ICH10R advanced format drives (4k) in RAID1 (mirroring) slow data transfer rates locally but not when accessed through Gbit LAN?

[idata]

Hi!

First of all my system specs:

I7-975EE

6GB OCZ RAM

Gigabyte EX58-UD5 motherboard (X58 chipset)

MSI 5870 Lightning

X-fi Titanium

Intel 80GB G2 SSD

Samsung F1 1 TB hdd

Pioneer BDR205

2x WD15EADS 1,5TB in RAID1 (non advanced format drives, for simplicity "array A")

2x WD20EARS 2,0TB in RAID1 (advanced format drives, for simplicity "array B")

Corsair AX850 power supply

O/S: Win7 x64 Professional

Intel RST 10.1.0.1008

Nothing is overclocked!

Here's my problem:

I have 2 RAID1 arrays as I described above. Each array has one partition. Everything is fine for array A. It's data transfer rates are as espected. (However they're "green" drives so their speed is behind 7200rpm drives.) I've measured about 60-80MB/sec in read/writing and that's acceptable for me because I've purchased these drives for high amount of data storage.

But I have data transfer rate issues for array B. It's speed is half like array A. The transfer rates are never exceed 20-40MB/sec locally which is I think ridiculous nowadays.

However I recently discovered that when array B is accessed through Gbit LAN the transfer rates are right. (60-80MB/sec instead of local 20-40MB/sec)

So I assume that it has to be local software related otherwise accessing through LAN wouldn't be so fast...

The differences between the two kind of drives:

-WD15EADSWD20EARSsize:1,5TB2,0TBcache:32MB64MBuses advanced format? (4K)noyesmeasured transfer speeds in RAID1 array when accessed locally60-80MB/sec20-40MB/secmeasured transfer speeds in RAID1 array when accessed through Gbit LAN60-80MB/sec60-80MB/sec

They belong to the same family/speed ("Green drives").

There's no sense that array B should be that slow locally...

Here's what I've tried so far increasing the transfer rates for array B:

1:changing default cluster size for ntfs volume --> small improvement

2:using jumpers for the advanced format drives (WD says jumpers are should used for hardware aligning the first partition to 64th sector, not 63rd as default for non advanced format drives) --> small improvement ( however I use Ubuntu rarely and because of the jumpers at the booting process it gave me "DMA ERROR : DR NOT RDY" or something similar errors, removing the jumpers the boot went fine again)

3:trying to align partition in different ways (to match 4k sector boundaries) --> no improvement

4:enabling-disabling "write back cache" in Intel RST multiple times --> no improvement

5:updating Intel RST from 9.6.0.1014 to 10.1.0.1008 --> no improvement

6:defragmenting array B --> small improvement

7:disabling anti-virus software --> no improvement

8:re-creating array in Intel RAID BIOS (CTRL+I during boot) --> no improvement

(But none of them gives the same transfer rates like array A even if I combine them.)

1:Does anyone have any idea of what's going on? What should I try?

2:Nobody encountered any issues when built an array of advanced format drives?

The Intel RST FAQ mentions the 4k (same as advanced format) drives that they're supported from RST 9.6 however there's no mention regarding 4k drives used in RAID arrays...

However it's weird that in Intel RST when I check the physical properties for the advanced format drives that part of the array (or any 4k drive that is connected to the motherboard sata ports) under "advanced" the "physical sector size" is 512 bytes not 4096 bytes...

I just simply want the same data transfer rates for my array B because it's really annoying being so slow...

(The migration for advanced format drives already started so sooner or later advanced format drives will be the only available drives on the market.)

[idata]

This has nothing to do with advanced format and 4K sector drives I did some testing with some 320GB drives that are 512 bytes sector drives and RAID 1 with Intel had about half the Sequential speed lost of a single drive!

Test of a single drive ST3320620AS

RAID 1 using 4K cluster

RAID 1 using 8K cluster

The fact that you can pull data off it faster over LAN then locally is interesting...could you try swapping the ports over with your other array and another test you can do is disable the LAN in the BIOS and see if your speeds are better locally after that.

Note: the more stuff you have on the array any test files you write will be at the slowest part of the array.