Hard drive speed

[ex_reven]

I have a RAID setup and I just ran HDD tach. The burst speed seems decent but the sequential read seems very slow...or am I paranoid?

http://img.techpowerup.org/070705/Untitled-1.jpg

Feel free to post your own benchmarks images/speeds too :)

[Kursah]

Well with my one 500gb SATA2 drive I average 54MB/s...I forget what burst is.

I'd say you're doing pretty damn good dude! :toast:

[Namslas90]

I have a RAID setup and I just ran HDD tach. The burst speed seems decent but the sequential read seems very slow...or am I paranoid?


Thats about right, the speed won't double in RAID. Seq-Read is dependant on DATA Density.

[DOM]

http://i199.photobucket.com/albums/aa306/dom71/zxcv.jpg

[nora.e]

http://img.techpowerup.org/070705/Capture009676.jpg
Here's mine, though I have no idea what it means. Someone care to inturpret?

[ex_reven]

Thats about right, the speed won't double in RAID. Seq-Read is dependant on DATA Density.

I knew it wouldnt double in speed. I just expected around the 120-150mb/sec mark. Any increase is good though :D. Any way to further increase performance? (Apart from defrag and doing that "enhanced performance" thingy in properties).

[Chewy]

It does seem a tad low I get 105mb/s with my 2 drives.. I think the 320's use 2 heads from what I heard, maybe thats why. Still pwning though Raid owns.

[Kursah]

Here's mine, I didn't break 52MB/s, but I usually average around 52-54. So I'm still close! But you can see how much faster your 500GB Raid is compared to my 500GB SATA2 drive. :toast:

[hat]

He's showing off :o

[ex_reven]

He's showing off :o

who?

[hat]

who?
Oh, you know who... :nutkick: :cry:

[ex_reven]

oh.
Im not, it was honestly a serious question. Pinchy said he got 150mb/sec, and i was just wondering why my setup didnt reach that.

[DRDNA]

In raid 0 it does just about double ! At least on all my arrays it does (that is in Raid-0)
And yes thats about right for your array;)I am running the Perps and there sata2's running on a sata 1 mobo my burst suck azz but thats cuz my mobo wasnt for hard drive speed more for overclockin the ram and cpu....on my two 250 gigs in raid 0 =ing 500ish gig array give me this and the avverage is double for my mobo.

http://img.techpowerup.org/070705/HD TACH.jpg

[ex_reven]

Looks pretty good. Do you recon the stripe size in the raid affects performance a heap?
My stripes are 120kb Im pretty sure (or maybe 64, meh) it was the default size.

[DRDNA]

mine is the default as well:toast:

[ex_reven]

Is it possible to change the stripe size? Or would that require a format...
Id like to experiment with different sizes

[DRDNA]

it would require a full reformat and not the quick option either:eek:

[Chewy]

I got 32bkb stripe size, it depends on the avg file in your system like if you want a little boost in gaming performance use smaller if the files you use alot are bigger (movies?) use a larger stripe size... I dont see it making a huge differnce though.

If you ahve the option you can disable some cache and stuff listed in the vista tweaks thread.. my raid controller does not let me do that though.. but the one prob with doing that is if you dont have a back-up power supply and your houses power fails you lose all your data and have to reformat.. so to do it you should have a battery back-up power supply.

I guess disabling that thing gave some people a nice performance gain, but its not guaranteed to,

[DRDNA]

Is it possible to change the stripe size? Or would that require a format...
Id like to experiment with different sizes

Or you may be able to ghost over the array to a big enough single then reformatt using desired stripe size then ghost the drive back from the single on to the newly striped array... I am certain i have done this in the past.

[mas0n]

ex_reven

You have a bottleneck somewhere, there is no way you have a consistent read speed across the first 450GB of that array.

[Kursah]

Ex, did you pull the jumpers for SATA2 mode? Judging from burst speed you did, but I know on my 'Cuda, I had to remove a jumper that sets the HDD to SATA 1 as default. Just a thought, but I think it's a wasted thought considering your speed. It's still damn fast.

[ex_reven]

ex_reven

You have a bottleneck somewhere, there is no way you have a consistent read speed across the first 450GB of that array.

Do you mean how it doesnt vary alot? Why would that be impossible? (Theyr fairly new drives and they only ever have about 50-60gb of space used at any one time if that makes a diff)

Ex, did you pull the jumpers for SATA2 mode? Judging from burst speed you did, but I know on my 'Cuda, I had to remove a jumper that sets the HDD to SATA 1 as default. Just a thought, but I think it's a wasted thought considering your speed. It's still damn fast.

Wasnt aware they had jumpers. Well I did think they had jumpers, but when i looked into my case the other day I saw what looked like a place for jumpers, but when I used my tweezers I couldnt seem to find any bridging piece that goes on the jumpers. Maybe if i took them out of my case I could have a better look, not atm though.

[Kursah]

The jumpers are very, very small...very small..did I mention they are small? I had to use a small pick to get it out! I lost it immediately! That may help you. My performance increased a little, mostly my burst which went from about 120MB/s to 255MB/s. My overall increased a few MB/s. I wish I could get more out of it, but I don't think there's really anything I can do, I have a partition for Windows Vista that's approx 20GB, and then the second partition for programs/games/downloads, etc. I may get another and go raid someday, but 500GB is sufficient for me ATM. :toast:

[ex_reven]

http://img.techpowerup.org/070705/Untitled-1147.jpg

I feel so clean lol...Im gonna defrag and rerun the test :roll:

[mas0n]

Do you mean how it doesnt vary alot? Why would that be impossible? (Theyr fairly new drives and they only ever have about 50-60gb of space used at any one time if that makes a diff)

If you are not sufferring from a bottleneck the results in HD Tach should always follow an arc. This is because the entire disk is read from start (outside) to finish (inside) and the outside tracks will always read faster because they are spinning at a higher rate of speed than the inner tracks.

[ex_reven]

Redid it...pretty much the same result

http://img.techpowerup.org/070705/Untitled-1960.jpg

[nora.e]

This is because the entire disk is read from start (outside) to finish (inside) and the outside tracks will always read faster because they are spinning at a higher rate of speed than the inner tracks.

how is this posible? if a wheel is turning at 7,200rpm at the outer edge, and solid and has no flex, it will also still be turning at 7,200rpm at the center hob area.

[ex_reven]

how is this posible? if a wheel is turning at 7,200rpm at the outer edge, and solid and has no flex, it will also still be turning at 7,200rpm at the center hob area.

No, use your head. Think about it.
The outer part is moving faster. If I swing my arm to throw a punch the bit of my arm a few cm from my shoulder only moves a few cm while my hand has moved about 4 feet in the same period of time.

Say it takes me 1 second to throw the punch. My fist is travelling at 4 feet per second, and that topmost part of my arm near the shoulder is travelling at say 5cm/sec.

Roughly:
Hand = 120.4cm/sec
Top of arm = 5cm/sec

Hand = 433440cm/hour
Arm = 18000cm/hour

Hand = 4.3km/hour
Arm = 0.18km/hour

My hand is to the arm what the outer edge of the disk is to the inner edge of the disk.
While my hand rotates around my shoulder, the outer edge of the disk rotates around the centre of the disk. They both operate around a central point.

[driver66]

LOL roll a paper cup on the floor and see which end is traveling faster:toast:
That's a simpler way to explain it:roll:

[ex_reven]

Woah...sorry about this double post but I just hit a extremely anomalous result.
I ran HDtach again after finding and removing the jumper on both hard drives (which theoretically changes it from Sata 1.5gb/s to Sata 3gb/sec) and look at my new result :eek:

http://forums.techpowerup.com/attachment.php?attachmentid=8502&stc=1&d=1183698536

LOL driver! Your analogy works better :p

[nora.e]

LOL roll a paper cup on the floor and see which end is traveling faster:toast:
That's a simpler way to explain it:roll:

Ya but if you stick a pencil threw it lenth wise an spin the pencil both ends are traveling at the same rate of speed again.

[ex_reven]

Ya but if you stick a pencil threw it lenth wise an spin the pencil both ends are traveling at the same rate of speed again.

I dont get it? Think of it this way, anything that travels a different distance in the same amount of time is travelling at a higher speed. The outer edge of the disk travels further because its circumference is greater than the inner edge.

[nora.e]

No, use your head. Think about it.
The outer part is moving faster. If I swing my arm to throw a punch the bit of my arm a few cm from my shoulder only moves a few cm while my hand has moved about 4 feet in the same period of time.

Say it takes me 1 second to throw the punch. My fist is travelling at 4 feet per second, and that topmost part of my arm near the shoulder is travelling at say 5cm/sec.

Roughly:
Hand = 120.4cm/sec
Top of arm = 5cm/sec

Hand = 433440cm/hour
Arm = 18000cm/hour

Hand = 4.3km/hour
Arm = 0.18km/hour

My hand is to the arm what the outer edge of the disk is to the inner edge of the disk.
While my hand rotates around my shoulder, the outer edge of the disk rotates around the centre of the disk. They both operate around a central point.

what you're desribing here is more an isue of leverage, not speed. If you extend your arm out to the side and turn at the waist half way around both your hand and your sholder have done the same thing, make a half of a revolution.

[ex_reven]

what you're desribing here is more an isue of leverage, not speed. If you extend your arm out to the side and turn at the waist half way around both your hand and your sholder have done the same thing, make a half of a revolution.

Leverage is much more focused on energy efficiency through changing fulcrums. The fulcrum in this case is the centre of the disk. It is the leverage that achieves the different speeds, which brings us back to the point that the outside of the disk is indeed faster.

Another (more stable looking) result:

http://forums.techpowerup.com/attachment.php?attachmentid=8503&stc=1&d=1183699610

[nora.e]

I don't see us coming to an agreement on this one:rolleyes::toast::rolleyes:

[ex_reven]

I don't see us coming to an agreement on this one:rolleyes::toast::rolleyes:

thats ok, you can have your opinion
Back to the point it sprung from, the whole "my graph should be a curve rather than be going upwards after its starting position because the disk spins slower as it moves away from the edge" comment made by mas0n.

Why isnt mine doing it? I can see in everyone elses results that were posted that they had the symptoms he described.

[mas0n]

It's simple if you break it down.

The entire disk must spin at 7200RPM.

The outer tracks are longer than the inner tracks.

The outer tracks therefore would allow for more data to be read in a single rotation.

Also, If you have an 12 speed bicycle and peddle at the same rate through all 12 speeds, the 12th speed (the one with the largest gear) allows for the highest speed.

[driver66]

Ya but if you stick a pencil threw it lenth wise an spin the pencil both ends are traveling at the same rate of speed again.
:banghead: :banghead: :banghead:

8504

This is all my old beast does :toast: It seems ok.:D

[GLD]

@ ex reven

I see you are running 250g and 320g hdd's in your raid 0. Isn't it considered best to run matching capacity hdd's? Could this be the cause of the anomaly you are seeing?

[ex_reven]

@ ex reven

I see you are running 250g and 320g hdd's in your raid 0. Isn't it considered best to run matching capacity hdd's? Could this be the cause of the anomaly you are seeing?

well, looking at my last benchmark my average sequential read speed seems alot more steady.
As for matching hard drives, im pretty sure the size isnt a problem, since they are partitioned evenly anyway. Id say that them being the same model drive (7200.10 in this case) in comparison to two different models has a bigger impact on performance.

[mas0n]

do both use perpendicular recording?

[ex_reven]

This is all my old beast does :toast: It seems ok.:D

Lookin good

do both use perpendicular recording?

My drives? Both of them are 7200.10, so yes. I should buy a third. These things are pretty damn cheap too.

[mas0n]

What confuses me is that it looks like it is a bottleneck at the controller but your burst speeds are right where they should be.:confused:

[ex_reven]

What confuses me is that it looks like it is a bottleneck at the controller but your burst speeds are right where they should be.:confused:

I thought the same.
I wonder how much benefit a hardware based controller would provide then on my particular setup.

[hat]

Ok, NOW you're showing off :laugh:

[ex_reven]

nope...pinchys score still pwns mine :p

[Wile E]

how is this posible? if a wheel is turning at 7,200rpm at the outer edge, and solid and has no flex, it will also still be turning at 7,200rpm at the center hob area.Yes, they will be rotating at the same speed, but the outside edge must travel a greater distance than the inside hub, to make one complete rotation. Therefore, the surface speed of the outer edge is greater. IE: The drive head passes across more data per rotation on the outer edge, compared to the inner edge.

@Reven - Those new benches are looking better. The different sized drives could cause the discrepancies, when compared to others.

[regan1985]

speeds are also different due to what your running like my onboard runs at 94mb but on my pci ex runs at 98mb and if i had a good controler i could probibly get 108mb, but burst speeds dont really matter unless you are moving loads of small 1k files, its better to have a line that curve down on a gentle slope and have a read time of under 13,

[Verikon]

http://img174.imageshack.us/img174/9923/hdtachkb6.th.jpg (http://img174.imageshack.us/my.php?image=hdtachkb6.jpg)

Errr... I don't like that downward trend near the end. I want magic SATAII jumpers on my HDDs!


No fair...

[TXcharger]

wow pitiful mine was under 60mb/s at all times and had random drops with a huge curve

[Wile E]

http://img174.imageshack.us/img174/9923/hdtachkb6.th.jpg (http://img174.imageshack.us/my.php?image=hdtachkb6.jpg)

Errr... I don't like that downward trend near the end. I want magic SATAII jumpers on my HDDs!


No fair...
That actually looks about right, Verikon.

[Verikon]

That actually looks about right, Verikon.

It's just sad that my results have to curve down like a frown while everyone else's are nice and straight.

But it is good to know my results don't spell doom! I'm always paranoid my RAID array will collapse, but since my WD drives are so new (under a year) I think they are single platter drives which, from what I've read, means they are more reliable. Fewer parts, fewer places for failure. Keep it simple, stupid.

Hopefully they will last me a few more years.