|
|
Post by Johnny Grenade on Oct 2, 2006 14:43:00 GMT -8
If you have a Maxtor drive, it comes initialized in quiet mode. This is to minimize the sound the drive makes. The penalty is that the seek times go up. End result is your drive is slower than it should be. Most PC's these days have fans that are significantly louder than the drive. There is a utility called AMSET that can be used to turn off this feature. Check out the difference it made to my hard drive.
Quiet Mode enabled (Default)
Sequential Writing : 18.63 MB/s
Sequential Read : 24.05 MB/s
Buffered Writing : 59.64 MB/s
Buffered Reading : 83.68 MB/s
Random Reading : 23 MB/s
Quiet mode disabled by using Amset utility.
Sequential Writing : 27.14 MB/s
Sequential Read : 26.28 MB/s
Buffered Writing : 68.03 MB/s
Buffered Reading : 98.26 MB/s
Random Reading : 23 MB/s
|
|
|
|
Post by Johnny Grenade on Oct 2, 2006 18:29:35 GMT -8
BTW, Here are the sandra2007 numbers.
Now Sandra
SiSoftware Sandra
Benchmark Results
Drive Index : 39 MB/s
Results Interpretation : Higher index values are better.
Random Access Time : 8 ms
Results Interpretation : Lower index values are better.
Performance Test Status
Run ID : X2-4800 on Monday, October 02, 2006 at 8:48:08 PM
Processor Affinity : No
System Timer : 3.6MHz
Operating System Disk Cache Used : No
Use Overlapped I/O : Yes
IO Queue Depth : 4 request(s)
Test File Size : 1.5GB
File Fragments : 1
Block Size : 1MB
File Server Optimised : No
Benchmark Breakdown
Buffered Read : 30 MB/s
Sequential Read : 44 MB/s
Random Read : 32 MB/s
Buffered Write : 67 MB/s
Sequential Write : 43 MB/s
Random Write : 35 MB/s
Random Access Time : 8 ms (estimated)
|
|
|
|
Post by unisol on Oct 2, 2006 18:52:52 GMT -8
Ineresting you should start a discussion on this very topic. You should try testing your hard drive performance with HDTach. Much better and more accurate and reproducable results then Sandra. There are three modes of operation of acoustic management. FAST/QUIET/OFF. Both my Maxtor and Samsung drives shipped with AM in FAST mode. Here are the results of my testing: ...snip... See post two below this one. The big thing that AM effects is the seek times, with Maxtor having very aggressive AM and practicaly sending seek time performance into the toiliet. With the Maxtor drive, I found that FAST was better then OFF, since FAST used less CPU time than OFF did. Samsung on the other hand, is known for very quiet and cool running drives, and their AM has little effect on the drives performance, most likely because it is not very aggressive since it does not need to be. These tests were all run with HDTach 3.0.1.0 from www.simplisoftware.com/Public/index.php?request=HdTach and were reproducable when I ran them. Sadly though, the Maxtor drive is in another PC now, and my Spinpoint is still with me in my current PC, which is a different build with a different controller. So, I can not test them to get comparible results from other software packages. |
|
|
|
Post by unisol on Oct 2, 2006 19:06:35 GMT -8
On a side note, when I was doing all my reading on this back in late March, I found that it all depends on when you bought your drive and what channel you bought it from. Drives that were destined for OEM usage and purchased as OEM packages, are more likely to have AM in the quiet mode then drives purchased for retail usage. This seems to be, because OEM packaged drives tend to be for OEMs like Dell, HP, Lenovo, Gateway, etc and their focus is on ergonomics, where as the retail sector such as nuts like you and I, tend to focus on sheer performance over ergonomics. Also, the time frame of your purchase dictates alot. Older drives were very noisy and also created alot of heat, this being compared to modern drives.
|
|
|
|
Post by unisol on Oct 2, 2006 19:10:33 GMT -8
|
|
|
|
Post by Johnny Grenade on Oct 2, 2006 19:31:08 GMT -8
Cool info. My drive seek time dropped >4ms when I turned it off. The first set of results I posted was with the /FAST setting. The seek times were initially calculated in the 12ms range. This is a pata drive, also udma mode was 4 with fast enabled and now udma-6 with it off. Im leaving it off...
|
|
|
|
Post by unisol on Oct 3, 2006 9:02:50 GMT -8
You should goto StorageReview and look around. They have an article about their testing methology. While HDTach is not anywhere near perfect in representing real world performance, it is far better then what you will get with Sandra, PCWizzard, Everest, or the likes of any of those programs. In a nutshell, they are junk for testing drive performance. Their results are not reproducable and swing back and forth. HDTach is a good tool to compare drives and performance changes from various tweeks, but like I said, it does not represent real world performance at all, but it has it's place in giving you the peformance results of tweaks and comparing the relative performance between two diferent drives.
|
|
|
|
Post by Johnny Grenade on Oct 3, 2006 9:52:09 GMT -8
Yeah. The benchmark proggies dont test the entire drive. Just a portion. So the numbers will vary if you are writing near the edges of the drive platter. I dont remember which is worse, inner or outer, but it will skew your results. ( btw:you can see that in your hdtach pics ) The hdtach is better than most probably.
A few years ago we were working with a Medical imaging company that needed a special system for archiving x-ray data in real-time. Basically they had to ensure as soon as the raw data came from the CT imager, it needed to be written to disk and processed simultaneously. FDA mandates that xray data has to be saved to storage immediately before any other operations so a system crash doesnt require a patient to be re-radiated. So anyway, they tested all kinds of drives for perfomance to meet their particular bandwidth and reliability needs. The data transfer rate varied according to the head position on the drive platter. Also error correction was utilized closer to the inner circle. Was interesting, but troublesome due to the non-linearity of these things. Also, the drive vendors play all sorts of games with the MTBF and specs to make their drives look better than they really are.
|
|
|
|
Post by unisol on Oct 3, 2006 13:39:57 GMT -8
The outer tracks are faster. The drives are CAV (Constant Angular Velocity), so they rotate at a constant RPM unlike early CD-ROM drives that were CLV (Constant Linear Velocity) where the RPM of the disc was constantly sped up as the heads moved from the out to the inner tracks to keep the data rate the same through out the reading of the disc (CD-ROM's actually start on the inner portion of the CD and read out). Also, the there are more sectors per track on the outer edge then the inner edge, so with the faster rate of surface area running under the heads, the performance goes up.
|
|
