Thursday, April 7, 2011

Vertex 2: After the 25 nm and 34 nm, 32 nm

This morning we reported qu'OCZ would indicate the memory used on some models of Vertex 2, adding a ".25" or ".34" depending on the type of memory. But if there is nothing to indicate what memory is it? And although there are now three possibilities. Indeed, OCZ uses memory in 34 nm IMFT, memory IMFT in 25 nm, which is slower but less expensive, but also, and again, Hynix memory in 32 nm.

And this memory is - according to preliminary tests - slower than 25 nm in IMFT. With compressible data, which allows the controller SandForce give the best of himself, the 34 nm memory reaches 215 MB / s and the other two about 221 Mb / s, which seems interesting. With more conventional data, the controller can not (re) compress the game is changing: the first tests provide a write speed of 136 Mb / s in 34 nm, 118 Mb / s in 25 nm and only 95 MB / s in 32 nm.

Recall that data from a computer are often incompressible: video, audio, data programs, etc.. are already compressed upstream. Having an SSD that transfers a text file full of 0 to 220 Mb / s is one thing, struggling to see up to 100 MB / s when transferring a video file in H.264 is another ...

In fact, the easiest way is to wait for the next revisions of Vertex 2 and choose according to your needs: capacity / price model with a 32 or 25 nm and raw performance with a model in 34 nm. Note however that if the Vertex 2 in 25 nm and 32 nm is slower than the models at 34 nm, they are not necessarily bad.

No comments:

Post a Comment