The SSD field has been at a crossroads for two years that no one has been able to solve yet in order to launch products on the market. It’s not that R&D isn’t invested, the problem is the voltages that new cells have to withstand due to the increased bits. In this direction Kioxia has something to show, because when Intel completes the 5-bit QLC, the brand has already nand flash for SSD of 7 bit in the oven… well, rather in the freezer.
Neither Micron, nor Intel, nor Samsung, the leadership of NAND Flash is in the hands, for the moment, of Kioxia with the support of WD. A year ago the company announced the 6-bit-per-cell study and now it’s surprising just 12 months later with the 7 bit or HLC.
Super-high-capacity 7-bit SSDs are a ‘hot’ reality
If you know how the voltages of a NAND Flash work and how they change with each increase in bits, you will understand that what is done is a technical milestone at the level of CFET Transistors. According to Kioxia research, in these 7 bits per cell, the common material should be replaced with polysilicon for lower voltage resistance, reduce current leakage that occurs with so much voltage jump, and minimize the threshold of voltage tolerated in read and write operations. .
Why are so many changes in materials, thresholds and resistance necessary? Well, because in 7-bit cells you don’t need anything less than 128 different voltage levels to save or read the information. Let’s imagine a CPU as a single cell with 128 different voltages as much as Cores has one of the new AMD EPYCs and also imagine that said voltage is 10 times higher than the current one because the complexity of controlling them is needed for the LLC.
Logically, this may not seem like much, but the processor would literally boil if it was not cooled accordingly. Well, that’s precisely what’s happening with the new 7-bit NAND Flash, at least for now.
Liquid nitrogen to cool them and poor performance
If we have something this hot and we want to make it work in order to understand its reactions and its nature, then we will need nothing less than LN2 mix with helium approach absolute zero. This is what Kioxia has done and to provide a relevant demonstration of it, it has put its cells at -200ºC.
The performance per cell has not been revealed, but if with QLC it is already lower than TLC, with HLC at the moment we can only expect it very few MB/s. And those are precisely the two challenges, because the temperature must approach positive degrees with consistent performance, and it must increase by almost double digits to make sense for the domestic market. There are no Kioxia dates to see them as a final product, but it is estimated that there are at least 7 or 8 years old have something tangible in your hands.
It’s a very precious time for hard drives, because SSDs have stagnated and magnetically assisted recording technologies for the former are only the first step to regaining the rhythm that a decade ago had lost in the market. . It remains to be seen whether the two types of units will be able to meet the deadlines.