Forum Vs Naplam - Fasted real world storage solution

[One_Hertz]

You second quote suggests that when you disable the cache in WINDOWS, the drive uses 0 cache in the drive. (Atleast for your quote/point to make sense that is what it would need to suggest)

The first thing that doesn't sound correct, is that you have previously stated that a drive with 0 cache will not work at all.

So which is it?

The drive either uses the minimum 256k cache but not the remaining 15.75mb used for wear leveling when you disable caching, or it uses NO cache at all and as you stated previously the drive will not function.

If the x25-m still uses the 256k cache then my point still stands because the x25-m is still opperating with 1600% more cache than the j-micron even with the wear levelling cache disabled.

So with performance already hit with the wear levelling cache disabled, how do you think the x25-m would fare if you replaced the 256k cache with 16k?

You can't turn off the cache that is used for wear leveling. It doesn't let you because the drive wouldn't work at all. When it erases a block, where do you think it would keep the previous contents until it has to write the modified data back?

My second quote suggests what it says. When you click the turn off cache button in windows, there is no stuttering. We are clearly talking about the write buffer at that point as I've already said wear leveling cache can not be disabled.

It lets you turn off the write buffer because that is just for performance purposes. You can turn off the read/write buffer an an HDD too. Will be 5-10x slower. It is the same idea but SSDs don't suffer anywhere near as much. My point stands, the X25 would not stutter even with a 0kb write buffer. It would be slow, but it would not stutter. Stuttering is due to bad controller design and not due to cache. Cache can somewhat remedy the bad controller design, but if the controller is fine to begin with then there is nothing to remedy.