hard drives & raid - benchmark and compare!

[Mayhem]

Right well i did all that and reinstalled windows and guess what - its the same - as above. There are very minenute diffrences. but over all it still spikes like hell. Maybe one of my leads is a bit crappy but cannot understnd why as the leads are brand new.

 

I may buy some braded cables later on to reduce noise but other wise if these came with my DFI and there not upto spec i think thats a bit rubbish as these MB DFI Lan Party SLI-Dr cost a dam fortune and dont see the point od supplying low grade cable with a High grade bord.

[Prostock62]

Hi Guys iam doing some testing with the size of my clusters and this is what i came up with.The HD Tack results barly changed with diffrent clusters so iam using the HD Tack for both.ATTO iam using a 64 Stripe and 4 cluster on the first shot.The second shot i used 64 stripe and 16 cluster.

Qusstion i realy dont know how to read ATTO.Can some one explane what is better a higher number on the Read & Write?

 

 

 

[red930]

Bad things happen to good cables.

 

I'll be honest, I replace all of the original SATA cables that came with my first DFI nF4 board. All of the recent builds I've done have gone off flawlessly.

if these came with my DFI and there not upto spec i think thats a bit rubbish as these MB DFI Lan Party SLI-Dr cost a dam fortune and dont see the point od supplying low grade cable with a High grade bord.

[dred]

Ok guys i moved everything over to NVraid.. Trueimage would have worked but blue screened cause it loaded the wrong driver for the controller.. =( anyway fresh install.. and ya big performance improvement. Why does the sil3114 have to be such a pos? having the extra SATA was one of the main reasons i paid extra for the SLI-DR.. =(

 

 

 

 

 

At least we get a nice compare of exactly how much crappier the sil3114 is compared to nvraid. =P

[Millzee]

Looks OK on ATTO I guess but the average HD Tach figure is a bit diappointing consdiering there are 4 disks.

[red930]

It's the nature of the beast! lol

 

The main use of the SIL controller is for RAID-5. The only problem is that since it's a "software" solution, the overhead keeps the bandwidth very low. While the data is protected you aren't going to win any benchmarks.

 

The SIL controller is on the PCI bus while the nV RAID is integrated into the chipset on the PCIe bus.

 

You can put SATA optical drives on the SIL controller leaving the nV RAID controller for hard drives.

 

Have you seen the benchmarks with four Hitachi SATA II drives on the nVidia controller. FAAASSST!

 

Why does the sil3114 have to be such a pos? having the extra SATA was one of the main reasons i paid extra for the SLI-DR.. =(

[red930]

By setting your stripe and cluster to the same size you are killing your performance.

 

See here for an explanation of RAID...

http://www.storagereview.com/guide2000/ref/hdd/index.html

 

Pay special attention to the RAID section on performance...

http://www.storagereview.com/guide2000/ref...perfStripe.html

 

Looks OK on ATTO I guess but the average HD Tach figure is a bit diappointing consdiering there are 4 disks.

[Millzee]

By setting your stripe and cluster to the same size you are killing your performance.

 

See here for an explanation of RAID...

http://www.storagereview.com/guide2000/ref/hdd/index.html

 

Pay special attention to the RAID section on performance...

http://www.storagereview.com/guide2000/ref...perfStripe.html

 

Changing it did improve anything though?

 

[red930]

You have four drives in the array. Try a 128K stripe with 32K cluster. You will have to rebuild the array to select the 128K stripe.

 

I've found that a 4to1 ration gives the best overall performance envelope. Try to keep your stripe size at four times the cluster size.

 

But don't be afraid to try a lot of different settings.

 

Most forum members have settled on a two drive array using the Hitachi SATA II 3.0Gbps drives. It offers the most "bang for the buck". Remember that the more drives in an array the greater chance of array failure. The spindle speed of 7,200RPM is a limiting factor in the array performance. There's still nothing like the Raptor for mind numbing performance. It's the cost that gets in the way.

Changing it did improve anything though?

[Mayhem]

I got my Raptors form ebay £180 inc P&P for 4 x 37 gig raptors and they were only two months old and still sealed not a bad price all though my cables are rubbish that i got with the DFI SLI bord (maybe just bad luck but would expect better from DFI).

 

I how ever followed the guides on here (thks all for the help BTW) and now they peak at the drive speeds on the reads but the wrights are way to low and i still get lots of spikes.

 

I have orderd 4 x braided sata cables costing £24.00 from over-clocker.com hopefully these will solve my problems. All though to get the speed you have to tweak your OS a little as well.

 

So far i have reinstalled over 15 times and seem to be hammering partition magic good style lol.

 

when i eveventualy get my cables ill re post my pics.

[red930]

Writes will always be slower than reads due to the overhead involved in verifying the data.

 

I got my Raptors form ebay £180 inc P&P for 4 x 37 gig raptors and they were only two months old and still sealed not a bad price all though my cables are rubbish that i got with the DFI SLI bord (maybe just bad luck but would expect better from DFI).

 

I how ever followed the guides on here (thks all for the help BTW) and now they peak at the drive speeds on the reads but the wrights are way to low and i still get lots of spikes.

 

I have orderd 4 x braided sata cables costing £24.00 from over-clocker.com hopefully these will solve my problems. All though to get the speed you have to tweak your OS a little as well.

 

So far i have reinstalled over 15 times and seem to be hammering partition magic good style lol.

 

when i eveventualy get my cables ill re post my pics.

[Mayhem]

well thes ar my stats atm

 

4 x raptors stripped for get what settings