:huh:
I was jealous so I posted my crappy results :(
Western Digital Caviar Black WD6401AALS 640GB
http://img686.imageshack.us/img686/3972/capturegm.jpg
BASTARDS!!!
Printable View
:huh:
I was jealous so I posted my crappy results :(
Western Digital Caviar Black WD6401AALS 640GB
http://img686.imageshack.us/img686/3972/capturegm.jpg
BASTARDS!!!
Well, I've done something to my setup that has borked the results.
Dunno what.
Average read has dropped to the mid 300 range and burst to the mid 2k's...about a 30% decrease.
I've also been unable to detect what firmware version my drives came with since they're shielded by the RAID controller.
I read in the OCZ forums that that Vertexs started shipping with the 1.4 firmware in October but it's possible that Microcenter has been sitting on their stock for longer than that, so it's conceivable mine are an earlier version.
Given my degraded performance and mystery firmware, I think I'm going to reinstall the OS with the drives just set up normally and see what's going on.
This time I'll also apply the recommended "tweaks" one at a time instead of all at once and try to isolate what's screwing things up.
I'm also going to set it up with the OS on one drive (the 30GB drive size is no problem for me) and put my browser caches and page file on the other.
I think that the biggest advantage (for my usage at least) of the SSD is the access time, and RAIDing them really does nothing to improve the already insignificant access time of a single drive.
D's chart shows exactly why I used to RAID...a mechanical drive's performance slows as you get to the inner sections of the spinning disk and by using the Intel RAID you could cherry pick the fast sectors for the partition and maintain a pretty consistent speed.
Since SSDs obviously don't suffer from this defect (phenomenon?), it could be that RAID isn't the performance enhancer it used to be.
We'll see.
Have a look at the matrix storage settings. Maybe the write back cache I saw disabled was something the drivers do on their own?
Also, I noticed a discrepancy between the long and short test in HD-tach. Not as big a difference as you'd have to have for that to be the reason here, but still, if you ran the short instead of the long, you'd notice a difference in average read, I think. My run of the long test shows an average read about the same as yours, in your first post in this thread.
I tried the tool posted in one of the links I posted, btw, it does most of the tweaks at once. It didn't do much. What I don't have is a block size of 64k or whatever they recommended.
And I've noticed the matrix software is easy to sink, maybe you need a reinstall of that.
/halping
My array is doing better since I started tinkering. I think I speeded up my boot time with like a second, even, but the write speed is still capped at around 150MB/s.
You may well be right about raid 0 not doing much for us, though, but I like striping stuff :( I'd like to think it helps with handling smaller files as well, my controller is supposed to do worse at that.
Oh, and ATTO's good for testing.
EDit: Also, if you ran a hell of a lot of funky benchmark tools, maybe you suffered performance degradation from writing to all blocks or something? Unlikely, that, but still.
Well, I'm not sure.
I did however luck out, both my drives are the 1.4 firmware.
When I installed the Vertexes (Vertexi?) I simply unplugged the two drives that made up the previous array and left them in situ.
I removed the Vertex drives and plugged the other two back in and the old array was up and running.
I then plugged one Vertex in at a time, converted it to a non-RAID disk, reentered Windows and initialized it and was able to see the hardware ID details, which confirmed the 1.4 BIOS of the drive(s).
I then used the recommended Sanitary Erase tool and wiped them both.
Supposedly, I'm now good to go for a clean install.
ckrit: I checked all my Intel settings and everything else I could think of and found nothing which explained the performance hit.
That tool you refer to just changes all the settings I'd already done manually, so I'm not sure how useful it is.
So, it's now time to take the plunge and reinstall Win 7...on a single drive this time.
See ya'll in a bit.
Stupid question, maybe, but do you break the array if shifting which connector each drive is connected to around?
I'm always very careful about connecting them exactly the same way if disconnecting them, but do I still need to be, or do these modern-ish controllers go by serial number or something?
Mine's an intel controller too, probably the next generation down from yours, or the same. ICH10R.
We have the same controller.
Doesn't matter how you reconnect them, the pertinent data is on the drive so as long as they're on the same controller as before, the array will rebuild itself.
I'm back in Windows, folks.
Here's what the drive does right out of the box- absolutely nothing done, did this right after Windows hit the desktop for the first time...
http://img.photobucket.com/albums/v7...er/Vertex1.png
Now, let's do some shit and see what happens...
Is this a single drive? Is this formatted? This seems almost slower than my single 640 gig :huh:
The burst looks awry, the average speed beats your wd black badly. Could be a bit higher though, but that was with no tweaking or anything.
The low burst is prolly cos the matrix controller driver/software aren't installed yet.
I think OCZ recommends to use the ms drivers for ahci, if you didn't go with raid mode. But you'll have seen that, clocker.
oooo I missed that. I assumed the red br was average. my bad NM
My drives are sitting in my hallway ATM my wife says I have to wait until the 25th DAMNIT!!!
Well, kinda...but too late.
I didn't install in AHCI mode and now it can't be enabled without bluescreening in BIOS.
I can't install the Intel software cause it tells me there's no relevant hardware.
So, I can either wipe the drive and reinstall in AHCI or go back to RAID.
I'm thinking RAID would be the best use of the space.
What would AHCI mode get me, anyway?