ideal env to test HDD speed

July 18, 2011 at 11:22:33
Specs: Windows SERVER 2008
Okay, recently purchased new SEAGATE SATA 3.0Gb/s which translates to
3 Gigabits/sec =>
3,000,000,000bits/sec / 8 = > 375,000,000
375,000,000 / 1024^2 = 357 Mbytes / sec.

Here are the tests I ran:

Athlon Phenom 3 core
4GB DDR2

- Large data transfer (64G) between two partitions in Linux Ubuntu 11.04: 12-13M/sec (initial few seconds 50-60MB/s after that rate steadily dropped. Roughly same result in Windows 7.
- Between two same HDD, surprisinly the rate was higher 30-40MB/s
- Used GHOST to clone the whole partition with data with 64G, took about 40mins which translates to about 27MB/s.

This is far below the manufacturer's spec, now I know the ideal transfer rate can not be achieved in real life but neither do I expect it to be this low speed either.

Here I am taking inputs from user people who have done similar test and I am especially interested in inputs of how to make the test environment as close to the ideal speed as possible.



See More: ideal env to test HDD speed

Report •


#1
July 18, 2011 at 12:30:47
NO!
The sata interface could in theory reach 3.0Gb/s.

The hard drive can not.

Max transfer speeds on a drive is posted in specs or on their web page.

1/3 of highway deaths are caused by drunks. The rest are by people who can't drive any better than a drunk.


Report •

#2
July 18, 2011 at 13:25:51
The maximum speed of the SATA2 interface is 3 Gb/s. That is the rate that data can be transferred between the hard drive and the computer. But there is far more to real world hard drive performance than interface speed. In the case of a conventional drive this includes internal transfer rate, rotational latency, and seek time. The interface transfer rate is usually the least important. These other factors reduce actual transfer rate to well under 3 Gb/s.

At the present time only a fast SSD can achieve this speed. Most are slower.


Report •

#3
July 18, 2011 at 14:06:44
I know, but if you guys read my posts carefully, i am already mentioned that I am aware of it. I was more interested in how to tweak (if there is any) the environment to make it toward ideal speed limit.

Report •

Related Solutions

#4
July 18, 2011 at 17:26:12
Have only 1 drive task running. No Windows.

Report •

#5
July 18, 2011 at 18:07:23
OK. You are correct in that I didn't fully read the post.


Ghost is so old I can't say how it can support newer stuff. Might be a command line switch for dma or other settings.


Run a hard drive diagnostic program from the OEM to decide if it is bad.

1/3 of highway deaths are caused by drunks. The rest are by people who can't drive any better than a drunk.


Report •

#6
July 18, 2011 at 20:49:21
If you copy from partition to partition you have to add track seeking and head switching everytime the buffer fills and it switches from read to write. If the partitions are on different disks then there is less track seeking. Add directory and FAT updating on the written drive. Reading only is always faster than writing. Spec speeds are instantaneous data rates, not continuous data rates.

Report •

#7
July 19, 2011 at 10:07:46
Okay, looks like the discussion is going in the direction I wanted. I just thought about it, perhaps using SSD drive will took away overhead related to head seeking, track etc.
But the overhead related to updating FAT etc still there. I dont know the FAT32 but did some work on Linux EXT3 which involves updating inodes tables etc. That is not counted as a real data transfer and eats away the performance. THanks,

Report •

Ask Question