18000 in SiSoft Sandra

[VotTak]

My old KT7RAID's BIOS was flushed to new 7n which includes 2.32 BIOS for HTP.
Freshly installed WinXP( with 2.32 drivers ).
Your suggestions and thoughts are greatly appreciated( as always ).
Thnx

[speculative]

Hmm... is your setup as listed in your sig? With a 45GB IBM of the same make I get around 21000 in Sandra, on a KT7 with an older BIOS (think it's the 3a or something or other, I'm really out of date when it comes to BIOS's).

-spec

[VotTak]

Yeah... setup is "as listed".

The thing is that before the flush it was around 40000-45000 but I can't remember the conbination of BIOS version and HPT BIOS version.
:mad:

[mdzcpa]

Try a defrag than run Sandra again. Sandra is a poor indicator of RAID performance and is very sensitive to fragmentation. Give it a shot and report back:)

[Player0]

Agree with mike...Sandra is a poor drive benchmark.

[VotTak]

Well... thanx for your input guys... although... I'm going to surprise you.
As soon as I convert from NTFS to FAT my speed went back to normal( 42500 ).
That is very unexpected for me... I don't know about you but it was very surprising for me.
Read and write perform better, although access time become 8ms( it used to be 7ms with NTFS ).

[mdzcpa]

Although there is a known small hit to performance under NTFS, I still think your results are based more on how the Sandra bench works than the actual performance of your RAID set up.

To really bench a RAID set up with consistency you need WinBench or another similar program. ATTO is not as good, but better than Sandra HD bench too.

just food for thought:)

[VotTak]

Just to stress out those results ... tests were done on partition with no files... nothing on it( so there is nothing to defragment ).

And, as it was pointed out by mdzcpa SiSoft might misinterpret that data.
One more thing about Sandra... On my O/Ced Northwood ( after some changes ) Sandra tests shows performance of around 112000-113000 for RAID of 2 60G IBMs 60GXP. Before changes it showed regular 52000 or so.

So... anyway I thought that 2.32 HPT is not working... aparently it is.