Benchmarking for FCP X media drives

I'm using QuickBench 4.0 to benchmark my different hard drives. I can't seem to find a straight answer on the web. . .

Specifically for video editing drives, is it better to have the "Allow Cache Effects" box checked or not when benchmarking media drives? OWC seemed to think so. But then again, they've been wrong about a few minor things in the past (depends who you talk to there).

Thanks,
Dave

When comparing drives you should do the test the same way for each. That way the results are relative. There would be no harm in testing each drive both ways. You just can't compare a drive tested with cache effects enabled to one with cache effects disabled. It's not apples to apples.

anickt wrote: When comparing drives you should do the test the same way for each. That way the results are relative.

Yes, I figured that part out. But, which is more accurate for the way a media drive is being used for editing in FCP X? No, I'd rather not do both. I only want to know the results as how they best relate to editing. The results can be quite different with it on or off and it also seems to make a difference as to how full the drive is.

Thanks,
Dave

- To simply test raw read/write speeds turn caching off.
- To test with a specific app for real world performance, turn the caching on.

BenB wrote: - To simply test raw read/write speeds turn caching off.
- To test with a specific app for real world performance, turn the caching on.

Which would you say is more valid for what we do in FCP X?

(By the way, I'm a bit fried at the moment to add all the details, but I have the benchmarks with the hybrid drive when connected via eSata Expresscard. Also, it's now installed in the main drive slot and I moved over the boot drive to the optical drive slot. The 2 edits I did were just fine connected via eSata. I'll try an edit tomorrow with it as an internal 6GB connected drive and see how it feels. It benchmarks much faster as an internal drive. I'll update the other post with the benchmark numbers and conclusions in the next day or two.)

Thanks,
Dave

- To test with a specific app for real world performance, turn the caching on.

That includes FCP X, as the caching is on when you're working with apps on a daily basis.

It is turned off for "level playing field" benchmarking because caches on various drives are not created equal, and it can cause benchmarks to vary quite a bit. But in the real world the cache is working.

Excellent. I have been writing down both versions, but I think I'll just stick now with "Allow Cache Effects" checked from this point on.

The QuickBench manual recommended to set the Read and Write Delay to 5ms. Does that make sense to you? Before I changed them, the defaults were at 20ms and 30ms. These do seem to make a bit of difference in the results, but it also seems to make a difference how full the drive is also.

The trick of course is to do all tests with the same parameters set and it seems best to do the comparisons of drives all at the same time.

Thanks,
Dave