www.supertalent.com The Official Super Talent Support Forum

FAQ MembersRegister

SandForce Controller base SSD 5 - How to choose the best SSD

SandForce Controller base SSD 5 - How to choose the best SSD

Postby SSD-James » Tue Apr 17, 2012 8:49 am

Now, I would like to review how to choose the best SSD in the market.
There are hundreds of SSD makers right now and it is hard to make a good choice of the reliable product.

Here's best tips for you.

1. Original SSD Makers - such as Intel, Samsung, Micron, Toshiba
1-1 Samsung started SSD building first with its own controller about 6 yers ago. At that time, it had a lot of problem at garbage collection algorithem and TRIM support function.
Of course, there was no TRIM concept at that time. So slow... But it was reliable.
Nowadays, its 830 series is perfect for the consumers. (end customers)
It supports Back Ground Garbage collection adn TRIM... It is fast too. Access time is fast too.
But it cannot be used for data center driven 'Steady State' purpose.
If GS Choi had been still in charge of Semiconductor business at Samsung, they should have made a good sense of "Steady State" driven SSDs... Long way to go for that server and mass storage market.

2. INTEL/Micron - They started SSD business a little later and especially Intel made a good SSD with its big capital. In a word, Intel made the first reliable SSD for consumers.
Intel used the "Underfilling' assembly technoloy and it gave us the reliability.
Performance was so good and unbeatable at that time.
It had the perfect Clean Status performance and fairly decent 'Steady State' performance.
Intel, as a market leader in the PC/Server industry, made a sensational progress with its first SSD.
But for the 'Steady State' driven application market like Mass Storage application, which has a lot bigger market size, its product line just cannot satisfy the performance.
So, they finally released SandForce controller base SSD product line such as 520.
It is good approach, I think.

3. Toshiba - It is THE ORIGINAL NAND inventor and it also has HDD technology.
So, people think Toshiba product should be reliable... It is. My kid are using this from 2.5 years ago and it just does not give him any single problem... Better than expectation.
It aslo does not fit into the 'Steady State' driven application use.

4. SK Hynix - It just started the SSD business but its response and technical knowledge about SSD is not much accumulated. Its SSD is based on SandForce (LSI) controller 2281...
To make up the late market entrance, it insteady used the fairly reliable technology.
It presort the NAND chips for SSD and it used the underfilling.
It only uses the Sync (ONFI) NAND chips so that its performance is best at the market.
Reliable and high range of performance.
The above Original makers' SSD is reliable and fast but expensive... about 20~30%
Let's review the 3rd party SSD makers:
3rd party SSD makers' market is blood fighting maketing to get the market share.
The only tool for that purpose is the cost reduction.
To reach that arena, 3rd party SSD makers have to use a little cheap NAND which should be used at USB flash drives...
As of today, to make the decent SSD with decent reliability, it is almost impossible to make it under $110 for 120GB SSD. But there are a lot of SSDs in the market offered at $120. or less.
It has a high possibility to pay the cost such as RETURN and REWORK... A Lot of embarassing works for customers...
Nobody wants to do that.

It has the early adapters' market and it has a fairly good brand for the SSD industry.
Oxx used to use only original good NAND chips at the early stage when I was working there.
I do not know the situation right now. But per the market complain about its SSD product lines, it seems like they use some cheap NAND to reduce the cost.
From my perspective, it is OK to use some cheap NAND for SSD but it should be tested and presorted for the defective ones. If they do that, it would be OK.
VERTEX3 uses Sync NAND chips and AGILITY3 uses Asynchronous NAND chips.
For the end customers, AGILITY is cheaper and decent reliable. Good enough.
Actually, there is not much difference between 2 products for the end customers. PC USERS.
Its latest product line VERTEX4 is based on Indilinx firmware on the MARVELL controller.
It is also good for the PC users...Not for "Steady State" driven applications.
Oxx is doing a lot of good works mostly due to Ryan Petersen, CEO. He is God of 'market driver'
He is one of the best CEO I have met. So aggressive and so brave on every matter.
If he were the Samsung president, Samsung would become $1T market cap company. hahahaha.
But I am not sure about 'longer term'

6. Cxxxxxx - Force, Force GT
Cxxxxxx started SSD business a little late. But it starts it with decent/half verified firmware.
That is good for Cxxxxxx because it has less return rate...
Most of issues are coming from the firmware update. So bad.
Cxxxxxx uses mostly good NAND chips so that they must have much less return rate.

7 Super Talent - TERA drive product line.
I only use GOOD NAND chips so that the return rate is so low.
It also knows how to test the sort out the possible defective SSDs at production lines.
It is experience base know how.
It has the STANDARD firmware and GOLD firmware both. But as I mentioned earlier, standard firmware is better for the reliability and endurance. I do not know why people need GOLD firmware.
I also actually think SATA II SandForce Controller is better than SATA III controller.
Reliability, Firmware, Compatibility tells me to put higher score on SATA II controller.

8. Super Talent - MX drive
It is based on JMicron controller 616. The reason why I am mentioning this product is:
- I have done a lot of hard test with this and that configurations... NO PROBLEM at all.
- JMicron has a long history of SATA controller business for more that 15 years so I believe its know-how is so much accrued and they put all kinds of scenarios on their chips.
The one thing I can tell you is this MX drive has a lot better COMPATIBILITY than any other SSDs.
Even better than SandForce.
It shows Un-Beatable comaptibility.
But the speed is a little slow. It is SATA II.
SATA II and SATA III - Is Performance a lot different? - The answer is NO... Absolutely no.

9. A-DATA - I'd better not make any comments for this maker.
10. Other TWN/CHINA SSD maker - same as above. Too many and I am not much comfortable with their operations on "what chips should be used"
I have checked a lot different makers' SSD but I am not much comfortable to recommend my customers.

Oh,... I forgot to tell you about the engineering points as a summary.
1. Original Makers' SSD - Better, but expensive. Very low return rate.
- Why?
- they use pre-sorted NAND chips for their SSD
- they use 'Underfilling'
- they test a lot before the shipping. It is actually good and bad.
2. SandForce Controller 2000 series and 1000 series both have potential heat problem.
- Is it good for longer terms? One of My kids use it at his notebook and it does not have issue until now.
- We are also testing both at our LAB computers as OS drives. So far so good.
3. Current SSD is mostly using 8KB page NAND chips and it will use 16KB page NAND chips later
- But I expect NAND makers ultimately build 4KB page NAND chips for SSD in a long run.
- Why? I would like to leave it a homework.
4. Firmware Difference does not make any sense to the real end customers.
- All of decent level of SSDs are now all good for the end customer use and application.
- No big differneces at all.
5. Buying criteria:
- SSD makers' reputation
- Consider the total cost including Return and Servicing
- If you want to use it for specific board and application, please contact the SSD makers' FORUM site.

For any kind of questions, please post it.
Next session: I plan to tell you more on "Application Specific SSDs", Mass Storage application SSDs.
It may take a time to prepare the story...

Posts: 2072
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