Surprising T43 HDD Benchmark Results!

[JHaislet]

I think the results were quit surprising! They were performed on a T43 (2687DSU) and the drive tested was the drive it came with.

Hitachi 80GB, 5400rpm, 8MB cache (HTS541080G9AT00)

Sandra 2005 SP2 HDD Benchmark Results:
http://img57.imageshack.us/img57/6430/h ... ark1no.jpg

I was completely surprised by the Hitachi 7K60 results (60GB 7200rpm 8MB cache) as this is the EXACT hard drive that comes with the T43p's. While Sandra doesnt say what the test platform was (if it was even a notebook or what speed), I would assume that any modern notebook would be able to accurately "drive" any 2.5" form factor HDD as the HDD has been the limiting factor in notebook performance for many years now.

Also, this is was run with a relatively clean install of Windows XP (~25GB of used space for programs & such). The IBM Recovery partition was deleted and one huge ~75GB partition exists on the disk. Also the IBM Shock Protection program WAS running and the benchmark was done on a solid surface (desk) with Centrino Hardware Control running the CPU @ 2.0GHz with the video card downclocked by ATI's PowerPlay.

[FTC]

Hi,

Well, there are a couple of things to say.
1. First, Sandra posted result for 7K60 is clearly mistyped. I've got a 7K60 in my T40 and get Sandra scores in excess of 34 (when clean and just installed). No way you get 19 with that drive *unless* the disk is almost full and badly fragmented.
2. Sandra is a *very bad* test to measure hard disk performance, since it is too dependant on the fragmentation level of the drive, the amount of free space,... it really measures your 'file system performance'.
3. Your results are in line with what should be expected. In a T42 with the same drive (but hard disk filled up to 60%) I get a score of 27. Only suprising/wrong number is the estimated seek time (10ms). This is due to your disk being almost empty. 10ms is the real average seek time of a 7K60. the Hitachi 5K80 has a real seek time around 12ms.

[JHaislet]

Yeah, I too think it's rather strange. I think your correct in that it's a mistaken value for the 7K60. The "19" was provided by Sandra as one of their drives to compare my results to.

Part of me though, tends to believe most of the scores for my drive as it opens stuff like Firefox and Word 2003 almost as fast as my desktop w/SCSI drives.

Regardless, I've been pleasantly suprised by the performance of the 80GB 5400rpm Hitachi drive. I thought it would be painfully slow and had actually planned on upgrading to a 100GB 7200rpm drive (seagate or hitachi), but I don't think it'd be worth the minimial speed upgrades. That money would probably be better spent on a 400GB USB 2.0 external drive setup. That way I can keep enough free space on the 80GB so as to not slow it down.

[alpuzz]

Hi,

3. Your results are in line with what should be expected. In a T42 with the same drive (but hard disk filled up to 60%) I get a score of 27. Only suprising/wrong number is the estimated seek time (10ms). This is due to your disk being almost empty. 10ms is the real average seek time of a 7K60. the Hitachi 5K80 has a real seek time around 12ms.

so how does "real seek time" relate to actual results? I ran HD Tune on my 5k80 and got the following: Is my seek time lower than it should be? I'm using the preload but have disk keeper set on a smart schedule to keep the drive defragmented:

HD Tune: HTS541080G9AT00 Benchmark

Transfer Rate Minimum : 16.6 MB/sec
Transfer Rate Maximum : 36.9 MB/sec
Transfer Rate Average : 28.4 MB/sec
Access Time : 16.7 ms
Burst Rate : 79.2 MB/sec
CPU Usage : 4.0%


--Al

[FTC]

Hi,

HDTUNE seems accurate. Generally, access time is defined as the time it takes to position the read/write head at a given spot. This translates to :

ACCESS TIME=SEEK TIME + ROTATIONAL LATENCY TIME

The seek time is the time it takes to move from the original track to the destination track... this is my previous figure of a published average of 10ms for 7K60, 12ms for 5K80.

The rotational latency is the time it takes for the head, once over the destination track, until the desired sector is under it, which on average equals the time it takes to do half of a turn for the platter. This is 5.55 msec for 5400RPM drives, and 4.16 msec for 7200 RPM drives.

So, in your case. Your access time should be around 12+5.5 msec, or about 17.5 msec. HDTUNE got you 16.7, which is close to 5% from the theoretical (17.5) data (acceptable error margin).

[alpuzz]

Thanks very much for your informative response, I learned something new today

--Al