Surprise, surprise - Samsung has won a game of flash leapfrog and is producing a 20nm part way ahead of Toshiba and IMFT.

Smaller flash memory processes mean more flash chips can fit on a wafer, lowering their individual cost. It also means that a given device can be loaded up with more flash capacity in the same space, so a smaller process brings both cost and capacity advantages.

Samsung, which has been building 30nm parts, is making 20nm-class - it's not actually saying what the actual nanometre dimensions are - process SD cards, with multi-level cell (MLC) 32Gb chips and sampling them. Capacities include four, eight, 16 and 32GB with a 64GB part coming. Mass production is slated to start by the end of the year and it reckons it'll get half as many chips again from its 20nm process as it does from the current 30nm one.

It should herald a speed increase as well. SD cards using this flash are up to thirty per cent faster at writing data, Samsung says. This means 10MB/sec - hardly lightning speed. Read speed is said to be 20MB/sec. That will be good news for digital camera and video recorders that could use Samsung flash chips.

The three main flash foundry players are IMFT (Intel Micron Flash Technologies), Samsung and Toshiba. Toshiba is beginning a transition from 32nm parts to an undecided but sub-30nm process. Intel and Micron are strengthening the use of their 24nm process, though they seem intent on enhancing profit margins rather than lowering prices. Samsung's move could change that. It's likely that Toshiba will try and make a jump to a process size as near 20nm as practicable, rather than targeting a 29-25nm process.

Interestingly, the owner of the Korean Samsung 20nm process foundry has just taken a stake in Fusion-io, a supplier of PCIe-connected flash solid-state drives. This should mean an increase in Fusion-io product capacities, once Samsung makes parts for Fusion using the new process. ®