i680 RAID 0 Performance - is it bunged?

[neonflx]

P5N32-E 680
2 x 75 WD Raptor

[scarface]

weird how my sata2's beat out your raptors. wait, are you running raid-1?

[neonflx]

i m running them on Raid 0 (stripping)
i will set up my bad axe 2 board later on today after i get back from school and do a comparison

[dinos22]

do everyone a favour and run some real world test where raid would benefit performance and also one chipset vs. another

[scarface]

i m running them on Raid 0 (stripping)
i will set up my bad axe 2 board later on today after i get back from school and do a comparison

so how come it says 68.95GB's on your stripe?

do everyone a favour and run some real world test where raid would benefit performance and also one chipset vs. another

like what?

[dinos22]

run some 3DMARK PCMARK, VIDEO ENCODING TEST, ANTIVIRUS SCANNING, GAMES, TRANSFERRING FILES............i mean there is a heck of a lot of stuff you can do

[scarface]

well hdtach is quick, simple and universal. not everyone uses the same AV or plays the same games or uses the same video encoding software.

[OBR]

My testing ....

[dinos22]

and is UNRELIABLE and does NOT indicate real word performance.......you will see when you do the tests

[day187]

Abit IN9MAX
2 x 74 GB raptors

[scarface]

and is UNRELIABLE and does NOT indicate real word performance.......you will see when you do the tests

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

[neonflx]

sorry

my hd are 34 a piece my mistake

[virtualrain]

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

QFT! (This thread is not about comparing performance but about gaining consensus on whether an issues exists with RAID on 680i)

It definitely seems like something is throttling STR. If the HD Tach curve does not taper off in a natural curve and is flat all the way across, then something is getting saturated and/or throttling data transfer. The curve should naturally taper off as the heads move from the outer cylinders of the platers to the inner.

It seems like the majority of EVGA boards and at least one ABIT board suffer from this problem.

Some more Striker RAID performance curves would be helpful.

Perhaps there is a potential issue with 680i RAID performance depending on how the manufacturer implements it (BIOS/board layout/etc)?

[dinos22]

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

look like Lestat pointed out earlier that's what these "raid" controllers are supposed to do....all the peak measurements and other gimicks HD and other software do doesn't exactly translate into mindbogling real life performance hence why i'm telling you to do some real world tests

these raid controllers are not really true raid controllers but i hope you guys know about that right? if you want true raid performance you need to buy yourself a dedicated PCI Express 8x raid controller (EXAMPLE) with 2 or 4 Ultra320 15K SCSI drivers in Raid0. Only then will you actually get the hard drive performance that is good for 300 or greater I/Os per sec and some nice seek times but it will cost an arm and a let and it will be hot and loud like you wouldn't believe...yeah i've run multiple 15K scsi in the dual xeon days and i still build high end workstations with people demand peak performance in HDD speeds particularly

you guys are kidding yourselves with raid setups on any of these desktop motherboards....the only raid useful on them is Raid1 so you can at least protect yourself from a single drive failure and nothing else

what we need are solid state disks.....hopefully 2008 will bring some joy to HDD performance and main bottleneck in PCs today

[virtualrain]

Dinos22, again, this thread is not about the merits of on board RAID vs dedicated hardware RAID, nor is it a thread about the merits of HD Tach vs. other drive benchmarks, nor is it a thread about the merits of RAID vs. single disks, nor is it a thread about the pros and cons of different RAID levels... sheesh!

This thread is about whether the 680i implementation of on-board RAID by EVGA and/or other manufacturers is actually being artificially bottlenecked, capped, throttled, or otherwise buggered, by something.

[Grinch]

2 x 150 raptors on 680i motherboard:




The trick to getting all the speed is here:




go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..

[virtualrain]

2 x 150 raptors on 680i motherboard:




The trick to getting all the speed is here:




go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..

Thanks Grinch (BTW, isn't that an oxymoron? )

What are the default driver settings?

Can other EVGA RAID users check these settings and report back?

[Grinch]

Thanks Grinch (BTW, isn't that an oxymoron? )

What are the default driver settings?

Can other EVGA RAID users check these settings and report back?

everything is checked by default...and any 680i based board should be able to do this trick...

[scarface]

okay here's the difference with the Grinch suggestion on my rig.
Before

After

[virtualrain]

Scarface, your RAID 0 on EVGA is not as bad as most... As you've seen, most have a flatline at 110MBps. Since you are one of the few NOT using Raptors, I wonder if that has anything to do with it?

Never the less, the settings did improve your burst and CPU utilization.

[scarface]

i was actually thinking of buying a couple of raptors but i noticed they are Serial ATA150 & not SataII. maybe that's the problem?

[gr8golf]

Grinch's suggestion improved mine - thanks - it worked!

[dinos22]

Dinos22, again, this thread is not about the merits of on board RAID vs dedicated hardware RAID, nor is it a thread about the merits of HD Tach vs. other drive benchmarks, nor is it a thread about the merits of RAID vs. single disks, nor is it a thread about the pros and cons of different RAID levels... sheesh!

This thread is about whether the 680i implementation of on-board RAID by EVGA and/or other manufacturers is actually being artificially bottlenecked, capped, throttled, or otherwise buggered, by something.

i realise that and i am trying to tell you that the way you are testing doesn't exactly give you the right answers as people have posted intel matix figures which are more than double the figures of 680i but i bet you anything in real world hard drive intensive tasks you will not have double the performance difference.......i would even be surprised if there was any noticable difference in all honesty across the board......

[Grinch]

on a sigle 150 raptor it would take me 2:54 to unrar a 4.7gb file...when using raid 0...it takes about 1:32 to unrar same file..

[Sumanji]

My NF4 Mobo can do 150MB/s with old 74GB Raptors in RAID 0... this i680 performance is terrible!

No matter what I try I can never get an average read much over 100MB/s with my Raptors in RAID-0

The rig I had before it (P4 3.2C and Abit IC7) used to destroy the nF4 in HDTach...