Nforce4 Ultra-D/SLI-DR/SLI-D RAID Tutorial (updated first post)

[red930]

Try running a benchmark with HD Tach. You can google it for a download. Once you're done with the benchmark run, press and hold Alt and Print Screen. This will capture the active window.

 

Post the results for us.

[GNU]

Okay here is the first one with short run.

The second one with long run.

[cpuz]

Nice thread. Great guide!

[paulbaker]

For those of you that aren't experienced with setting up RAID, here's a quick, easy guide with lots of pictures to help you along.

 

http://www.angrygames.com/nf4raid-1.htm

 

feedback is appreciated!

 

 

I will possibly do a SI3114 RAID Setup if there is demand for one...

 

Nice guide.

Managed this all myself when I 1st set it up a few weeks ago, and had no probs at all,

But I am curiouse to know an ideal strip size as at present I am setup with 64k,

also can I change this without loosing (having to format) all my data, and is it worth the hassle, I guess what would be ideal is to do it in windows, I do have 2 spare 60gig drives if that's any help.

also when you say 3 drives would the 3rd drive have to also be identical to the others?

[red930]

CPUZ setup an 8X32 array and did see some performance gain with the larger cluster. This also means that even a 1Byte file will take up 8K of drive space. This leads to a lot of slack on the partition.

 

Three drives will all need to be the same type to get the best performance otherwise they will default to the slowest drive.

 

RAID-0 seems to scale well up to four drives.

[GNU]

CPUZ setup an 8X32 array and did see some performance gain with the larger cluster. This also means that even a 1Byte file will take up 8K of drive space. This leads to a lot of slack on the partition.

 

Three drives will all need to be the same type to get the best performance otherwise they will default to the slowest drive.

 

RAID-0 seems to scale well up to four drives.

 

And about my Benchmark. It's fast or not? It's too slow so i think.

[red930]

And about my Benchmark. It's fast or not? It's too slow so i think.

Sorry, I missed your post earlier.

 

Yes, that does look awfully slow and erratic. Now that we have a starting point we can go forward with trying to fix the problem.

 

I would try the drive in another computer if possible.

 

I would swap the SATA cable with the other one from the box the motherboard came in. You should have gotten two.

 

I would try another cable from another computer that you know works.

 

If all the cables give you the same benchmark results, I would download and run the Seagate drive utility to see if there is something wrong with the drive.

[Angry_Games]

Nice guide.

Managed this all myself when I 1st set it up a few weeks ago, and had no probs at all,

But I am curiouse to know an ideal strip size as at present I am setup with 64k,

also can I change this without loosing (having to format) all my data, and is it worth the hassle, I guess what would be ideal is to do it in windows, I do have 2 spare 60gig drives if that's any help.

also when you say 3 drives would the 3rd drive have to also be identical to the others?

 

ideal stripe size + ideal cluster size = you will have to test and retest and retest again to find what is right for you.

 

If you only have large files (1MB or more like ExRoadie) on the array, then you want the biggest stripe size + biggest cluster size.

 

if you use it for normal things, like surfing web, creating small text docs and such, then you probably want to go 16k/16k

 

i personally use 64k/8k as I don't really care so much about being the fastest dude in the west...I just want stuff to load relatively fast.

 

In my main machine (the NF3 250Gb is still my main rig at this time) I do not use RAID at all anymore. I use a single 36GB Raptor drive.

 

In my server, I have a single 80GB drive for OS/games/cd-dvd ripping etc, and 2x 160GB drives in a RAID-1 set (mirroring, so if one 160GB drive dies, I have an copy of it on a good drive, and can simply add another good 160GB drive in and rebuild the data from one drive to the second, thereby having 2x drives with mirrored info again!)

 

The NV controller scales very well with 3 drives, and even better with 4 drives.

 

The Si3114 is the slower controller, and does not scale well at all with more drives (nor does it have support for NCQ SATA II drives...they will work on the Si controller, but the features are not used).

 

The NV controller uses more cpu time, but the increase in speed is well work the extra 5%-8% cpu useage over the slower but more cpu-efficient Si3114 controller.

 

Please see this thread for more in-depth benchmarks that I have done with the NF4 and 2x, 3x, and 4x drives in RAID:

http://www.dfi-street.com/forum/showthread.php?t=4177

 

its a bit of reading, but by page 4 i think, the testing reallllly begins (but it is always good to read the entire thread to get an idea of what is on everyone's mind)

[paulbaker]

Thanks off to follow your link

[paulbaker]

Okay read your guide, and wanna try 16x16, but say i was to do the pe builder Post 12 by XPTEX thing to resize, would it then be possible to just then use PowerQuest Drive Image 7.0, to restore a backup image of my drive, there by avoiding the sp2 slipstreamed bootup issue thing,as linked out by thunda Here as I dont want to go through a full install atm, and I dont really know how you change them files,Post 2 by Big Poppa Pump as what I can remeber I used a small programme that done the slipstreaming for me

[GNU]

Sorry, I missed your post earlier.

 

Yes, that does look awfully slow and erratic. Now that we have a starting point we can go forward with trying to fix the problem.

 

I would try the drive in another computer if possible.

 

I would swap the SATA cable with the other one from the box the motherboard came in. You should have gotten two.

 

I would try another cable from another computer that you know works.

 

If all the cables give you the same benchmark results, I would download and run the Seagate drive utility to see if there is something wrong with the drive.

 

Thank you for the information but which utility? It's maybe Seatools (Desktop)?

And about cable test. See my post. I have tested with an Epox SATA cable and the yellow from the DFI (now my HDD is connect with a red cable). Not exactly the same bench but nearly.

[red930]

Thank you for the information but which utility? It's maybe Seatools (Desktop)?

And about cable test. See my post. I have tested with an Epox SATA cable and the yellow from the DFI (now my HDD is connect with a red cable). Not exactly the same bench but nearly.

Yes, the tools that Seagate has on their website for testing their drives. I know that you have tried two cables but the possibility does exist that both are bad. I had a guy go throug four cables before he found a good one. The problem was that he was bending the cable right at the connector causing them to give lots of errors. When he installed the last one he didn't bend the cable so tight and it has worked fine since then.

 

Just trying to cover all the possibilities. Hang in there!