Single large dip during benchmark runs

A

AMD_Gamer

Fully [H]
Joined
Jan 20, 2002
Messages
18,287
I am benchmarking RAID 0 and RAID 5 on my Perc 6 controller with 4 500GB TLER drives and for some reason I get this single large dip during the test, is this normal? could it be a bad drive or what could cause this?

jkimr.jpg


GB2BP.jpg
 
Yeah, I had something similar, I took apart the RAID and retested the drives individually and found one drive causing the dips. I replaced the drive and the performance dips were gone from the benchmark.
 
Your first benchmarks looks invalid; try CrystalDiskMark instead or the "File benchmark" in HDTune Pro. Depending on controller, HDTune would give you incorrect output with default settings on many RAID arrays.
 
sub.mesa said:
Your first benchmarks looks invalid; try CrystalDiskMark instead or the "File benchmark" in HDTune Pro. Depending on controller, HDTune would give you incorrect output with default settings on many RAID arrays.
Click to expand...

Yes I know that, it was the block size in HD tune being set to 64kb, so it was writing to one drive only, increasing the HDtune block size to 8MB solved the problem.
 
Still it is single queue depth I/O; not how any actual program would access your disk(s); it would use the filesystem instead. CrystalDiskMark, AS SSD and the File Benchmark in HDTune Pro use the filesystem instead to do I/O. Even with 128KiB blocks, that would mean it uses multiple queue depth and send several requests at once for sequential I/O; read-ahead and write buffering.

The small dip may just be some reallocated sectors in that spot; shouldn't be anything to worry about.
 
sub.mesa said:
Still it is single queue depth I/O; not how any actual program would access your disk(s); it would use the filesystem instead. CrystalDiskMark, AS SSD and the File Benchmark in HDTune Pro use the filesystem instead to do I/O. Even with 128KiB blocks, that would mean it uses multiple queue depth and send several requests at once for sequential I/O; read-ahead and write buffering.

The small dip may just be some reallocated sectors in that spot; shouldn't be anything to worry about.
Click to expand...

Would there be any way to fix it?
 
You must log in or register to reply here.
Back
Top