New Intel 320 SSD: Smaller, cheaper, less reliable?

SSD (Intel)

Hurrah! SSDs are getting less expensive. And the reason is our old friend, Intel co-founder Gordon E. Moore and his eponymous law -- Moore’s Law forecasts that the density of integrated circuits doubles every two years -- which usually causes falling prices. Read how Intel's new 320 Series SSDs are an unusual illustration of how this works, in The Long View... SSDs are following the classic pattern. Let's use Intel's SSD product history as an example: Its first generation drives used 50nm process technology and were eye-wateringly expensive. Its followup X25-M G2 drives used a 34nm process and were (relatively) inexpensive -- something around $2 per GB. Now we're finally seeing the next jump in density: 25nm. This is going to drive street prices of mainstream SSDs down, say, 20% soon -- dropping even more once production volumes ramp up. I'm pleased to see that my worst fears about Intel's 25nm fab seem to be unnecessary. Here's why: read on... Four weeks ago, I "graced" this blog with my concerns about the state of 25nm NAND-flash production:

What are we to make of the no-show 25nm parts? This has all the hallmarks of a fab in trouble ... [which] usually means a low-yield fab ... [because] the defect rate is too high.
...
Perhaps the parts coming out of the new fab worked OK up to a point, but were more restricted in their reliable write cycles than expected? ... If the 25nm technology isn't up to even ... typical consumer workloads, we could be looking at a serious miscalculation by Intel.

Intel's new 320 Series SSD line is an unashamedly mid-range product. Its high-end desktop drives are the 510 Series, which still use 34nm flash chips. However, as I also mentioned four weeks ago, the 510 line is new. That's weird: technology companies usually sell the new stuff to high-end users at a premium price.

What's going on here? How come the 510 drives use the old flash technology?

The key reason is reliability: the new 25nm process means smaller NAND cells. And smaller NAND cells retain significantly less charge. Less charge in a cell means they're more prone to charge leakage -- which is especially worrisome when the cells are MLC (multi-level cell), as opposed to SLC (single-level cell).

To continue reading this article register now

  
Shop Tech Products at Amazon