• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Are my 2 Raid-0 WD Raptors performing up to snuff?? It doesnt seem like it....

Flakk

Senior member
I'm running two Western Digital SATA Raptors in Raid-0 on my P4P800 Deluxe motherboard's on-board Intel Raid controller.

Attached are some images of benchmarks I have run.. If anyone has a similar setup, can you compare to what you are getting for benchmark scores?

HD Tach benchmark
Sisoft Sandra benchmark
ATTO benchmark

How my raid is setup (BIOS photo)

I thought Raptors were faster than this?? I had my my hopes up, expecting scores almost double what I am seeing.. am I expecting too much?

Input greatly appreciated

--edit-- added photo of my bios to show how the drives are configured.. properly i think
 
Mine reach 130MB/s Read 90Mb/s Write. What you have there looks like a single Raptor. Maybe you didn't setup the RAID0 right? Thats my only guess because the drives should be "up to snuff"

Maybe you did RAID1 instead not sure. What stripe size did you use? mine is 64kb or 32kb can't remember which. Ya might have it too big/too small. Ya might wanna try updating the drives for the raptor too.

BTW i find sandra's HD bench to be unreliable for me. Try to find ATTO drive benchmark or something that tells you write and read.

Does it at least feel likes it;s 10000rpm?
 
Originally posted by: OverVolt
Mine reach 130MB/s Read 90Mb/s Write. What you have there looks like a single Raptor. Maybe you didn't setup the RAID0 right? Thats my only guess because the drives should be "up to snuff"

Maybe you did RAID1 instead not sure. What stripe size did you use? mine is 64kb or 32kb can't remember which. Ya might have it too big/too small. Ya might wanna try updating the drives for the raptor too.

BTW i find sandra's HD bench to be unreliable for me. Try to find ATTO drive benchmark or something that tells you write and read.

Does it at least feel likes it;s 10000rpm?

Cool... i havent tested mine yet... since i havent installed windows... but your stripe is 32k? did you leave your cluster size alone or did you change it too? i have my stripe at 16k and cluster at 4k... i get conflicting advice that changing my cluster to match the stripe should increase my score but some say it's wrong and should just leave the default 4k alone... i dunno which one is right. 😕
 
Originally posted by: OverVolt
Mine reach 130MB/s Read 90Mb/s Write. What you have there looks like a single Raptor. Maybe you didn't setup the RAID0 right? Thats my only guess because the drives should be "up to snuff"

Maybe you did RAID1 instead not sure. What stripe size did you use? mine is 64kb or 32kb can't remember which. Ya might have it too big/too small. Ya might wanna try updating the drives for the raptor too.

BTW i find sandra's HD bench to be unreliable for me. Try to find ATTO drive benchmark or something that tells you write and read.

Does it at least feel likes it;s 10000rpm?
No, it doesnt really feel like 10000rpm.. at those speeds how could it :\ And yeah, I double checked, it is set up in Raid-0. The stripe size I used was 128kb. See the photo I took of my bios. I'll look for some sort of driver updates but I don't know what else I can update..

I'm seriously begining to think it is my motherbaord (the P4P800).. because when I raid-0'd two 15gb maxtor 7200rpm scores on the OTHER raid controller (the on-board Via Promise Raid) it also scored very low scores.. infact, lower than my other computer's SINGLE 40gb 7200rpm 2mb maxtor drive...

I just plunked down $270 for these two drives.. 🙁 I really would like to know whats going on here

OverVolt: what kind of controller are u using? the on-board one from your P4C800?


Originally posted by: LED
Well they look alright. I'm running @ 2.2GHz on an Abit NForce 2 MoBo with NTFS in RAID0 on SI chipset and here are my scores :
HDTach
SiSoft
Raptor Single
Yes, your scores are closer to mine.. but still they are better. you get an average of 80mb/sec on HDTach, while mine is around 54mb/sec. Quite a big difference actually....

And then I hear about the great success stories such as OverVolt's, where he gets over 100mb/sec and I cant help but feel a little jipped
 
ok i just added a photo of my bios setup and my ATTO benchmark results...

How my raid is setup (BIOS photo)
ATTO benchmark

why the heck is the 'write' so much more than the 'read' in that ATTO photo??? I thought the reading was supposed to be faster..
and the write goes up to 163mb/sec??! holy sheit thats fast.. lol?

can someone else put up a screenshot of their ATTO scores with two raptors in raid-0? especially u OverVolt plz
 
I see your stripe size is 128K which is fine for larger files (ie. Vid), however, smaller stripe size increases the speed of the RAID Array
 
the raid program suggested 128kb for a home user.. and i figured since my files are all mostly above 128kb, they would almost all be split up between the two drives (and as you know, files less than 128kb would be stored and accessed by only one drive)..

i mean mp3s, computer game levels/maps, and many other files are all more than 128kb, so my hopefully not failing logic would indicate this to be ok.. but feel free to give your view and opinion on this area.. i know strip size is a pretty debatable area.


i just ran sisoft benchmark again and somehow scored around 63mb/sec... about 10 better than before (didnt change anything).. this is a little better, but still not like other people's 100mb/sec success stories
 
No I think you have it set right after reading your list of data. You might try it ewith the USB2 Drive not running or USB off although logic states that your MoBo passes the PCI bus but I'm curious as well because of that fact you should be running faster then me...Did you defrag the Drives?
 
This is what they are capable of in an optimal environment.

Please set your test length to 32MB in ATTO. The reads are running high because the test data is being fetched from the drive's 8MB buffer instead of the recording media.

Smaller stripes will show better throughput in HDTach. This often translates to better performance in everyday desktop use. For pure STR sensitive applications such as NLE, 128 is fine.

Using the test platform that produced the above results, 8KB stripes were found to be ideal.

EDIT: Your BIOS shot reveals the drive SN! You may want to blur this!

-DAK-
 
Your motherboard has been in tests with other similar boards and was considered the best of the lot. http://www.hothardware.com/hh_files/Motherboards/springdale_showdown.shtml I was planning buying this board for a new system and running s-ata raid on it as well.Now I am not sure whether to cough up the extra 75 cnd for the asus p4c800e-dlx.even though i have studied the specs its hard to see the difference between them,could anyone tell us what advantage their would be on the newer of the two.?....As far as your scores go keep tinkering and check for any firmware and driver updates,you have a screaming system and should be ok.
 
i updated my ATTO benchmark with a 32mb test file size.. it does seem a lot more accurate now. The 'write' doesnt fly up to 160mb anymore. Rather it stays more consistant with the 'read' now.

Also, i blurred the ol' serial numbers.. thanks for the heads-up shuttleteam.. i didnt even think of that as being a big deal

i will get back into this discussion in a few hours when the company I have over leaves..
be back in a bit
 
Originally posted by: 440BX
Have you read this article http://www.anandtech.com/storage/showdoc.html?i=1799
no, i havent.. but i will now. Sorry for ignoring u that first time, i was in a hurry. Thanks for that link. */me reads away*

Originally posted by: StraightPipe
Sandra usually has you defrag the drives multiple times and then reboot before it tests. did you do this?
didnt defrag... but i had just installed windows on it freshly.. not much there to fragment
 
I must thank you -DAK- for your clear and thoughtful answer.. it very helpful to me in seeing how the performance should be with these drives. One thing though, you said the screenshot results were from an 8k stripe size.. but it looks like they all say 4k in the pictures? is this correct? I am surprised at such a small stripe size, but if it gets results such as those in your screenshots, then I am definitely for giving a smaller stripe size a try. Thanks again.

there doesnt happen to be some magical way i can restripe my drive with a new stripe size without having to rebuild the array is there? lol .. i didnt think so 🙁
 
well off i go to re-stripe at 8k and reinstall windows..

when i come back, i will do all the same benchmarks and do a comparison between the two stripe states: 8k and 128k

be back later! expect results tomorrow..
 
I must thank you -DAK- for your clear and thoughtful answer.. it very helpful to me in seeing how the performance should be with these drives. One thing though, you said the screenshot results were from an 8k stripe size.. but it looks like they all say 4k in the pictures? is this correct? I am surprised at such a small stripe size, but if it gets results such as those in your screenshots, then I am definitely for giving a smaller stripe size a try. Thanks again.

You're welcome. 🙂

Yes the flurry of benchmarks run was originally performed with 4KB stripe size. I tried all sizes later and found 8KB to be the best.

Unfortunately, you must destroy the array and (your data!) and start over every time you want to change stripe size. Once you install, use a product such as Symantec Ghost and create an image of your array to another IDE disk. This can be reloaded within minutes and you're up and running again where you left off!

-DAK-
 
Originally posted by: shuttleteam
You're welcome. 🙂
Yes the flurry of benchmarks run was originally performed with 4KB stripe size. I tried all sizes later and found 8KB to be the best.
Unfortunately, you must destroy the array and (your data!) and start over every time you want to change stripe size. Once you install, use a product such as Symantec Ghost and create an image of your array to another IDE disk. This can be reloaded within minutes and you're up and running again where you left off!
-DAK-
and thanks again.. i will try the Symantec Ghost procedure. Only problem I can see myself encountering is how to format the drive once I restripe it. Last time I ran into difficulty with this, and the only way I ended up being able to format it was to go through Windows XP setup, load the intel raid drivers, and then let setup format the disk for me. Not sure why I had a hard time but I did. Normally I just Fdisk the drive after booting with a DOS-boot-floppy, and then format it. But when I ran Fdisk, it gave very odd results.. it showed two drives of differing size (5gb and 12gb) and I just couldnt figure out what was going on... Anywho, if you know what might be going on, or have any advice on how to format a raptor raid-0 drive without using Windows, let me know! otherwise I'll do it the old fashioned way 🙂
 
FDISK will see the first 64GB so it tends to do wierd things...

It's best to create the array and install XP onto the fresh logical volume. Let it (quick) format and you'll be fine.

Once your Ghost image has been created, you can change stripe size as you wish. Ghost will see the new volume and copy there.

-DAK-
 
Alright! I have restriped my Raptor Raid-0 array with a 8k stripe size instead of the 128kb size i used initially. And I have to say it was a good move. I noticed right away the windows installation was zippier.. it felt more like people described when they spoke of "15 minute windows" installations and such (dunno about the exact time, but it was fast)

here's some benchies:
ATTO benchmark (8k stripe)
HD Tach benchmark (8k stripe)

I'm really happy about that HD Tach one. If you compare to my benchmark using 128k stripe, it looks so much better now. The Sisoft Sandra benchy was a bit dissappointing, as it was around 60mb/sec, only about 6mb/sec more than before... I'll put up a screenshot later..

Oddly, when I very first installed Windows XP and did my benchmarks, i was getting 10mb/sec higher with ATTO, and i *think* (if my poor memory isnt lying) i was getting like 90mb/sec on Sisoft Sandra's disk benchmark (now only 60....) something definitely dropped in performance after installing a lot of software on here. Note that I did run a Norton Speed Disk to defragment my drive after installing all my software, and prior to running these benchmarks... so fragmentation *shouldnt* have been a problem...

Anywho, I am happy with the new results and am glad i posted here. I would never have tried as low a stripe size as 8k. Thanks to you all, especially DAK for the assistance in getting this raid to perform better. Lata!
 
OMG!!! I've had enough! Off to order a pair of Raptors we go! Can I expect the same performance out of the onboard SATA raid on my NF7-s? And how long does a cold boot to Desktop take with XP? I've felt that my WD SE 120mb drive was "fast enough", and didn't think I would see much of a diff in apps and games that, but those numbers are 3-4x what my drive runs across the board. There is no doubt that it has to feel nosebleed fast.
 
Expect more like what I posted except my Stripe size is 64K cause I have large and long Music/Vid Files so you'll see a little better increase but I'm using that MoBo Rev 1.2. I believed shuttleteam used an added on a hacked SATA card and pumped the PCI bus running @ 66 with smaller stripped size whereas Flakk has the new Intel mobo which by passes the PCI Bus so that explain their speed. I saw some Maxtors actually beating some Raptors that were faster then mine somewhere....I guess I'm stating don't expect to see God with them, although fast and Bootup is not to shabby once past the RAID0 Read 😉
 
what do you guys think about raid 0 on a p4 board with the ICH5 not the ICH5R. Doesn't the ICH5R just help with making the array bootable, aren't all the calculations still done in software? If so is it possible to make a completely software raid 0 array bootable with xp? If not would it be possible to use them for the os if you put some boot files on a usb memory device or a floppy?
 
Back
Top