How much do access times matter?

ShadowTiger

Distinguished
Apr 19, 2010
14
0
18,510
Hi,

I've got a new hard drive that supports Automatic Acoustic Management (AAM).

I can only set it to 128 (quiet) and 256 (performance). I definitely can hear the difference.

When I run HDTune I see a difference in access time (16.2 vs 11.2ms, respectively) but my transfer rates and burst rates are identical in both modes (average 112 MB/s burst 231 MB/s...it's an SATA 6.0GB drive/controller).

So here's my question: am I going to notice any sort of real-world performance improvement running the drive at the 256 setting? (The system is primarily used for browsing, office, photoshop, etc.)

Thanks,
Peter
 

sub mesa

Distinguished
If its your system disk: yes. If its a download disk to store large files like .iso .avi .mp3 etc, then no you won't feel a thing.

Access times is the primary performance spec on your system drive, not throughput.
 

ShadowTiger

Distinguished
Apr 19, 2010
14
0
18,510
So, it is my system disk. What kind of performance improvement would I really feel? Are programs going to open faster, run faster? The system boot up and shut down faster? Are we talking about a second or two or dozens?

Thanks!
 

sub mesa

Distinguished
Increasing the access time by 20% would lead to a performance degradation when doing anything on your system disk (i.e. launching firefox) by the same ~20%.

Access time = performance (on your system disk)

throughput = performance (on your data storage disk)
 

ShadowTiger

Distinguished
Apr 19, 2010
14
0
18,510
Not that I didn't believe you, but I decided to run a quick benchmark to see what the effect actually is.


If you'll recall, I saw a ~30% decrease in access time (11.2ms vs 16.2ms). I timed Windows boot times (from when I first see windows logo to when the windows startup sound is played).

In silent mode I saw an average of 38.7s (the times were pretty consistent).
In performance mode I saw an average of 33.5s (again, consistent).

That is a performance increase, but it only cut ~14% off the times. It wasn't 1:1.

So, now I need to decide if shaving off 14% for disk-intensive tasks will be worth the noise (this thing is LOUD in performance mode). :)

Thanks!
Peter



 

sub mesa

Distinguished
Booting involves more things than just reading from your HDD; some parts are 0% disk bottlenecked. So the overall effect of your 30% less disk performance would not translate into 30% slower booting.

But, launching an application should be close to this, especially on a well used system with a decent amount of fragmentation.

However, if you prefer it being silent, then you may opt to just live with the lower performance. Its not like 20% faster or slower would change your experience all that much. Going to an SSD would be a huge step, can be 10.000% (100x) faster than HDDs in some realistic scenarios. But again, that would mean other bottlenecks would present itself; it would not mean things go 100 times faster.