OCZ Vertex 2 - SSD settings/tweaks

[Proday]

Any idea way I have this dissapointing write values (as ssd and crystalmark tests; atto is ok but as u sad dont believe in him)

I did all tweaks for win 7, and im using Ocz Vertex 2 and WD green 1TB! Ill try to unplug WD to see results!

Broo20000: do u use ide or ahci (iaStor,msahci or u put on micron controler on the board)?!...u hawe gr8 write numbers man!! Btw do u use vertex2 as a system disk or?!

[flamenko]

flamenko, Do you have any idea why does to the boot slow up when in addition to OCZ Vertex 2 100GB SSD (firmware 1.11; latest intel rapid storage driver) a WD VelociRaptor 300 GB is connected ?

With default Win 7 x64 AHCI driver, instead of latest IRS driver, the boot is fast with both OCZ Vertex 2 100GB SSD and WD VelociRaptor 300 GB connected.

I am going to guess this is BIOS related. I would carefully walk through your BIOS screen and make sure everything is as it should be because it seems like the BIOS is trying to recognize the HDD after the fact. Also, is the pause occuring before or after the initial post?

And Proday... Your answer is above. AS SSD and Crystal cannot obtain accurate results with respect to drives which utilize storage compression.

[bro20000]

Any idea way I have this dissapointing write values (as ssd and crystalmark tests; atto is ok but as u sad dont believe in him)

I did all tweaks for win 7, and im using Ocz Vertex 2 and WD green 1TB! Ill try to unplug WD to see results!

Broo20000: do u use ide or ahci (iaStor,msahci or u put on micron controler on the board)?!...u hawe gr8 write numbers man!! Btw do u use vertex2 as a system disk or?!

I use ahci and i did a new install of windows 7 x64! For better results in crystal mark use test data all 0x00(0 fill!

[Lu(ky]

Any idea way I have this dissapointing write values (as ssd and crystalmark tests; atto is ok but as u sad dont believe in him)

I did all tweaks for win 7, and im using Ocz Vertex 2 and WD green 1TB! Ill try to unplug WD to see results!

Broo20000: do u use ide or ahci (iaStor,msahci or u put on micron controler on the board)?!...u hawe gr8 write numbers man!! Btw do u use vertex2 as a system disk or?!

I was told by OCZ Forum that running crystalmark will make your drive lower its performance within time? Not sure why but ATTO shows correct numbers for you, same as mine..

Note newegg has Vertex 2 60GB on sale now AR $125.00

[bro20000]

Am looking to get a other 2 vertex 2 drives for raid have heard there amazing in raid!

[Ao1]

Any idea way I have this dissapointing write values (as ssd and crystalmark tests; atto is ok but as u sad dont believe in him)

The test data that AS SSD & Crystal generate is not compressible. A Sandforce based drive needs to be able to compress data to achieve its rated read & write speeds. If it can’t compress the data write & read speeds slow down significantly. The 0 fill option on Crystal allows the data to be 100% compressed. It can therefore read and write much faster.

In real life data will be somewhere in-between. Think of it like zipping a file. Some files can be compressed to reduce the file size significantly. Those files will work very well with a Sandforce drive. Other files are already in a compressed format. If you try to zip an mp3 for example it won’t get much smaller.

This makes it very difficult to compare Sandforce based drives with SSD’s that do not compress data because performance is directly related to the data that you are writing and then reading.

[Ao1]

I was told by OCZ Forum that running crystalmark will make your drive lower its performance within time?

Benchmarking a fresh drive gives better results as the block allocation and wear leveling algorithms have not kicked in. Running a benchmark generates a lot of writes and if you keep running benchmarks in quick succession performance will drop as the block allocation and wear leveling algorithms kick in. This will typically resolve itself over time, but regardless there is a difference between steady state and fresh state.

[Anvil]

+1 with Ao1,

Someday we'll get to the point where we bench the drives in steady-state. (at XS?, not so sure about that )

The SF drives are fast but real life speed will never be anywhere close to the advertised speed.
(real life data will never consist of 100% compressable data)

@Proday,
Your results does look a bit low on writes, the reads are just fine.
There are a lot of factors that could influence your result. (chipset, bios, onboard storage controller, OC, OS, ...., the list is long)

[Ao1]

Anvil have you ever tested with Microsoft SQLIO.exe? We need something that can be configured for a standard test that has a fair mix of compressible and non-compressible data, which mixes reads/ writes and really hammers the drive in a used state to be able to sort out the wheat from the chaff. SQLIO could do that but I don't if it compresses data or not.

[Proday]

Thank u all for explanations people! now with 00000 crystal mark is better! Bro20000 can u do defoult crystal test just to see youre writes in that case?!

[Anvil]

No, but it's on my list of things to do.

SQLIOSim does not require the SQL Server to be installed, should be doable for most people.

Link to MS

[Ao1]

No, but it's on my list of things to do.

SQLIOSim does not require the SQL Server to be installed, should be doable for most people.

Link to MS

Anyone should be able to run SQLIOSim but the problem is getting a universal configuration based on the same data.

If only there was someone that had every SSD known to man it would be possible to get a truly comparative set of results.

According to the link btw:
File creation
The SQLIOSim utility creates separate data files and log files to simulate the I/O patterns that SQL Server generates in its data file and in its log file. The SQLIOSim utility does not use the SQL Server engine to perform stress activity. Therefore, you can use the SQLIOSim utility to test a computer before you install SQL Server.
When you run the SQLIOSim utility, make sure that you specify the same file location that you use for your SQL Server database files. When you do this, the utility simulates the same I/O path as your SQL Server database.
You can enable the compress attribute or the encrypt attribute for the existing test files. You can also enable these attributes for the existing directory where the test files will be created. The corresponding options to enable these attributes are located in the Properties dialog box for a file or for a directory.

[bro20000]

Thank u all for explanations people! now with 00000 crystal mark is better! Bro20000 can u do defoult crystal test just to see youre writes in that case?!

Here the test!


[IMG][/IMG]

[Proday]

Tnx m8, now i see that ewerything is ok. I only dont know how -JJ- god twice higher write numbers on defoult cm test?!

Btw way u have so high core voltages for 3.6ghz on i7?! It should woark below 1.27xxxV

[Ao1]

There you go

Some types of files compress better than others. So performance will vary depending on the data being written or read.

The extract below is from Winzip site. Obviously Winzip and Sandforce are two different things but compressibility works on similar principles.

For example, most multimedia files already exist in a highly compressed state. The standards for these file types usually specify efficient techniques to compress the data they contain. Examples include, but are not limited to, graphic (picture) files (GIF, JPG, etc), music files (MP3, WMA, etc), and movie files (AVI, MPG, etc).

(Most people store these types of file on HDD and reading from HDD to SSD will be limited by the HDD anyway).

There are other file types that don't compress well. For example, certain types of encrypted data files, such as those used by home finance programs and some database products, will not compress very much.

By contrast, some data files (such as text files and picture files in the BMP format) can often be compressed by 90% or more; while some types (such as program files) are often compressed by 50% or so.

[Proday]

Tnx for explanations Ao1

[bro20000]

Tnx m8, now i see that ewerything is ok. I only dont know how -JJ- god twice higher write numbers on defoult cm test?!

Btw way u have so high core voltages for 3.6ghz on i7?! It should woark below 1.27xxxV

Have still gotta tune system i just put in a random number for the voltage!

Am still waiting for new parts to come next week!Mem and a new psu!

2 More OCZ Vertex 2E 60GB i will post the results!

[-JJ-]

If its any consolation the problem has also been reported with Intel drives, so it looks like it's a RST driver problem.

(Not really a fix but if you have another controller on your mobo you could put the hdd on it as a workaround until the driver gets sorted out).

Thx, I will use the default AHCI driver then.

[-JJ-]

Well I don't know whether its from my hangover that's killing me today, but I actually just did find a solution.

1. Plugged both SSD and HDD to the system;
2. Booted into OS with default MS AHCI driver.
3. Manually updated the MS AHCI driver to RST latest driver (control panel/device manager/IDE ATA/ATAPI controlers)
4. Restarted the OS
5. Boot went down with no delays!

Now its boots as fast as it did with just SSD attached and HDD gets initialized and all.

[Ao1]

Interesting. So you updated the driver only without installing the control center and it worked OK?

[-JJ-]

Interesting. So you updated the driver only without installing the control center and it worked OK?

Well I dont remember the sequence of actions for my first try, when I simply installed RST pack by running setup.exe without the control panel, but it didn't work anyway. The boot still slowed down and the HDD was not initialized, however with the above mentioned sequence everything works fine and moreover on the first boot after manually installing RST the OS while installing driver for HDD particularity indicated that WD3000GLFS has been initialized and driver installed.

[tiro_uspsss]

I should probably do this too...

[PnoT]

I was told by OCZ Forum that running crystalmark will make your drive lower its performance within time? Not sure why but ATTO shows correct numbers for you, same as mine..

Note newegg has Vertex 2 60GB on sale now AR $125.00

I ordered mine last week and it shows a delivery date of today but seeing as I do not have the product yet newegg price matched itself and gave me $14 off my original purchase. Sweet!

[eva2000]

+1 with Ao1,

Someday we'll get to the point where we bench the drives in steady-state. (at XS?, not so sure about that )

The SF drives are fast but real life speed will never be anywhere close to the advertised speed.
(real life data will never consist of 100% compressable data)

so true with sandforce ssds. Their speeds should come with disclaimer - * for compressable data only heh

[PnoT]

I finally got my OCZ Vertex 2 installed with Windows 7 x64 and the speeds are right on par other posts in here and 0fill.

I'm rather disappointed in the jump from my Raptor R0 array to the SSD. I was hoping to pick up a lot in loading my games and map loading but I can honestly say it's slim to nill.

Windows boots like a madman but I never turn off my machine but instead put it into sleep so that's 99% of my usage. I'm saving on power and heat so that's a plus I guess.