View Single Post
  #34 (permalink)  
Old February 14, 2011, 09:54 PM
AkG's Avatar
AkG AkG is offline
Hardware Canucks Reviewer
Join Date: Oct 2007
Posts: 5,032

Originally Posted by draemn View Post
Benchmarking was not the problem. I only used AS-SSD once and already had horrible results... Ran it a second time to confirm. That's how I found out that I had messed up and needed to "sanitary erase" my free space. Ran it again after that and got great benchmarks. I'm pretty careful about not killing my limited writes :)

Not 100% sure what you mean by "tony TRIM" unless you are referring to the CCleaner's "securely erase all free space."

After a bunch of reading online (before I used CCleaner), I was reading a couple articles where they did some testing and found that if you select the "single pass" option, it doesn't necessarily restore full performance and you need to use a 3-pass (sigh, lost a couple more write cycles... but it was worth it).

Either way, I know I was getting slow performance, so that's why I benched it with AS-SSD and I know that doing this 3-pass nuke on the free space restored performance and didn't touch any of the data that is supposed to be on there. The only logical answer to me is that the "free space" was not empty due to not having TRIM enabled during certain erases (i.e. quick format of windows install, non AHCI mode) and causing the slowdown. Haven't tried to benchmark it since or erase free space, but not having any real performance issues, so no need to.
Tony over at OCZ is who coined the term Tony Trim (its covered in the FAQ above...the one you should have read before posting in this thread). While it is a two part process (and the second can vary in how you do it) what you are doing is basically a simplified Tony Trim (second part only)...or that is at least what it is colloquially called. HOW you do it is of no importance as 1s are 1s.

The reason you have to do it multiple times (sometimes) is because SF compresses ALL data that is written to it (remember they claim a .6x write amplification) even when you think it has written 1s to ALL the really hasn't as its pretty darn easy to compress data that consists of all 1s into a small'ish pkg.

As I said earlier. SF drives do what is called Lazy House cleaning. This is done on PURPOSE as the NAND they use is (usually) the lower (medium) grade MLC good for 5-8K writes instead of (the high grade) 10K MLC NAND. My guess is the controller tries to reuse as many of the "dirty" blocks as possible so it can further reduce write cycles (in addition to less writes from the compression)...but it is only a guess.

Just be careful with the heavy duty nukings.

Originally Posted by s2g-unit View Post
Thanks a lot for the detailed explanation.
NP. Happy to help.

Originally Posted by birthdaymonkey View Post
Is it really the case that the only AMD chipset to support trim is the 890? I was under the impression that the AMD AHCI drivers didn't support TRIM with AMD chipsets but the MS ones did.

After reading AkG's FAQ, I did some more research on this topic to see if the general consensus had changed since I did research leading up to my purchase of a Sandforce drive using an older AMD chipset, but I could only find some confusing indications that TRIM might be working with the latest AMD drivers and the 8xx chipsets. Nobody seemed to be suggesting that the MS drivers didn't support TRIM on AMD chipsets.

This post by QB the Slayer on this AMD forum page suggests that TRIM has been tested working on the 790 chipset using the MS driver but not the AMD driver as of Oct. 2010:
AMD Processors Forums - AMD AHCI Driver SSD TRIM Support

So it seems that TRIM can work with AMD systems as long as you stick with the MS driver. Of course, maybe there is other evidence that suggests otherwise... I did find some information which suggested that TRIM (working or not) doesn't really do much with SF drives anyway.

Confusing stuff.
Yup it is confusing, but AMD has stated that the only SB that they make which is guranteed to pass the TRIM command (with their own drivers and assuming all other necessary conditions are met) is the 8 series. Does this mean that the 7 cant pass on idea. All we no is that if you want guaranteed TRIM support with AMD rigs you need a 8 series mobo. There is so much conflicting info on MS drivers + 7 series that it could even be on a case by case basis or only some 7 series support it or its a case if ITGC kicking in and making it appear as if TRIM is "working". I personal don't intend to find out as the 7 series is getting long in the tooth. ;)
"If you ever start taking things too seriously, just remember that we are talking monkeys on an organic spaceship flying through the universe." -JR

“if your opponent has a conscience, then follow Gandhi. But if you enemy has no conscience, like Hitler, then follow Bonhoeffer.” - Dr. MLK jr
Reply With Quote