Please explain why SATAII not faster than IDE (for my test)

[lastxj]

This isn't an issue, but rather a request for explanation
on some unexpected transfer rate comparisons.

Relevant Specs -
OS: XP
D1: SATAII, 160GB, 16MBcache, 7200rpm
D2: SATAII, 500GB, 16MBcache, 7200rpm
D3: IDE ATA100, 250GB, 8MBcache, 7200rpm

All are connected to mboard via SATA. Note - **For D3, I
use and IDE->SATA adapter that connects directly to the
IDE drive.**

Observation -
Using a single 700MB .iso image for a test file.

D1 > D2 = 31 sec
D2 < D1 = 20 sec

D1 > D3 = 28 sec *
D3 < D1 = 18 sec *

D2 > D3 = 22 sec
D3 < D2 = 13 sec *

Based on the specific test (ie., large single file), Disk 3 appears to be significantly faster.

My Question -
Please explain why, in this case, the IDE drive connected via an adapter is faster than a SATA II directly to mboard.

Thank You

 

[cdogg]

First off, I just want to say that physically, there isn't much difference between an IDE drive and a SATA drive. SATA has additional features like Native Command Queuing (NCQ), and in this case, a larger cache. But such features do not matter when you transfer a single 700MB file, which is why you aren't seeing any real benefit of SATA in this test.

___________________________________________________

You need to explain your use of < and >. "D1 > D2" mathematically means the same thing as "D2 < D1". I'm assuming that in your test results the first drive is the source and the second drive is the destination.

Let's take a closer look. The problem with interpreting your results is that you don't have a good point of reference to start from. You put an * next to the times you see as the fastest, because you were coming from the perspective of D3. If I approach it from the perspective of D2, these are the ones I would mark as the fastest:

D1 > D2 = 31 sec
D2 < D1 = 20 sec *

D1 > D3 = 28 sec
D3 < D1 = 18 sec

D2 > D3 = 22 sec *
D3 < D2 = 13 sec *

The results can be manipulated in this way to explain the following:
D2 to D1 is faster than D2 to D3 (this shows a little lag with the write speed on D3)
D3 to D2 is faster than D3 to D1 (this shows an issue with the write speed on D1)
D2 to D3 is faster than D1 to D3 (this shows an issue with the read speed on D1)


So the end result would tell me that the test was inconclusive. I show problems on D1 and D3. It would be better to test each drive independently of one another using standardized tests like the ones listed at this site:

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

You have to get a good sample of large and small data transfers to determine a drive's true performance capability. Don't put too much stock into a single file transfer. Remember that most drives will only average 50-60 MB/s transfer speeds on a "good" day. So even though SATA II bumps the interface of IDE up from 100MB/s to 300MB/s, it doesn't matter to most drives since they are still averaging well below 100MB/s.

You can widen the highway, but the speed limit doesn't change

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[wahnula]

It would be better to test each drive independently of one another

Good summation cdogg, but I would add that the only fair test would be (1) IDE drive using PATA connection, cable, and controller, (not just adapting SATA to IDE) vs. (1) SATA drive connected to a SATA cable & controller.

Then, run a series of HDD benchmarks (at least 3 times for each scenario) including Sandra and HDTach. This removes the variables of file reading errors and bus inconsistencies that can occur within the system when file copying.

Tony

Users helping Users...