Intel showed off the performance gains you can get with SSD overclocking at this year's IDF in San Francisco, but then admitted it was still very much a work in progress.
A few weeks back it was revealed they would be showing it off in an overclocking session and some enterprising chaps went digging around in Intel's Extreme Tweaking Utility and uncovered code relating to it. Unfortunately in the end we were only given a very quick demonstration of the possibilities of SSD overclocking, but I had the chance to chat with Dan Ragland, one of Intel's Senior Systems Engineers, afterwards.
“To be honest, we're just playing with it in the labs,” he explained. “We're still in the experimental stage at the moment.”
In the short demo Ragland showed results from overclocking the memory controller from the standard 400MHz up to 625MHz. That extra clockspeed ended up delivering around a 10% performance boost in a general AS SSD benchmark.
I asked whether that ten per cent would apply to both random read/write performance as well as sequential read/write performance, but because of the very experimental nature of the whole process right now he wouldn't be drawn into specific benefits.
“I gave you a snapshot of one benchmark, it varies,” he said. “There are some areas where you'll see a whole smattering.”
Probably more important than any actual or perceived performance improvements though is just how stable overclocking your SSD would be. For me, the benefits to overclocking my boot drive would be irrelevant if I couldn't guarantee the integrity of my data.
So I collared a representative from the NAND Solutions Group (NSG) to see what assurances they could give me about tampering with your drives. “Any risks you take with your data in overclocking the memory controller has to be monitored,” he explained. “But we have those monitors.”
There are then these 'monitors' which keep an eye on both the memory controller and the NAND chips themselves. If these systems start detecting elevated levels of stress inside the NAND then the software would be alerted to them. In a commercially available application, possibly as part of the XTU software, it would then dial back the overclock to protect your drive.
Ragland also showed controls to offer different NAND frequencies, but if overclocking the memory controller is fraught with danger then messing with the NAND could be positively suicidal.
“Controller overclocking is what you want,” explained my new friend from NSG. “The intelligence is in the firmware and you don't want to stress the NAND so much.”