Generally speaking, RAID 5 uses block-level striping with parity data distributed across all member disks. RAID 5 can suffer from poor write performance itself because parity must be updated on each write. This is because parity must be calculated on each write, requiring read-modify-write sequences for both the data block and the parity block. A discrete RAID controller will help with this because the controller processes the parity calculations. With embedded RAID, the parity calculations must be calculated using the system board processor and memory.
Embedded RAID also writes using a write-through cache setting. Write-through mode means that any cache is only used as a buffer and the drive will immediately write data the operating system instructs it to. The operating system sends a command to the drive and the operating system is forced to wait while the drive completes the write.
I know that a embeded RAID Controller is not that fast as a external Controller.
I don't know if the board uses the CPU or an other onboard Chip for calculating the parity...
The CPU Cores are always between 0-2%.
I also can change the Cache Settings to Write-Through, Off or Read Only...
It makes not a big difference to change they...
Be sure that Disk write cache is enabled in Windows* and also try enabling write-back cache in Intel(R) Rapid Storage Technology and in device manager. You should also check in the Device Manager in the Policy tab to select “Turn off windows write-cache buffer flushing on the device”.
In addition, keep in mind that parity calculations in Intel(R) Rapid Storage Technology are done using system memory and processor as this RAID solution does not have dedicated memory and a processing unit, therefore, the software running in the background (including the benchmark tool being used) may affect the performance.
The Disk write cache is enabled... I ckecked that..
I cant enable the Write-Back Cache Modus... But I can use Write-Through.
The Memory or the Processor can't be the Problem... Its a new i7 4770-k and new Vengance RAM's
I found old Threads in the Forum who peoples tried several Strip Sizes and Allocation Unit.
Maybe i should try others than this one...
I found something out.
At first, i tryed out some other strip sizes and Allocation Unit.
The Results with 5 HDDs are ~550 MByte/s Read and ~25 MByte/s Write Speed.
Tested with the biggest and the lowest stripsize (128k and i belive it was 16k) and several Allocation Units.
It didn't make significant Speed differences if i activate one of the Cache Mode.
The interesting thing is, i tryed also to make a RAID 5 with 4 HDD's.
The Results are ~440 MByte/s Read and ~25 MByte/s Write Speed.
Then i activated the Write-Back Cache Mode and i got ~350 MByte/s Read and ~350 MByte/s Write Speed.
The Results with 5 HDDs i got with a mini RAID 5 about 100 GB over all HDDs, the Results with 4 HDDs are from a RAID about 75 GB over all HDDs.
All this Results let me belive that the RST driver (or what ever) ist not correctly working on my Board...
I have the Version 18.104.22.1686.
Currently i initialize the full Storage from 4 HDD's.
Now is the Question which mistake is it...
Is it Asus, because they maked something wrong on the Z87-Pro or is it Intel, because your driver isnt actual (the Version 22.214.171.1246 is from 3.22.2013) or is it my mistake...
I belive its Intel, because i don't belive that on the date 3.22.2013 it gave Boards with 6 SATA Ports (Maybe i am false).
Intel(R) Rapid Storage Technology 126.96.36.1996 is also functional on new motherboards based on Intel(R) 8 series chipsets. If you may like you can test using the driver provided directly by Asus* and check directly with them to see if these values match what they have tested on their systems or see if they can test your motherboard for you