I don't think that mass or density play much here as do things like drag coefficients (shape).
Drag coefficients depend on flow direction, speed, body shape, and aspect ratio. For instance, if the CG of a cylinder is the same as the centre of volume (buoyancy centre), then the cylinder will fall horizontally, and the drag, a function of the area presented to flow, is governed by length and diameter. If, on the other hand, the CG is offset from the CV, then there is a tendency for the descent to slip sideways, presenting less of an area. For significant differences between CG and CV, the aspect will favour a blunt end-on descent, and given that the area is smaller, lower drag and higher terminal velocity.
The US Navy conducted a number of studies circa 2000 for determining the location of mines on the sea bottom from airborne and ship launch platforms. I recall there is also data from research carried out by the Oil industry where the interest was in determining what happened (trajectory) to drill stringers when they fell off rigs in deep water.
The link below shows the behaviour of a cylinder (1M L x 0.17M D) when descending to sea bottom, aprox 12M deep, at a US Navy test range.
ImageShack® - Online Photo and Video Hosting
a) CG = CV horizontal entry
b) CG = CV vertical entry - trend is to revert back to horizontal descent.
c) CG slightly deviated from CV angled entry with trend to 'See - Saw' descent.
d) CG deviated from CV more than case (c) - angled entry with trend to plane in the direction of CG displacement from CV.
Given the above and assuming the memory module has a CG coincident, or nearly so, with CV, and that during the descent to the seabed the module was not wedged in any co-descending wreckage, it is reasonable to suppose that it landed horizontally on the silt bottom. Whether or not it is hidden under other wreckage is another matter.
I would be interested to hear if anybody on this forum has experience of how surface mount (Mil spec or otherwise) survive under 400 atmos pressure - I'm thinking here about the QDR memory, and the various non volatile memories in other subsystems FADECs etc.