RAID0 degradation?

[EnderW]

I've noticed my RAID0 array is much slower than it used to be. Is this normal? I have a lot more stuff saved on it now.


11-23-2009
4 x Samsung Spinpoint F3 HD502HJ RAID0 on ICH10R

6-6-2011
4 x Samsung Spinpoint F3 HD502HJ RAID0 on P67

 

[Vengance_01]

Looks like the array is fragmented.

 

[EnderW]

Looks like the array is fragmented.

 

[ugly]

Doesn't the access/write time of raid array still depend on where on the platter the data is being written? Even without fragmentation, the drive will be accessing data from inner portion of the platter if there's data on the drive I would think.

 

[ghost6303]

if the first screenshot was taken on an empty array, where the data would be written to the outter edges of the platters, and the second was taken with only 30% free space remaining so the benchmark data would be almost on the inside edge, those numbers could make sense.

 

[drescherjm]

if the first screenshot was taken on an empty array, where the data would be written to the outter edges of the platters, and the second was taken with only 30% free space remaining so the benchmark data would be almost on the inside edge, those numbers could make sense.

That is what I was thinking however I am not sure how CrystalDiskMark is reading / writing to the disk. If they are just allocating large files then this makes a lot of sense.

 

[ugly]

Yep, I think that's how crystalmark works. Open up the explorer while you're testing the drive - you can see the folder that crystalmark generated in preparation for the tests.

 

[drescherjm]

So it basically benchmarks what the current rate will be for newly added files.

 

[Old Hippie]

4 x Samsung Spinpoint F3 HD502HJ RAID0 on ICH10R

4 x Samsung Spinpoint F3 HD502HJ RAID0 on P67

I had always assumed the P67 equaled the ICH10 but it's still an apples to oranges comparison.

 

[EnderW]

I had always assumed the P67 equaled the ICH10 but it's still an apples to oranges comparison.

I tested the array on the original ICH10R a few weeks ago before I upgraded and got similar speeds to what I'm getting now

 

[EnderW]

if the first screenshot was taken on an empty array, where the data would be written to the outter edges of the platters, and the second was taken with only 30% free space remaining so the benchmark data would be almost on the inside edge, those numbers could make sense.

I think that may be right