Intel's turbo memory needs a boost

Despite impeccable engineering and massive marketing, part of Intel's mobile plan seems to have stalled
Written by Leader , Contributor

Those with tears to shed, prepare to shed them now. Two years ago, the finest minds at Intel started to prepare the specifications for this year's Centrino update by looking very hard at everywhere performance and power consumption came into conflict. Arguably, the company knows more than anyone about this: as creator of the most successful mobile PC platform and owner of the largest collection of suitably trained brains, Intel is no stranger to intensive analysis.

One of the components that came out of that consideration was code-named Robson, now known as Turbo Memory. The idea seemed impeccable: hard disks are slow and take lots of power, especially when you have to load in data from lots of different bits of the disk in quick succession. That happens particularly at boot-up, as well as for certain applications. Stick that data in flash memory, thought Intel, and you can boot far more quickly for less wattage, as well as getting useful acceleration during actual work. As befits such a meticulous company, it did tests and presented the incontrovertible findings to the world. Intel was ready for the future.

That future has arrived — and Robson is getting a cold shoulder. HP and Sony have both said that they're not going to bother — for now, at least — while other companies have said they will have it, just not in the first round of products. What went wrong?

Intel did the best it could. Its calculations were right, and we have no reason to doubt that the technology works as expected. The rest of the world, though, is not quite as Turbo-ready as it might be. HP has calculated that more main memory makes Vista and other operating systems work better overall. Sony is saying that for the user benefits to be apparent, Vista needs to be updated, so why spend the money now? Both statements highlight one of the main problems with any sort of caching system: by sitting between hardware and software, it needs both to be just right to make economic and engineering sense.

The lesson for us all is that even the best prepared, most iron-clad proposition may not survive contact with the real world. Intel made the smart decision not to make Robson a compulsory part of the Centrino platform: it has time to come good without dragging down meantime. Nonetheless, the company's still in minor denial. Don't let a surefire thing blind you to the reality that environments change without asking permission. Markets are what happen while you're busy making other plans.


Editorial standards