780G + SandForce = Need advice

[poohbear]

huh? are u guys serious? Sandforce is gonna isolate ALL of the AMD consumers out there by not having its SSDs work w/ AMD chipsets?? That's a huge portion of the market, how does that make any sense? Im sure it works fine in AMD systems. im about to purchase a Vertex 2 and never for a second thought my AMD setup would be a problem, is this for real?

[FlyingFish]

I have trim working using a Vertex2 drive and a 790XT board using the standard MS drivers.

[_dangtx_]

Originally Posted by poohbear

huh? are u guys serious? Sandforce is gonna isolate ALL of the AMD consumers out there by not having its SSDs work w/ AMD chipsets?? That's a huge portion of the market, how does that make any sense? Im sure it works fine in AMD systems. im about to purchase a Vertex 2 and never for a second thought my AMD setup would be a problem, is this for real?

even if it didnt work, but the guy above says it does. if you go amd youre probably on a budget and will probably get a wd black or sammy fx for os drive on the cheap .

[Sagath]

I think there is some misconceptions in this thread about how TRIM works.

TRIM is OS and SSD Firmware based. It does not matter what chipset you use. If the drive has TRIM firmware, and you're running an OS that supports trim, you have trim, unless you're running a raid array.

Run the following at the command prompt: fsutil behavior query disabledeletenotify
A response of disabledeletenotify = 0 indicates TRIM is active in the OS. (Note that this only means the OS is trying to send the TRIM command to the drive, if the drives firmware does not have TRIM, the drive will not TRIM.)

The following are lists (at the time of writing this post) of OS's that support trim (stolen from Wikipedia):


Windows 7 Final release - October 2009

Windows Server 2008 R2 Final release - October 2009

Linux 2.6.33 2.6.33

OpenSolaris July 2010

FreeBSD Version 8.1 (Only for low-level erase)
Version 9.0 (Proposed)

Mac OS X
Probably in the future

Hope that helps to clear things up.

[sswilson]

Sorry, but that's not correct Sagath, it's close but you missed a step.

disabledeletenotify = 0 indicates TRIM is active in the OS. (Note that this only means the OS is trying to send the TRIM command to the drive, if the drives firmware does not have TRIM, the drive will not TRIM.)

This indicates that the OS is indeed setup to send the trim command, but does not say whether or not the chipset drivers are passing the command along to the drive. As it stands... only intel chipset AHCI drivers, and the default M$ AHCI drivers pass the trim command through to a SSD. Other chipset AHCI drivers (such as AMD & Nvidia) do not currently pass the trim command through.

Up until now, this wasn't a huge issue as the M$ default AHCI driver worked fairly well, although with a small(ish) penalty to performance (guessing around 10%). Sandforce changes this dynamic though as it appears that it doesn't play nice in the sandbox with the default M$ AHCI drivers.

The lack of trim command for AMD AHCI drivers is a well known issue that has to date not been addressed by AMD. In fact, when I sent an inquiry off to AMD tech support about the issue, the answer I received suggested that I contact the motherboard manufacturer..... not much help there as it's AMD who both license the chipsets, and provide the drivers to the motherboard manufacturers.

[Sagath]

Wilson, I suggest you doublecheck that. I have TRIM working on an AMD laptop with IDE commands, verified by AS SSD.

Edit: For clarity sake, since my answer was vague and misleading: Steve is right, the drivers are at issue with all non-intel Chipsets. TRIM will be passed, if you use the default M$ ACHI or IDE driver for passthrough and forgo the AMD/nVidia chipset driver. (the generic MS in-box ones named PCIIDE.SYS (native IDE/SATA) and MSAHCI.SYS (SATA AHCI Mode) are the ones you want).

[sswilson]

We'll have to get AkG to chime in here, but I suspect all that crystaldisk is checking is the disabledeletenotify = 0 tag. Everything I've read to date says that the trim command requires AHCI drivers to be installed, and that it will not work when the drive is in IDE mode.

That said, I could be wrong (wouldn't be the first time it's happened... ;) ), but with all of the fuzzy information out there on the subject, I'd have thought that AMD would have released some form of press on the subject by now if their chipsets had no problems with trim.

One of the problems with the whole discussion is that it's hard to quantify and/or test for degredation over time.

[Sagath]

Yeah, like I said we are both semi-right and semi-wrong. I wasnt too clear. I edited out the CrystalDisk to AS SSD to show I was benchmarking the drives before and after...

To clarify again: you can get it to work on ANY chipset that supports ATA commands, however you have to use the default M$ drivers for ACHI and IDE.

AMD isnt on the ball, and dropping it alot.

[sswilson]

Looks like there might actually be a MS tool to test for trim functionality.... Trim Test (Windows) . If I'm reading this right, it looks like it might actually apply a trim command, and then test to verify that the actual blocks on the drive have indeed been affected.

Unfortunately, it also looks like it might be data-destructive, so it's obviously not something a person would want to run on a drive with anything other than OS on it.... ;)

[Teboga]

Hello people,

So I've been searching and asking through the OCZ support forums to try and find an answer to this. All I got is that the SSD will work fine with either driver MS or AMD. TRIM is available on the MS driver only, and no answer for TRIM on the AMD part yet.

The people there are running their own tests and its seems the new catalyst driver works pretty well, even without TRIM, with no performance loss over several weeks. I don't know if the time tested was too short to fill the entire drive, or if it's ITGC is actually good and can keep it up to speed even without TRIM. So I guess I'm gonna stick to the AMD driver and see what happens.

OCZ is also developing a tollkit for drive maintenance, but we don't know when it will be released.