hard drives & raid - benchmark and compare!

[Guest Kobalt]

Kobalt

2hrs to copy 160 gigs..

Wow man, that's along time...

I'm not bragging or anything but it takes me at tops 40mins to copy around 120gigs, normally around 20-40mins, depends on how much I copy.

I dn the exacts, but 2hrs is very very long, it's never taken me that long to do a straight copy.

 

I mean it may take me all night to copy stuff to here and there, remake partitions, etc etc, to get it all setup.

Then once it's all setup, copying the files to there correct location does'nt take to long.

 

For example a 80gig partiton to copy to a 100gig partitoin only takes me around 20mins probably.

Another 20mins to do my other drive's big partition.

Ruffly anyways, maybe add another 10-20mins, I forgot.

Not an hour though.

 

I just did this about a month ago...

 

It takes much longer on this seagate, it's a really old drive though.

So you were getting around 68 MB/s ?

These drives were not RAID or anything were they?

FYI, the 300GB was a Seagate Sata II 16MB cache, and the other was a Seagate Sata I 8MB cache.

CPU times was nill. (Well, you would expect that on a DC CPU. )

[red930]

naw ive got my 3 samsung 250gb and ive seen the sil3114 on my 3200dr is slow with raid5 n e way wether im right or not about how it works.

RAID-5 on the SIL controller is slow. There's not anything that can be done about it!

 

It's just slow!

[NEOAethyr]

So you were getting around 68 MB/s ?

These drives were not RAID or anything were they?

FYI, the 300GB was a Seagate Sata II 16MB cache, and the other was a Seagate Sata I 8MB cache.

CPU times was nill. (Well, you would expect that on a DC CPU. )

 

 

Nah I could'nt of been getting that high of performance.

 

I just benched with atto on one of my drives, the messed up one.

And I'm getting 58megs a sec reads and 55megs a sec writes with 256m sizes "files".

That's on one of my 2 temp file partitions though, very small partition(4 gigs).

 

I probably get around 40m a sec or maybe slightly higher in real world on my bigger 100gig partitions.

 

Hmm maybe I had that backwards...

 

Anyways as an edit...

 

Just tested very half @ssed using atto but wihtout direct I/O.

Holy freaking crap...

 

Reached over 500megs a sec and 700megs a sec.

 

I will post a screeny.

 

 

Editing to add the pic.

Not best of course, on the net, all that etc etc, unclean partition, blah blah.

 

More wierd results.

Ok, wierd results with atto and hd tach, any other programs?

[Guest Kobalt]

Lol, give me that HD

 

700MB/sec reads... Almost like a RAM drive.

 

I did atto with the same settings as your pic, and I get around 34MB/sec on my new Sata 2 300GB HD.

I also didn't know that there was a speed test built into the Nvidia drivers, according to that, I have 214MB burst, and 67MB sustained.

It does say it is 2nd generation SATA drive.

 

*edit:

Whoops.. I UNchecked direct I/O, and got some wacky readings:

 

 

Guess it was reading from the cache or something along those lines.

[NEOAethyr]

Yeah it must be the windows subsystem performance.

 

Strange that we can sustain so much though.

Then again those are in 64k chunks, but still, it's reading/writing out a 256m sized like file so....

 

Using the nv drivers here.

Onboard.

 

That was my 1200jb, I forgot what partitions were what at that time.

 

I'm on my 1600jb right now.

Using it as my only drive, I'm moving to a new case...

No hd bays or anything .

Net yet anyways.

 

Here's what it looks like, I'll attach the image.

I don't have the time to crop it correctly, just enough to get it to fit as an attahcment.

I don't have any fans on my drive ...

 

I bet it'll do better with more i/o depth.

 

Anyways I gotta get and work on my rig more.

I did some benches comparing the diffrences between how you have a hd set.

Normal - lable up.

Upside down.

And both sides.

Only for one hd, only 1 pass each.

 

Seems as if arm pointing down towards the center or upside down is best on this hd.

I will test my other 2 drives though to try getting a better idea, the diff seems very nil.

I'm just curouis since I'm setting upa new case and all, as to what way my hd's should be set in my case.

 

 

 

Oh yeah, I figuered out why hd tach said I had 0% cpu usage.

It was because I was runnig 2 instances of prime and etc at the time.

[NEOAethyr]

Ok...

 

Got both my western digital drives hookd back up.

 

Figuered out that cooling the top of the drives does'nt cool as well as cooling the bottem.

 

My 1600jb, benching before messing around with it, then after and then again and again...

Well, upside down yields less then 0.0ms access time diff, somehting like 0.0xms.

Somehting I can't even see the majority of the time with the benches, but it's notcieable during windows startup.

The fake xp style progress bar when loading windows.

 

And in gaming it's slightly noticeable.

 

During benching, the 1st time I put it upside down, I got 0.1ms less.

 

To give you an idea:

The arm would swing down to the center of the disk: 19.8ms, 10% usage

The arm would swing up to the center: 19.8ms, 12% usage

Upside down(lable down): 19.7ms, 8% usage

Upside right, lable up: 19.9ms, 10% usage

 

Nothing to go by right?

After hooking it up by it's self, it's was'nt that great either, not the best of te above, while upside down.

After becnhing with both disks in, I got 19.8ms and 12% usage again.

 

Windows startup when alone and now together again is lower though.

Instead of the middle of the bar, it's probably 3 spaces, I did'nt count.

Same was with it alone, faster, but faster on avg.

 

Gaming alone was better, but there was high latency, but it was all good.

 

Gaming together, was'nt much diffrent then it was before I moved to my new case.

Not done doing that one at all...

Anyways, seemed slightly better.

 

Both hd's are upside down.

 

 

I notcied this one before, but shook it off and did'nt care, now I'm thinking about redoing it...

The swap.

I have dual 1 gig swaps, 1 each per drive on there primary partition.

 

Latency seems halfway between the 2 disks.

Slow gaming loading on avg, but ok gaming.

 

When on the 1600jb, quick loading, but some added latency while gaming.

 

I belive if I go back to using only one swap on the 1200jb, my latency will go way down, and i'll have the quick loading.

 

Last time I benched the 1200jb, a bit ago, it was 13.2ms of latency, with 6% cpu usage.

 

 

I had setup 2 swaps incase one drive failed temperarly.

However, for the time being, I think I will setup windows to only use one of those until I need the other one if such an occurance happens again.

 

 

Power on time count seems lower on the 1200jb now, instead of over 10000 or 11000, whatever it was, it's 9889(not great).

I think the power on cycle count is lower too, but I'm not sure.

I should of checked before hand...

 

I can allways flip them around later and do a through check.

 

 

The temp thing, one other thing I notcied is that the motor it's self on my 1600jb is 1c hotter then reported by smart.

At the time it was 5c above ambient, now smart for that drive is 1c above ambient...

Anyways, cooling the botten half of the drives seems most important...

 

 

Sorry, I hope you guys don't mind the non raid chatting for a moment.

It's all hardrive related...

Been years since I did any serious hd bences, since back when I had 1gig drives...

 

 

Edit:

After trying both hd's swap file, alone with both hd's installed.

 

Maybe 5 bars to load windows on the xp stle loading screen with the 1600jb.

3 bars when using the 1200jb.

 

2 bars when using both.

 

Gaming loading was no better either way, tiny tiny bit better on both drives.

 

The 1200jb is the lower latency, but slower drive, the 1600jb is the higher latency but faster drive.

 

It seems I need to put them back on there own pata controller channel, instead of sharing one like they are now(I've had it like this for some time now, about when my load times when up for gaming).

 

It does seem that using a dual swap, on 2 good drives per say, actually does better then only having one swap.

Kinda surprising...

 

On raid however one swap would defently be best.

[Guest Kobalt]

I was doing some more tests, and just noticed this when I booted the machine:

 

Now, it seems the BIOS allows you to pick the mode (which always should be LBA AFAIK), but *only* for IDE devices.

 

For SATA devices, all it has is AUTO & LARGE. No CHS or LBA. But then it goes on to show LBA & CHS.

 

I assume this is a BIOS bug of some type? I did search for a bit, and it seems that there is no official word on this?

 

Never mind the fact that a drive that big can't be CHS, so again, this makes me think it is a BIOS bug.

 

I also think that 2k/XP/linux ignores what the BIOS says anyway correct?

 

Speed wise it all checks out OK from the Atto & HD tach scores.

[newguy]

Any suggestions? Not very sure I am doing it right? Just setup the RAID on Tuesday... Thx!!

 

 

[soundx98]

Newguy - set "total length" to 32MB

and check "neither" under I/0

 

They look like Raptors with that access time

Try "unchecking" - ReadCachingEnabled under NForce Serial ATA Controller in Device Manager.

[newguy]

sound,

 

Will run that again after back home. And yes, they are Raptors in 80GB, which WD made for DELL only.

 

Thx!! Will post again.

[soundx98]

re above

http://www.dfi-street.com/forum/showpost.p...&postcount=1259

 

NewGuy - also disable "command queing" on your raptors

[green-man]

hi

ive just ordered xp pro 64bit and as i'll be reinstaling wanted to know how u lot managed to get 3 drives in raid 0 on the uli controller?:confused:are u setting up ur arrays in windows or sumthin? . ive got 2 spinpoints in raid 0 and 1 disc running seprate but on same controller. still the speeds are still sweet on the raid 0 setup, hd-tech 8m test 145mb/s-80mb/s, burst around 300mb/s, 12gb partition 32kb stripe..