How much faster is SATA really?

[CoonDawg]

I did a Google search on this and the only articles I could find were 3 or more years old. I'm just curious, I hope to buy a new HDD (a SATA, thanks to my new motherboard) and I'd like to know how much of a speed increase I'll find over an ATA100. I've heard "5%" but that was for hard drives years ago. I'm wondering what the reality is these days.

It annoys me to no end that we're spending all this time and money and research for processors that can do petaflops and we still have to wait 20 minutes for windows to load. I'd give one of my cores for a hard drive that can load faster.

[JLK03F150]

You can download & run HD Tach on your current drive, then compare it against various models. The difference in Average Read will be the most noticed improvement.

Here's a screenshot of my 7200rpm 60gb ide notebook drive compared against a pair of WD 74gb raptors. Burst and Average Read time differences are tremendous. Burst speeds are similar with only one raptor, but the Average Read is still much better than my old ide drive.

[EXreaction]

That chart looks so lopsided because those are raptors, in a raid array no less.

Most hard drives are simply not fast enough to fill even the ATA100 bus to the limit.

If you see a performance increase it is because of the drive itself, not being SATA over ATA.

[JLK03F150]

Yeah, I just picked a comparison for comparison sake. Raptors are fast & raid 0 raptor are really fast.
Tom's Hardware has a HDD Chart too.
http://www23.tomshardware.com/storage.html

[Peter M]

The same (mechanical) model of harddisk will be exactly zero faster in SATA than in IDE.

Why? Because the higher interface rate on SATA does nothing beyond marketing. The actual data throughput on today's desktop drives isn't fast enough yet to exceed even UDMA-100 interface bandwidth.

The only performance benefit comes in server-like load patterns when Command Queuing is being used.


Generally speaking, your complaint is entirely valid.

Twenty years ago, typical harddisk size was 20 MB, and average throughput was 0.5 MB/s.
Today we have 200GB and 50 MB/s, which is a 10,000-fold size increase, but only a 100-fold speed increase.

Graphical interface bloatware does the rest.

[dunbar]

Peter, I believe you are right, because the mechanical system remains one head per platter which is rotating at some common RPM. Same RPM and head count for SATA or EIDE = same throughput.

But lets not forget the sizes of the SATA cables are significantly smaller; this may lead to better air flow.

And the quantities of conductors involved drops significantly from IDE to SATA, which can lead to greater I/O reliability (ever break just ONE conductor in any IDE ribbon? What a BEAR to find 'root cause' on that system unless you have decent diagnostic tools).