would probably require two controllers, a cache for each, and then a third cache/controller that would then have to "assemble" the data stream to send to the SATA controller/CPU.
But it seems to me that the overhead and additional cost to do that would make it unfeasible.
#7
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 12:40 PM
What's wrong with just plain Solid State Drives? (SSD) They are just gigantic flash drives....
#8
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 12:44 PM
I wish I could afford a big SSD
#9
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 01:04 PM
#10
Re: NEW HARD DRIVE IDEA!
Posted 24 May 2010 - 01:21 PM
LetMeFinclOut, on 24 May 2010 - 08:04 PM, said:
You don't understand what I mean, so I've drawn a crude sketch.

Let's say top'left is arm1, top right is arm2 etc
Arm1 is reach the end of it's track of bits, next bits and on next part of file is on a different track.
Arm3 is already queued up to read it. Arm1 is ready to be queued for the next.
Arm2 and Arm4 can operating on a completely different file.
Or a different scenario to improve average seek times. Arm1 does to first quarter of the track, Arm2 the next quarter and so on.
#12
Re: NEW HARD DRIVE IDEA!
Posted 26 May 2010 - 08:01 PM
Wow No2, you killed that post!
#13
Re: NEW HARD DRIVE IDEA!
Posted 26 May 2010 - 08:09 PM
Chew-it Chew-it Chew-it Chaw...
Rah-Rah-Rah-RIP IT UP!
lol
Rah-Rah-Rah-RIP IT UP!
#14
Re: NEW HARD DRIVE IDEA!
Posted 29 May 2010 - 12:10 PM
hmmm... i think you are following me no2...
#15
Re: NEW HARD DRIVE IDEA!
Posted 29 May 2010 - 09:55 PM
Dark-Necklas, We've been stalking way before DIC!

New Topic/Question
This topic is locked


MultiQuote






|