How fast does an HDD have to spin to be as fast as an SSD?

[Smoblikat]

Hey guys, this is a fairly silly question, I know. But I was just wondering how fast a high density 2.5" HDD (velociraptor for example) would need to spin for it to have the access time an SSD does. I know it is completely impossible, but I just wanted to get some anecdotal opinions...........just somthing to take a break from work

1 150gb Vrapt has an access time of about 6ms, at 10K, an SSD has an access time of .1, so I theorize that an HDD would have to spin at 600K RPM to have an access time of .1ms.

 

[corkyg]

It could involve more than just spinning. The lateral movements of the arm on which the heads are mounted may not withstand the resulting vibration. My anecdotal sense is that it would self destruct.

 

[razel]

Got me thinking and researching fast motors. I would have thought it'd be in turbochargers for jets. No. The fastest motor in mass production is found in some Dyson vacuums as of 2009 spinning at 100K RPM. The fastest motor being used by people reliably is in dental drills spinning at up to 450K RPM. And the fastest ever is in an experimental lab since 2008 is spinning at 1 million RPM.

 

[BrightCandle]

Its actually hard to calculate. The seek time needs to be about 1/100th of what it is today, the data density needs to be about 4x faster but there is one aspect that is very difficult to match - the multiple channels that give SSDs their exceptional 4k read/writes. In order to match this a hard drive would need to really be made out of 8 independent disks.

The dramatic reduction in seek time would have to come partly through the arm moving considerably quicker across the disk and partly via rotation speed. My initial guess is you likely need about 10x faster rotation but with an arm seek that was perfect. In practice the arm movement across the disk can't be perfect so the rotation speed must be great than 10x (ie >100k rpm) and the arm speed likely needs to see similar gains as well.

 

[sm625]

Even the top tier SSDs can only transfer a couple megabytes during the time it takes to seek. The problem is random access. It takes an SSD basically the same amount of time to read 1KB from 100 random locations as it does to read 1KB from 1 location. On a HDD it takes at least 10 times as long, with NCQ.

 

[KingFatty]

No, better to just increase density of the platter. So you keep the 7200 RPM spindle speed and arm performance, and then you ask, how high would the density have to be?

 

[Smoblikat]

All of these are great answers, Si how about a 100K RPM drive, with a fiber optic arm, 10tb per platter, port and starboard attachments, and with turbo drive? Would that do it?

 

[KingFatty]

Wait wait wait.

You know how inkjet printers usually have an inkjet head that moves back and forth across the paper? That's like the arm of the hard drive moving back and forth to read data on the disc.

Now, there are inkject printers that have a row of jets spanning across the entire page. So, no back-and-forth scanning of the inkjet head, you just print all the ink across the whole page at one time, so it's like one page a second.

That's what they need to do for spinning discs. Just have one arm that spans across the entire radius of the disc, so it doesn't have to move, it just can read the entire radius of the disc simultaneously and focus on the desired part instantly, without having to move the arm.

So add that too, a high density disc spinning very fast with the "instant" arm for zero seek time.

 

[Smoblikat]

Wait wait wait.

You know how inkjet printers usually have an inkjet head that moves back and forth across the paper? That's like the arm of the hard drive moving back and forth to read data on the disc.

Now, there are inkject printers that have a row of jets spanning across the entire page. So, no back-and-forth scanning of the inkjet head, you just print all the ink across the whole page at one time, so it's like one page a second.

That's what they need to do for spinning discs. Just have one arm that spans across the entire radius of the disc, so it doesn't have to move, it just can read the entire radius of the disc simultaneously and focus on the desired part instantly, without having to move the arm.

So add that too, a high density disc spinning very fast with the "instant" arm for zero seek time.

 

[386DX]

A few months ago at work I had to destroy some old hard drives. Came across an old Connor hard drive that had dual actuators. In a drive like this you'd technically only have to spin the drive half as fast to get the same access time because the data should be at most half a rotation away from one of the read/write heads.