[thehybridkiwi]

I recently acquired a new Western Digital Caviar Black 640 GB with 64 MB cache and SATA III compatibility ( WD6402AAEX ).

The below results compare the performance of the hard drive when hooked up to the motherboard's SATA II port versus the SATA III port.

The hard drive is NTFS-formatted and empty of any content.

Notable differences:

• The burst speed is greater in SATA III configuration.
• The read speeds for file benchmarks are generally greater in SATA III configuration.

This post aims to provide some insight into the performance differences between using the SATA II port versus the SATA III port on your motherboard for conventional HDD's (that are SATA III compatible).

Benchmark (Read)


Random Access (Read)


Extra Tests (Read)


File Benchmark (32 KB)


File Benchmark (128 KB)


File Benchmark (1 MB)


File Benchmark (16 MB)


File Benchmark (64 MB)


File Benchmark (512 MB)


Additional Tags: plugged into attached to using compared to

 

[Haze80]

Not much difference huh... I wonder if the same holds true in a raid 0 setup?

 

[jchambers2586]

It should be because both drive have the same speed.

 

[Kramy]

I tossed your first two benchmarks into a spreadsheet.

Average 1% performance increase by running in SATA3 mode - not bad.

I suspect that a more modern drive that can push around 200MB/sec sequential would have a larger performance difference - but someone that owns one would have to run the benchmarks to find out for sure.

I didn't include the File Benchmarks because without exact numbers it's hard to be accurate. But looking at this 32K File Benchmark, it's easy to see the difference is far larger than 1%.

With the smaller File Benchmarks, what you are seeing is the effect of cache - the reads sit in cache, enabling the drive to perform very close to SATA2 or SATA3's theoretical limits. That's the reason you can get results like this out of somewhat slow 64MB cache 'Green' drives.

Knocks the socks off an SSD, right? No, not really. As soon as you bump up the size of the test set, you get more realistic numbers:

In your benchmarks once you reach 64MB test size the performance difference starts to diminish, because the drive doesn't have enough cache for the entire benchmark file. In the 512MB benchmark we can see that SATA2 vs SATA3 is very close, slightly favouring SATA3.

Small file benchmarks don't tell us much about real world performance, but there are a few select tasks that they may apply to - the autosave feature in Office, for example. If you're reading or saving the same file over and over, you're going to get the speed you see in in the smaller File Benchmarks. But most of the time the first two benchmarks in this thread are probably more relevant.

1%! Woohoo!

 

[rui-no-onna]

Quote:

Originally Posted by Kramy

Average 1% performance increase by running in SATA3 mode - not bad.

I suspect that a more modern drive that can push around 200MB/sec sequential would have a larger performance difference - but someone that owns one would have to run the benchmarks to find out for sure.

First, I'd like to say thanks for summarizing the benchmarks. As for modern HDDs saturating SATA 2, they're not quite there yet. SSDs can do around 250+MB/s sequential on SATA2. The 1TB VelociRaptor and modern high density drives are getting close to maxing out SATA2 but they haven't reached it quite yet.

 

[Kitarist]

Thanks for this benchmark really useful

 

[Kramy]

Quote:

Originally Posted by rui-no-onna

Quote:

Originally Posted by Kramy

Average 1% performance increase by running in SATA3 mode - not bad.

I suspect that a more modern drive that can push around 200MB/sec sequential would have a larger performance difference - but someone that owns one would have to run the benchmarks to find out for sure.

First, I'd like to say thanks for summarizing the benchmarks. As for modern HDDs saturating SATA 2, they're not quite there yet.

Not yet, but they're creeping up there. One thing to remember - HDDs are more bursty than SSDs, which have more constant throughput. HDDs send through a lot of data and then flatline, then send through another large batch of data. It won't have a huge impact on their performance, but as they get closer to SATA2's limits, the performance improvement will creep upward from 1%.
Quote:

Originally Posted by rui-no-onna

SSDs can do around 250+MB/s sequential on SATA2. The 1TB VelociRaptor and modern high density drives are getting close to maxing out SATA2 but they haven't reached it quite yet.

However, when SATA4 comes out (And who knows when that will be...!), I'm pretty sure almost overnight they'll have 1100MB/sec SSDs.

No creeping there - if there's a new standard, put out a new controller!