www.supertalent.com The Official Super Talent Support Forum

FAQ MembersRegister

SandForce Controller base SSD 3 - HD Tune and TRIM show

SandForce Controller base SSD 3 - HD Tune and TRIM show

Postby SSD-James » Wed Mar 14, 2012 9:22 am

This topic is to show the trimming feature using HD Tune.
HD Tune bench program is one of the best tool to show your SSD trim feature is perfect or how much perfect... as far as I know.
Graphical showing is 100 times better than numberical showing and analysis for 99.9% of people.

Now, let's see the samples I have tested for this experiment.
One is Class 2 base 120GB (See the 1st topic about classification.)
The other is old time most popular Indilinx controller base 128GB.

First, I would like to show the Clean status HD Tune bench results of each.
Secondly, I will show you the screenshot of HD Tune after FRAGMENTED
Thirdly, I must show you the same screenshot after TRIMMING.
1 samples.PNG
Samples tested for HD Tune and TRIM show
1 samples.PNG (123.14 KiB) Viewed 4516 times

2 Clean.PNG
HD Tune READ bench result of CLEAN status SSD
2 Clean.PNG (158.16 KiB) Viewed 4516 times

3 After Fragmeting.PNG
HD Tune READ bench result after Fragment using IO METER
3 After Fragmeting.PNG (184.31 KiB) Viewed 4516 times

4 After Trimming.PNG
READ speed come back? after trimming
4 After Trimming.PNG (187.97 KiB) Viewed 4516 times

With the sample 1, I have done a series of IDLE for 1HR, 15HR, 24HR. Then check the same HD Tune READ bench and it did not go up as some of you already know.
Meanwhile, Indilinx controller base SSD is decently working with TRIM feature.

As far as I remember the SandForce controller base SSD, TRIM was well working at old controller SFi 1222 with Micron L63B NAND chips which are 4KB page size NAND Flash chips.
Nowadays, every SSD makers have to use Micron L74A NAND flash chips which are 8KB page size NAND chips.
It may be due to that.
Fortunately, Indilinx controller does not support L74A and so the TRIM feature is well working.
To make TRIM feature work with 8KB page size NAND, it must be a heck of work for firmware guys.

Next session is about IO METER testing result.
I would like to show you "Meaningless CLEAN status bench result" and "Most Meaningful STEADY STATE bench result."

All of the marketing people of each SSD makers are using the 1st bench result as most of you know.
But, I would like to show you the real world meaningful Steady State numbers too.
It will be very interesting. At some points, this is why I strongly believe that there is no differences between GOLD firmware and STANDARD firmware.
I'd better say like this:
For the PC application, Standard firmware is definitely better.
For the mass storage users, it depends upon the application.
For the streaming audio/video file servers, it is definitely one of SSD application at first, Standard firmware is better.
For the data center server users, such as banking center, GOLD firmware is better.
But, data center users are not ready for the SSD. It is too much risky at this moment.
If we have FULL SAS SSDs, it is different story.
Some of third pary SSD makers already launched the SAS drive but it is actually not the FULL SAS.
It is bridge solution which is called as "INTERPOSER"
I'd better not use it. What for? Nahhhh.
I met HP server people to ask them about the interposer solution. Guess what?
Testing itself is OK. but will not use them...
That was why I started to study what SAS and interposer SAS are...
Posts: 2076
Joined: Fri Jul 16, 2010 12:41 pm

Return to James SSD Benchmarking

Who is online

Users browsing this forum: No registered users and 1 guest