Vertex 2 60 GB HDD slow/wierd

jewii

jewii

jewii

Associate
Joined
12 Jul 2010
Posts
334
well i've ran HD tune and i get this graph:
35n0xdz.png


while other reivews show this http://www.guru3d.com/article/ocz-vertex-2-ssd-review/9

i ran ssdlife just ot check and it says it needs replaceing and life is @ 33%, only ran it for a hour. only installed essential programmes. 50% free.
 
never mind, i was just been a fool. forgot to search forums first *DOH!* there was a thread nicely below with the same problem, achieveing nice speeds on atto.

but by the way. is there anyway to find out if trim is enabled because windows 7 didn't detect my SSD. i've disabled prefetch super fetch, defrag and page file anythign else to do, something about ready boost? also updated my sata controllers and intel chipset.
 
jewii said:
never mind, i was just been a fool. forgot to search forums first *DOH!* there was a thread nicely below with the same problem, achieveing nice speeds on atto.

but by the way. is there anyway to find out if trim is enabled because windows 7 didn't detect my SSD. i've disabled prefetch super fetch, defrag and page file anythign else to do, something about ready boost? also updated my sata controllers and intel chipset.
Click to expand...

DOUBLE *DOH!* a quick google search gave me the answer for trim, installed intel toolbox and checked it there, i do have trim enabled.

any other tweaks i can do?
 
Having read em number one on the list is not hitting it with benchmarking tools that put lots of data onto the drive;

To summarise what was said, if you put a lot of uncompressed data onto the drive, it naturally gets written onto it. This means to write new data the old benchmarking data must be unwritten; this obviously will slow the drive.

The sandforce controller compresses data where possible, uncompressed data from benchmarks does not work as well but real life this is not an issue.

If you benchmark your drive using some programmes, like assd you write a heck of a lot onto it.

The controller does not get time to catch up; if you are using the drive as an os drive it may never get chance.

Logging out of your windows and leaving a while will allow trim to work on an os drive. (Few hrs a week reccommended)

If you continue to benchmark using the wrong tools that write too much the problem will get worse. Some users get into a situation where they keep checking using benchmarks and naturally it keeps getting worse, so they keep checking, making it even worse.

Those are my impressions, they are totally played out by the fact the none recommended benchmark assd showed slow and wierd results for both my drives (intel and vertex 2), whereas atto was pretty much bob on both reads and writes and general consensus for both drives, even this though does not play out the bigger picture.
 
Last edited:
ATTO is what OCZ use to get the figures that they quote in their promotional material (mainly because it flatters their product). So not too surprised that ... "atto was pretty much bob on both reads and writes ".

OCZ Sandforce based drives are good at handling compressible data, hence why the results from ATTO look good.

Not knocking their SSD's in any way, I have one myself, but you do need to understand their limitations. I wish I had £10 for every forum thread that I've read (not just here) that says something like... every time I run CDM or AS SSD, the sequential write speed goes down. Err... well yes it would, because you are bombarding the drive with GB's of in-compressible data.

"Logging out of your windows and leaving a while will allow trim to work on an os drive. (Few hrs a week reccommended)"

Almost... The idle time processing that the controller does is actually what some folks call "Garbage Collection", though the Duraclass Technology phrase is re-cycling (basically the same thing ). And technically you don't need to be "logged off" for idle time Garbage Collection to work on Duraclass drives, any idle time will do, it's just that being logged off supposedly speeds the process up.
 
Last edited:
In theory all you need is idle time but where a drive has your os on it, logging off will help maintain a decent stretch of idle time. Probably only needed in unusual cases.

I totally agree atto is best case, but it was right on for my intel drive too and ocz arent really being cagey about this, nor does it seem to really matter real world.

It isnt so much as ssd gives slow readings (which it does), it is that after the test the ssd controller is left with a lot of work to do due to the tests putting a lot of data that doesnt 'suit it' onto the drive, it needs to take time to sort this out, if it hasnt caught up by the time you bench with as ssd again, you will obviously add to the mess.

The ocz forums inc their staff share the view that repeatedly doing this leads to slowdowns in drives over prolonged periods as a mess is made that the controller doesnt get chance to sort out; this is drastically compounded where people decide to do epic 'worst case' benchmarks with as ssd.

Bottom line is if the drive is reporting right in atto it will be working right- and thus performing fast, no need for anything else really.
 
Last edited:
el Kuarlos

Can't say that I really disagree with you much on anything you say. I've been a member of the OCZ forum for quite a while myself and very much respect some of the more experienced posters, especially a couple of the mods (who actually work for OCZ).

But sequential read / writes are but a very small part of the the performance characteristics of a drive... though you've got to admit that it's the one that so many people get hung up on!

"It isnt so much as ssd gives slow readings (which it does), it is that after the test the ssd controller is left with a lot of work to do due to the tests putting a lot of data that doesnt 'suit it' onto the drive, it needs to take time to sort this out, if it hasnt caught up by the time you bench with as ssd again, you will obviously add to the mess."

A fair summation. But the situation is immensely more complicated than this. For one, under certain circumstances (writing huge amounts of in-compressible data being one) the controller in the drive can actually "throttle" the drive speeds to protect the drive and try (in it's own mind) to maintain the life / health of the drive in general.
IE. it's not that it can't continuing writing at a higher rate, it actually cuts the write speed (to a certain degree) to protect the drive. All part of the incredibly complicated Duraclass technology in these drives. Even some of the genuine techies on the OCZ forum struggle to explain some of these things. If you start looking at issues like wear levelling, GC maps etc. etc. it's enough to make your head spin and I've been in the business for some 30 years.

Have a good Christmas and New Year.
 
Ha not much else to really say mike- fully agree, kinda wandered off on an as ssd tangent though, wonder if this applies to hd tune also as per ops question and example- its not mentioned in the ocz guide specifically but it does take a while to run and is clearly meant for platter hds...merry chrimbo and happy new year! :)
 
You must log in or register to reply here.
Back
Top Bottom