Spinpoint F3 500gb RAID0

[DUBSTEP]

Got these in raid 0 going through a silicon image 3132 sata raid card in a pci-e x1 slot on an Asus P5QL PRO.
Why is the read speed on HD Tune and ATTO so slow?

 

[kithylin]

That's about on par for a standard 7200 rpm sata-2 drive in raid-0, what were you expecting, 300+ from two drives? Each one alone would only score 50 MB/sec - 60 MB/sec.

Edit: Asus website says your motherboard has ICH10 on it, although it might not be raid.. i'm not seeing anything about configuring raid in your motherboard's manual, so maybe not... sorry i posted a reply eariler thinking you had the onboard-raid chip.

just my 2 cents anyway.

 

[Cattykit]

That's about on par for a standard 7200 rpm sata-2 drive in raid-0, what were you expecting, 300+ from two drives? Each one alone would only score 50 MB/sec - 60 MB/sec.

Edit: Asus website says your motherboard has ICH10 on it, although it might not be raid.. i'm not seeing anything about configuring raid in your motherboard's manual, so maybe not... sorry i posted a reply eariler thinking you had the onboard-raid chip.

just my 2 cents anyway.

50-60MB??? Check your calendar, it's 2010, not 2000.
Based on numerous reviews I came across, single F3 500GB should do 110mb.

 

[AdamK47]

Is that in Vista or Windows 7? I have two 1TB SpinPoint F3s in RAID-0. Here's my ATTO results from when I ran it in Vista x64.

That was with the 2TB array roughly 70% full. I tried running ATTO again in Windows 7, but the disk write results were very low. I don't think the version of ATTO I have likes Windows 7.

Here's HD Tune:

HD Tach:

The HD Tune and HD Tach results here are the same as what I get now in Windows 7.

 

[DUBSTEP]

OK thanks so does anyone know why im getting such a slow read rate? I thought it would be about the same as the one you're getting with the 2tb array.

 

[Qbah]

I have a RAID0 of my HD501LJs still operational, Win7. Gonna break it apart later today, but I'll run the test on my config and will give you my results - from what I remember, I was getting similar numbers...

 

[DUBSTEP]

OK thanks, its on win 7 btw

 

[Qbah]

OK thanks, its on win 7 btw

It's on Win7 here too, this is my result. The RAID0 is divided into two partitions, a smaller 200GB and a larger 800GB, both almost full - no idea if that has any onfluence on the results. Anyway, there you go:

Some music was running in the background and two more drives attached to the system - a slow WD 160GB and my new 1TB Samsung HD103UJ. Here are the results for the new Samsung:

As you can see the 500GB drives from Samsung, if exceptionally quiet, are no speed demons Two of them in RAID0 are around the speed of a new gen 1TB drive. So your results are pretty good, I think.

 

[DUBSTEP]

Thanks for the reply, appreciate it.

 

[Dorkenstein]

Why would other reviews and such say the F3 is really fast? Unless I am mis-understanding something here.

 

[Longspeak]

"As you can see the 500GB drives from Samsung, if exceptionally quiet, are no speed demons Two of them in RAID0 are around the speed of a new gen 1TB drive. So your results are pretty good, I think."

I'm confused as well. If I understand the above statement correctly, it suggests I get the same performance with 1 new gen 1TB drive as with a pair of 500GB drives in raid-0???

If my above statement is correct, would I be better off with a single new gen 1TB drive vs a pair of 500GB in raid-0??

Can someone help with this??

Thanks

 

[Zap]

I'm confused as well. If I understand the above statement correctly, it suggests I get the same performance with 1 new gen 1TB drive as with a pair of 500GB drives in raid-0???

If my above statement is correct, would I be better off with a single new gen 1TB drive vs a pair of 500GB in raid-0??

No. You are using Samsung F3 drives, right? Those are latest generation with a single platter, and should not be confused with older 500GB Samsungs (or any older gen drive).

Maybe it is your controller card? The faster results may be from using Intel ICH10R. I recall reading some reviews that shows Intel controllers to be faster (except for high end hardware RAID controllers).

 

[Qbah]

Hmm... looks like all my drives are the "F1" models. So you may want to disregard my results if the F3 is supposedly a lot faster. Sorry for the confusion

 

[WildW]

I may be talking from my rear here, but I notice that the OP's graph is very flat, where generally a HDTune graph will curve downwards as you go across the surface of the disk - as is seen on some of the other graphs posted here.

I believe this downward curve is related to the time taken for the heads to move out across the surface, hence lower average throughput the further the heads have to move. But the OP's curve is flat. . . so this mechanical issue isn't what's limiting the throughput.

My guess would then be that the controller is saturated - the disks could go faster, but the disk controller can't utilize it. Why that would be. . . <shrug>

 

[DUBSTEP]

Yeah that does make a lot of sense actually. The sata controller was only like £15 so I bet thats whats limiting it. Though what would explain the write rate that im getting shown in the atto benchmark, thats higher than the read im getting in both benchmarks. Could anyone recommend a sata controller that would be able to handle the load and that is not too expensive? Thanks

 

[Longspeak]

Hmm... looks like all my drives are the "F1" models. So you may want to disregard my results if the F3 is supposedly a lot faster. Sorry for the confusion

Qbah, thanks for the clarification-I thought I was losing my mind trying to follow this thread.

Soo, since I'm about to buy a couple of 500GB drives for Raid-0, is there a definitive choice between WD Black and Samsung F3?

I have read a couple of posts that WD Blacks don't play nicely in Raid, although not sure it was referring to Raid-0.

Thanks for the help!

 

[Sahakiel]

I may be talking from my rear here, but I notice that the OP's graph is very flat, where generally a HDTune graph will curve downwards as you go across the surface of the disk - as is seen on some of the other graphs posted here.

I believe this downward curve is related to the time taken for the heads to move out across the surface, hence lower average throughput the further the heads have to move. But the OP's curve is flat. . . so this mechanical issue isn't what's limiting the throughput.

Curve isn't from head movement. It's because disks spin in circles so the outer edge moves faster than the inner edge.

If the graph is flat, something else in the chain is limiting throughput. PCI-E x1 is 250MB/s, so maybe Raid controller, the PCI-E switch, maybe even CPU.

 

[dook43]

Here are my results on ICH10R/Win7:

 

[DUBSTEP]

Thanks dook for posting your benchmark.
Also i doubt it the cpu limiting anything because its an E5200 overclocked to 3.2 so i think its the controller?