SATA PERFORMANCE?

  • Thread starter Thread starter Arnold Packer
  • Start date Start date
A

Arnold Packer

Gentlemen:

Is there a real world difference in the performance
of a SATA 1.5 HD and a SATA 3.0 HD?

I am aware of the alleged doubling of performance,
but having been deluded in the past by PC hardware hype,
your experiences would be enlighting.

T I A,
A. Packer
 
In theory, yes. But given the current drive technology, I'd say no.
Perhaps if they come out with a 15,000 rpm drive there would be. If you
can get a deal on a SATA 1.5 drive, go for it. Most SATA drives are 3.0.

Don
 
Arnold said:
Gentlemen:

Is there a real world difference in the performance
of a SATA 1.5 HD and a SATA 3.0 HD?

I am aware of the alleged doubling of performance,
but having been deluded in the past by PC hardware hype,
your experiences would be enlighting.

T I A,
A. Packer
Click to expand...

The "head rate" or the rate that bytes are written on the platter,
is 60 to 70MB/sec at the beginning of the disk, and maybe 40MB/sec
near the end of the disk.

Compare that to the cable rate of 150MB/sec or 300MB/sec.

The SATA cable rate is good for filling the memory cache on the
disk controller board. So for short data transfers, which go into
cache first, there could be an advantage.

But to position the head of the disk, in preparation for writing,
that takes milliseconds. So even though the short burst of data is
faster, the "transaction rate" or number of operations you can do
on the disk per second, is dominated by head movement issues.

For sustained transfers, the cache won't help, and all you can sustain,
is limited by the media rate.

So performance is a function of what you are trying to do. The
cache on the hard drive controller board is capable of handling
data at the cable rate, but only for short transfers. For long
transfers, the media rate is still the limiting factor.

So there can be no "doubling" of performance, in any useful sense.
If all you run is benchmarks, the "burst" entry in the benchmark
table may show double, but on average, operations on the disk
are not twice as fast.

If you want to get as much improvement as possible, try a disk with
a higher rotational rate, such as a Raptor. A higher rotation rate
helps with average rotational latency, and average rotational latency
is a factor in how many transactions you can manage per second.

The tables here, compare SCSI, SAS, and SATA drives. 15K RPM drives
are available in SCSI and SAS, while the best SATA I think, is still
a 10K RPM drive. "Ordinary" desktop drives are 7200RPM.

http://www.storagereview.com/Testbed4Compare.sr

And if you dig around, you may find a benchmarking article, that
compares the things you are interested in. In this one, I see that
the difference between PATA cabling and SATA, seems to be insignificant,
compared to the difference that RPMs of the media makes.

http://www.anandtech.com/showdoc.aspx?i=1799&p=1

Here is another article, where some of the drives are 3Gb/sec SATA
and some are 1.5Gb/sec.

http://www.anandtech.com/showdoc.aspx?i=2454&p=8

Paul
 
Gentlemen:

Is there a real world difference in the performance
of a SATA 1.5 HD and a SATA 3.0 HD?
Click to expand...

No, if comparing a modern drive (so drive performance isn't
the variable) connected via either SATA speed. If your
particular use is more multi-access, server-like, you might
gain some by having NCQ but in general the drive itself and
it's inherant RPM and size will matter most.
I am aware of the alleged doubling of performance,
but having been deluded in the past by PC hardware hype,
your experiences would be enlighting.
Click to expand...


No doubling, the difference is not enough to notice in use.
 
Perhaps you could gain some speed if you have 2 drives in raid 0 or 4 drives
in raid 10. and 16 MB of cash on drives and if the files are small.



Boba Vankufer
 
Back
Top