HDD for video Question

[skiphunt]

Hi, 

 

I know SSD drives are recommended, faster, etc. but I didn't want to spend the extra $ right now. 

 

Got a Buffalo mini thunderbolt 1TB (5400RPM) that I've been using for almost a couple weeks. During that time I noticed that G-Technology had a mobile 1TB thunderbolt drive with a 7200RPM. 

 

Because I can return the Buffalo for a full refund, I went ahead and spent the extra $30 to get the G-technology 7200 thunderbolt (want to keep them mobile without external power sources needed)

 

Well, the G-Technology drive arrived today so I figured I'd test it before taking the Buffalo back. I figured the 7200 would be faster, but not night/day difference. When I run the Black Magic hard drive speed tests, and the AJA speed tests, they both show consistently on all tests that the G-Technology is faster like 132MB/s vs 100MB/s with the Buffalo.

 

Here's the strange part, when I copy a large folder or large file to the Buffalo and time it, the speed is about the same and sometimes faster than the 7200 G-Tech drive. Consistently.

 

How can it be that even though the 7200rpm G-Tech tests faster in every way via speed tests, the Buffalo 5400rpm actually writes large files at the same speed if not slightly faster?!

[Inazuma]

At a guess id say because the buffalo has a larger cache size

[skiphunt]

I don't know what the cache size is of either drive, but if that were the case, wouldn't the Buffalo have also shown an advantage using the AJA and Black Magic HDD speed tests?

 

There's only $30 between them so cost isn't so much of a factor. I just don't want to keep both of them. Would you keep the one that tests significantly better across the board with two different software speed tests and claims to have a 7200rpm drive? Or, do you keep the one that copies a large file slightly faster with the 5400rpm drive?

[Inazuma]

I think cache size is what ultimately helps with individual large files. I would keep the one that tests better across the board :)

[skiphunt]

That's what I was thinking. Just copied a bunch of FCPX libraries and such over to the one that tests better. Figured I'd see if in practice, it at least feels a bit snappier.