Show us your SSD performance

essenbe said:

I'm really the wrong one to ask. But, as I understand it the older Marvell chips, were slower and the drivers did not support trim. The newer ones are better and do support trim. That is the best explanation I can give. Maybe someone with more knowledge than me will come along and give you a proper reason.

Oki, but thanks for the input :)

But i now use marvell 9128 on my mobo with MSAHCI with TRIM and you guys said it would work. Why shouldnt it then work if the controllcard has 9128 chip if i use MSAHCI?

Mike - I tried to find where I read that the 9128 supported TRIM. I couldn't (at least after an hour or so of going through the "no support" links) so I went from a different direction. It could have been that I read the specifications from a motherboard manufacturer's specifications. (Which is responsible for programming the controller--Marvell just sells the chips.) Hmmmmmm. Could this be why Marvell is staying out of the fray?

I too have read many posts about TRIM not being passed through a Marvell 9128 controller. Probably the same ones. None of the "experts" have been able to explain how this was determined. (At least to my comprehension.) All seem to be regurgitating quotations from other discussions in other forums. It's snowballed into a negative simply because more posts have jumped on the bandwagon and the spin continues--All referring back to each other. (Just like this thread.) But nowhere, that I could find, offered the scientific method used to determine this fact. It's all here-say. Some of the "tests" don't make any sense and have nothing to do with TRIM or how it works. Others seem to expect a TRIM command to be acted upon immediately and when no "visible" change occurs, assume the command was dropped. There's simply is no way to prove one way or the other that the drive controller received the TRIM command. Unless someone can devise a way to monitor the controller to controller interface.

If we can concur that Anand Lal Shimpi can be considered an expert, you'll find on the Anandtech's website a description of what TRIM is and how it works. What I gather is the command may never be acted upon if the drive has un-used space. This would be disk drive controller dependant, of course.

"TRIM is an interesting command. It lets the SSD prioritize blocks for cleaning. In the example I used before, a block is cleaned only when the drive runs out of places to write things and has to dip into its spare area. With TRIM, if you delete a file, the OS sends a TRIM command to the drive along with the associated LBAs that are no longer needed. The TRIM command tells the drive that it can schedule those blocks for cleaning and add them to the pool of replacement blocks." (1)

From the tests that I've read about, users are running benchmarks, deleting files then running another benchmark or watching lights flash or not flash. With a drop in a benchmark results, users are assuming the drop was because TRIM did not work when in fact back-to-back benchmarks rarely report the same results anyway. The "disk light" test had me nearly on the floor from laughing so hard. The disk activity light does NOT report activity occurring on a drive. It simply illuminates when the motherboard controller reads/writes data from/to the buffer. There's no hard wire from the disk drive to the light. When/if TRIM is executed by the drive controller, the motherboard controller isn't involved and thus won't illuminate the light.

In any case, my opinion stands.

___________________________________________
1. AnandTech - The SSD Relapse: Understanding and Choosing the Best SSD

I haven't given up on the Marvell site yet. I know I read about their controllers and how "customers" can make use of the programming options in the chip. (Customers = Motherboard Makers) Is it just me or has everyone noticed the absence of drivers on the Marvell site? Where are all these driver updates coming from then? Users should be knocking on the MB makers' door, not Marvell. Your statement; "Well we know the MSAHCI works so HugeRaven has TRIM" fuels my thinking that TRIM works on the 9128 but the problem is with the drivers. Which started this whole mess in the beginning when I said the controller supported TRIM.

In any case, as you say, HugeRaven appears to be set.

Hopalong X said:

carwiz said:

Is it just me or has everyone noticed the absence of drivers on the Marvell site? Where are all these driver updates coming from then?

I noticed and posted a similar comment earlier today.

They are coming from Marvell but going to the retailers not us directly.
They do have drivers for their retail products but controllers on the moboards are wholesaled to ASUS, MSI, Gigabyte etc.

So we be......

Just so i´m clear then: the TRIM command doesnt get sent through the controllercard with 9128 chip and that is why it wont work because you have to use the marvell driver to get the card to work?

Or if the card works without having to install any Marvell driver it could use the MSAHCI with TRIM working?