Vertex 2 system underwhelms itself to death

A

Atheist

Limp Gawd
Joined
Nov 27, 2005
Messages
272
Background

 I replaced my Intel X25M 80GB with a Vertex 2 90GB 3.5" unit (put my X25M in my laptop) around the beginning of March. I ran a few benchmarks and have a thread on the OCZ forums asking why my speeds after a week or two were half of what they were when the drive was brand new and was told that I was benchmarking it wrong (Sandforce throttles if you write too much to the drive in a short time). I went along with that for the time being, as the only way to regain "fresh" speeds right away would be to format via a secure erase (OCZ tool) and I'm coming up on the end of my semester and have several large projects to finish.

Newest Problem - endless cycle of uselessness

A few days ago, I noticed when I woke up that my previously sleeping computer was running and at a failed boot screen telling me that my SSD could not be detected. I dismissed it as a fluke at the time. This has manifested again this morning, but now I'm at a point of inoperability. If I turn the computer off and then back on, I can usually boot to my desktop. Once there, I can do whatever I want for a few minutes (say, browse with either Firefox 4 or Chrome) until the system becomes unresponsive. I can move the cursor around (spinning loading circle) but cannot scroll on websites, manipulate tabs in any way, or even do anything with Explorer like close a window, bring up the start menu, or even control+alt+delete.

After a minute or two of this unresponsiveness, the system will crash to BSOD with 0xF4. Upon reboot, the BIOS AHCI check will take up to two minutes, and may or may not detect the SSD (it does, however, always detect my SATA HDD and DVDRW). If the SSD is not detected, I get the boot error that a system32 file is missing. If the SSD is detected, Widnows will either run a check disk or boot to desktop. If it runs a checkdisk of the SSD, it will run to completion or near completion and then crash, reboot, and this whole cycle starts over. If the system boots to the desktop, I get the freezing problem already described, and the whole cycle starts over.

A number of users on the OCZ forums have described this problem. The mods there suggested a secure erase (good old go-to solution!), which one user did to no avail.

My computer has BSOD'ed or hard-locked more times in the last several weeks than I have shut it down. I've learned my lesson and will stick with Intel SSDs in the future.

Do you have any ideas on what's going on? I've stripped my system down to RAM, GTX 470, Vertex 2, HDD, SATA DVDRW, onboard sound is disabled, RAM is running 1333 instead of 1600, and I've unplugged all USB devices but my mouse, keyboard, and printer, but nothing has changed. I'm running Windows 7 x64 and the Windows AHCI drivers. SSD Firmware is 1.32
 
A few days ago, I noticed when I woke up that my previously sleeping computer was running and at a failed boot screen telling me that my SSD could not be detected
Click to expand...
Go in to the Control Panel > Hardware and Sound > Choose a Power Plan. Choose High Performance and you may also want to change the plan settings and make the computer never sleep.

The idea is to never let the SSD sleep/hibernate or lose power.
 
Why shouldn't I sleep my computer? If I didn't sleep it, I would just turn it off. It has time during the day to handle TRIM, so is there something I'm unaware of regarding Sandforce drives that I can do with my Intel drive?
 
Solution?

1.) RMA the drive. OCZ will send you a new drive.
2.) Sell the new drive with your Invoice.
3.) Use the money to buy an Intel.

Problem solved!
 
I had similar problems with my OCZ Agility 2, which was a replacement for a dead OCZ Agility 1 (died during firmware upgrade, while my 2nd Agility 1 flashed successfully using same procedure - par for the course with OCZ - at least customer support was responsive and gave me an agility 2!!!).

However, my Agility 2 still was detected in BIOS, even though critical OS files were damaged beyond repair. I realized after searching around that my Agility 2 had older firmware and that it would kill itself during sleep. A firmware upgrade and ensuring the hdd never powers off during sleep seems to have resolved the issue after doing a secure erase. I got lucky.

I think my next SSD will be an Intel - at least they are reliable. My two Intel 80GB X18-Ms still work flawlessly...
 
yeah, my Vertex 2 120GB was my first SSD, but it, honestly, is noticeably slower than my 64GB C300... both after a few months of use. I've gotten a 128GB C300 to use in my main rig, now, giving my 64GB C300 to my AMD dev build.
 
MagicMan84 said:
Solution?

1.) RMA the drive. OCZ will send you a new drive.
2.) Sell the new drive with your Invoice.
3.) Use the money to buy an Intel.

Problem solved!
Click to expand...

You know, I've seriously considered this.

I ended up doing a secure erase and reinstalling Windows. the performance still lags behind my Intel (just got a 5.7 Windows score on Windows 7 - my fucking Seagate HDD gets the same). The OCZ forum mods say not to benchmark the drive heavily because Sandforce will throttle if you write too much too quickly (then what the hell is the point of a SSD? No, go ahead and write that at rotating disk speeds! Thanks!) Vertex 2's are apparently not supposed to sleep (they keep telling people this) which is a crock of shit, especially for something so appropriate for mobile applications.

This burns me up. I wish I had heard all of these complaints before I bought my drive - I'd have waited for an Intel drive for sure.

:mad:
 
Atheist said:
You know, I've seriously considered this.

I ended up doing a secure erase and reinstalling Windows. the performance still lags behind my Intel (just got a 5.7 Windows score on Windows 7 - my fucking Seagate HDD gets the same). The OCZ forum mods say not to benchmark the drive heavily because Sandforce will throttle if you write too much too quickly (then what the hell is the point of a SSD? No, go ahead and write that at rotating disk speeds! Thanks!) Vertex 2's are apparently not supposed to sleep (they keep telling people this) which is a crock of shit, especially for something so appropriate for mobile applications.

This burns me up. I wish I had heard all of these complaints before I bought my drive - I'd have waited for an Intel drive for sure.

:mad:
Click to expand...
But Intel are like 50-70% more expensive. OCZ sucks nowdays it seems with SSDs, Corsair isn't a serious competitor (very mediocre performance, 2 year warranty) , Intel is way expensive (albeit most reliable and best performer). Only choice for middle to high end with a reasonable cost is Crucial I suppose. Well perhaps Kingston as well, but I think that's more lower-middle end

Disclaimer: I'm a SSD newbie and this information is what I gathered in the past week during my first foray into the world of SSDs. More than likely everything I wrote is wrong.
 
Last edited:
Atheist said:
You know, I've seriously considered this.

I ended up doing a secure erase and reinstalling Windows. the performance still lags behind my Intel (just got a 5.7 Windows score on Windows 7 - my fucking Seagate HDD gets the same). The OCZ forum mods say not to benchmark the drive heavily because Sandforce will throttle if you write too much too quickly (then what the hell is the point of a SSD? No, go ahead and write that at rotating disk speeds! Thanks!) Vertex 2's are apparently not supposed to sleep (they keep telling people this) which is a crock of shit, especially for something so appropriate for mobile applications.

This burns me up. I wish I had heard all of these complaints before I bought my drive - I'd have waited for an Intel drive for sure.

:mad:
Click to expand...

The fools on the OCZ forum that suggest to not use sleep are stupid fucking morons. I'm familiar with their suggestions and workarounds. Bottom line is I shouldn't have to modify any power settings to get their shitty drive to work correctly.

It took me weeks to finally sort out how to get my system to sleep without a 0xF4 error. My solution was to use a compatible AHCI driver.

My next drive will probably be Intel. Price be damned, at least they go through rigorous testing.
 
Atheist said:
Background

 I replaced my Intel X25M 80GB with a Vertex 2 90GB 3.5" unit (put my X25M in my laptop) around the beginning of March. I ran a few benchmarks and have a thread on the OCZ forums asking why my speeds after a week or two were half of what they were when the drive was brand new and was told that I was benchmarking it wrong (Sandforce throttles if you write too much to the drive in a short time). I went along with that for the time being, as the only way to regain "fresh" speeds right away would be to format via a secure erase (OCZ tool) and I'm coming up on the end of my semester and have several large projects to finish.
Click to expand...

Slowdowns are caused by the following:
1. Writing too much data to the drive in a short amount of time, aka, too much benchmarking.
2. Slow down from "regular" use. Basically your write speeds are reduced by half when writing incompressible data (video for example). There is nothing you can do to prevent this. It WILL happen.

The only way to regain performance in either case is a secure erase. You can prevent #1, but #2 cannot be prevented. See here:
http://www.anandtech.com/show/4159/ocz-vertex-3-pro-preview-the-first-sf2500-ssd/6

How fast you reach #2 is debatable. For me to reach #2 I would have written to the drive much more in less than 3 months. My drive was never more than 18% full and I still saw performance degradation.
 
Here's a brief history of my benchmarks:

Left is Vertex 2, right is Intel X25M. Both are in the same system, Vertex 2 is OS and for this screen shot the Intel is slave. This is just after I had installed the Vertex 2 back in early March.
WXkQK.png


This is the Vertex 2 2 weeks later.
9jVqx.png


Then I was told not to use anything but ATTO to benchmark the drive. This is ATTO 2 weeks after initial installation (same day as the image above)
zOwpO.png


After the OCZ forums suggested that I not let the computer sleep but instead idle at the login screen for a couple nights so that Windows can run TRIM, this is my ATTO score three days after the above benchmark.
9Y1sZ.png


I was told that my speeds would pick up if I secure erased and formatted, which, thanks to the SSD shitting on my OS, I did yesterday. Here's my new superfast benchmark.:rolleyes:
Tdy57.png



For comparison, this is my X25M in my laptop.
l87hb.png
 
Intel SSDs aren't magical, they also had their problems (including faulty firmwares that would make partitions disappear), performance also drops if heavily written...

Every "affordable" SSDs use MLC nand with its limitations, so there is a need to understand how it works if you want to use it without problems. A casual user may not need to, because a casual user will not do things like writing 10GB on the drive for a stupid bench that doesn't even show what a SSD is all about (response time), but an advanced user should read some articles and adapt usage accordingly.

Now, I'm not especially defending OCZ, I was active on their forum but I stopped because there was so much misinformation and stupid solutions by people who don't understand the underlying technology (like the "garbage collector" scam), but at least there is a real community, reactivity, and support, unlike, say, Crucial (I was active on the OCZ forum when I actually had a Crucial M225).
 
They explain that the drive knows what to clean up without trim, which is impossible (unless you use some Samsung controller that is NTFS aware). Without trim, the drive is considered full of data, and the only margin is what is kept away by the controller (the "missing" capacity).
 
^^ that makes no sense because when you RAID it loses trim & the drives function just fine.
 
MagicMan84 said:
Solution?

1.) RMA the drive. OCZ will send you a new drive.
2.) Sell the new drive with your Invoice.
3.) Use the money to buy an Intel.

Problem solved!
Click to expand...

Agreed, this is the best advice one can give for someone with a faulty Vertex 2 drive.
 
GC works independent of TRIM. in raid arrays this can be proven easily by before/after bench sessions.
 
Just repeating it doesn't make it true. I'd like to see a write test writing a significant amount of random data (more than what is set aside by the controller) without a drop in performance after a few GBs. Usually on the OCZ forum they suggest not to do this.
 
You must log in or register to reply here.
Back
Top