Intel says SSD overclocking is possible, will demonstrate at IDF 2013 (Update: Confirmed)

Intel says SSD overclocking is possible, will demonstrate at IDF 2013 (Update: Confirmed)

At the Intel Developer Forum in San Francisco, Intel will perform the first public demonstration of an overclocked SSD. Details are very slim at the moment, but in theory it should work: Solid-state drives, like your DRAM or CPU, are driven by a clock — it should be possible to boost an SSD’s performance by meddling with that clock, or perhaps altering the timings. Whether the overclock reduces stability and data retention/integrity, though, remains to be seen.

The SSD overclock will be demonstrated at IDF 2013 on September 10, at a session titled “AIOS001 – Overclocking Unlocked Intel® Core™ Processors for High Performance Gaming and Content Creation.” Most of the session deals with overclocking Haswell-based CPUs using software and hardware tools, and also the upcoming Ivy Bridge-E-based Extreme Edition chips. Hidden among the session’s agenda is this little bullet point: “First public demonstration of overclocking Intel® SSDs.” [Intel has actually since updated the agenda to read "Demonstrating overclocking SSD technology."]

In theory, overclocking an SSD should be possible. SSDs are based on NAND flash, and just like your DRAM, CPU, or GPU, the actual performance of the NAND is governed by external factors such as the clock speed and timings. The chips that make up your SSD aren’t intrinsically fast or slow; they’re just a bunch of NAND storage cells crammed into a package. It is the SSD controller that actually governs how fast you and read or write to those cells. Just like other components in your PC that you can overclock, some of those NAND chips, due to their design or binning, will be capable of operating at higher speeds or more aggressive timings. In theory, you might be able to change the firmware on the SSD controller to feed those NAND chips with a faster clock, a higher voltage, etc.

The question, though, is whether it actually make sense to overclock an SSD, or whether this is just a fun proof of concept. As we move towards smaller NAND geometries and triple-level cells (TLC), one of the bigger issues is the reduced number of program/erase (P/E) cycles that each cell is capable of before it dies. (See: Self-healing, self-heating flash memory survives more than 100 million cycles.) It is very likely that anything more than the smallest of overclocks would reduce the life expectancy of an SSD. There’s also the matter of whether there’s even enough space on the SATA 3.0 bus for higher transfer speeds. (See: New server-class SSDs from Samsung and Intel reach 800GB, up to 800MB/sec.)

The most likely scenario is that Intel is just having a bit of fun. For enthusiasts, reduced drive life isn’t a huge consideration — or, to put it another way, if Intel can squeeze 20% more performance out of an SSD, but at the expense of 20% reduced lifespan, then I think a lot of overclockers would take that gamble. Once you’ve upgraded from a HDD to SSD, and perhaps set up a RAID 0 array, long-term storage is one of the biggest performance bottlenecks. If Intel has found some way of quickly and safely improving your SSD’s performance by even a few percent, it’ll be very interesting indeed.

ExtremeTech will be at IDF. We’ll hopefully witness the SSD overclock first-hand and report our findings.

Updated: Myce has delved into a hex dump of some Intel SSD software and discovered some strings (pictured above) that clearly show that Intel SSDs can be overclocked — either by changing the controller frequency in MHz, or the bus frequency of the NAND (either 83MHz or 100MHz).