RAID on nForce4: Worth it?

[Zak33]

Hi guys,

Well after the whole weekend of hell, trying to get my new SATA2 drives from Seagate to work on my Asus A8N-E, I'm there (or there-abouts) on an install.

I did a fresh XP install, just added Motherboard drivers, and nothing else, and then used HD Tachometer for a few benchmarks.
http://www.simplisoftware.com/Public...request=HdTach

I then installed XP on it's twin sister to confirm the speeds of both drives were equivalent (they were so close I was properly impressed at Seagate's consistency).

Then I set up a RAID array using those two disks, and ran the same thing exactly....just motherboard drivers, and nowt else.

Here are my results and you can decide yourself whether it's worthy of your time and money.

The Motherboard is an Asus A8N-E, with an Opteron 146 running at 10x250 HT4x =2500
2 gig of ram (4x512 sticks of Corsair Value Select)
2x Seagate 7200.9 drives, with Firmware versions of 3AAH (which I had to request from Seagate via email: the guys at Seagate always replied within 2 hours of my requests , so Seagate get 10/10 for effort here this weekend )
http://www.scan.co.uk/Products/Produ...oductID=323522

Single Disk:
Random Access Time : 13.6 ms (this is a lot slower than my previous 10k SCSI disks)
Average Read Speed: 64.3 mb/s (this is a lot FASTER than my previous 10k SCSI drives)
Burst Speed: 254mb/s (this is so fast I nearly expired at it!!...but it is right on BOTH drives )

RAID 0, 2 identical drives, striped. Option in nForce RAID setup for largest striping blokc of 128k for performance use)
Random Access Time: 13.9 ms (tad slower than single disk)
Average Read Speed: 126mb/s (now IL-2 will load faster )
Burst Speed: 196mb/s (big drop off but still waaaaay faster than my SCSI stuff was)

Now, I'm not sure if real life will feel faster, in games, XP, film work etc..but RAID is here to stay for a while, and if anyone wants to ask anything, fire away

It's been along weekend: thanks go to my missus as ever, fopr being patient beyond belief

[directhex]

remember that raid0 means "if one disk dies, ALL your data is dead"

[Paul Adams]

Indeed - there really isn't an "R" in "RAID0".
But there is an "oooooh", when it gives a noticeable difference.

Edit:
Make sure when doing synthetic benchmarks that you run the tests multiple times with the same configuration and take the average - did you use the short or the long test in HDTach?

[Zak33]

used the long test. Did it 10 times in each

Yes Hexman, I know...if one dies it's all dead

Luckily my PC only has a few critical items, and they get burned to CD once a month.

One of the reasons I wanted to stick with these Seagate drives, is that I've never had a Seagate fail and at the mo....I'm liking RAID

[Zak33]

if anyone else wants to download HD Tacho and run it, by all means do and we'll all compare

However...real life is where it counts...so games are gonna be installed next

[directhex]

remember, this is your CPU at work. nforce4 raid isn't raid - it's all CPU based. you'd get pretty good results on any fakeraid controller with a nice cpu like that at the wheel

[Paul Adams]

if anyone else wants to download HD Tacho and run it, by all means do and we'll all compare

3 SATA disks, all connected to the nForce4 controller on my Asus A8N-SLI Premium:
2x 74GB WD 10k rpm Raptors in RAID0 - partitions C,D,E
1x WD 7200rpm 250GB in third SATA port - partitions F,G,H

________________RAID0 Raptors_____Non-RAID WD_____
Random Access...........7.6ms..........13.4ms
CPU usage.................10%..............2%
Average.............129.3MB/s........54.8MB/s
Burst...............214.9MB/s.......122.5MB/s

Just ran the long test 4 times on each volume set.

However...real life is where it counts...so games are gonna be installed next

Time to logon prompt, time for initial load of desktop & apps, and loading timte for saved games or levels in games are my favourite indicators.
I was amazed at the loading and restoration of saved game time in Oblivion - completely at the other end of the scale from Doom 3 on my old system which took an eternity.

[Zak33]

remember, this is your CPU at work. nforce4 raid isn't raid - it's all CPU based. you'd get pretty good results on any fakeraid controller with a nice cpu like that at the wheel

troo

[latrosicarius]

Thanks for doing the benchmarks. That's very useful information.

I have some info too, but it's not statistical... I bought one of those expensive $350-dollar Areca PCIex8 RAID cards and hooked up 3 identical 300GB Maxtor MaxLine III drives in RAID-5.

Then for a backup solution, I bought a cheaper, XFX Renovo PCI RAID card and hooked up 2 more identical 300GB Maxtor MaxLine III drives in RAID-0.

So I have 600GB (actually more like 558 after formatting and everything) on each array.

I have about 400GB of data that I keep mirrored on each array. I can do a right-click > Properties on each of the roots and see how long it takes to scan the volumes and tell me how many files each one has.

Well the XFX array took like twice as long to scan the files than the Areca array. So basically, this just proves that the extra money spent on the Areca card was well worth it.

I knew that the XFX one wasn't going to be super fast, but its fine b/c it works well as a backup solution. BTW, I will NEVER use the motherboard's onboard RAID for storing my data ever again b/c it screwed me numerous times, on both nForce and Intel platforms. Always use some kind of hardware RAID card if you care about your data.

That said, I think it's still okay to use onboard RAID-0 for your Master drive, as long as you don't mind reformatting it and/or f-disking it if it craps out on you.

-- my 2-cents

[Zak33]

think you're right long term for pure power: as Directhex said, it's not hardware raid on my mobo, it's cpu driven.

Game loads are only about 10% faster (9.3%) on average for boot up of IL-2 from frest XP boot.

So not much better there to be honest. BUT.....I always wanted to try RAID.

Agent tells me I must Defrag frequnetly as the data is often dumped fast and all over the place but I defrag weekly anyway, so no fret there.

So far: DO I think it's worthy? Well, had it been an easy install, then yes. But it for ther SATA 2 firmware issues, it was a long laborious job, so currently....I'm on the fence

Its pretty fast, it was very good value and I'll live with it for a few months, and then maybe go Raptor

[Nox]

if anyone else wants to download HD Tacho and run it, by all means do and we'll all compare

Heres mine:

http://img.photobucket.com/albums/v2...Fox/hdtach.jpg

Nox

[Nox]

and yeah, i'd agree on the defrag comment - it goes everywhere all the time. I'm actually considering buying a 'proper' defrag program to run on idle. I've heard good things about disk keeper, tho there may be free recommendations out there somewhere?

Nox

[directhex]

shame the areca cards are over 500 quid new in .uk

then again, the lsi megaraid 300-8x is pretty much identical electronics on a fatter pci-x card, for those on a tighter budget

[Nox]

I was lucky - got it off ebay. Complete miracle really as I can't see these things cropping up much. It has been a complete <insert expletive here>er with linux though, everytime i've updated the kernel it doesn't work so stuck on an old one, which considering the kernels are meant to have support built in from .15 onwards I can't understand.

Nox

[Zak33]

Nox....mate....that is ROCKET FAST

maaaaan thats fast!!!!

I'm gonna post that piccy just here for peeps to see

[directhex]

I was lucky - got it off ebay. Complete miracle really as I can't see these things cropping up much. It has been a complete <insert expletive here>er with linux though, everytime i've updated the kernel it doesn't work so stuck on an old one, which considering the kernels are meant to have support built in from .15 onwards I can't understand.

Nox

yeah, that was the main reason i bought an LSI rather than the Areca, despite the added cost of PCI-X

now, how about some "hdparm -tT" figures to compare?