Intel is getting more serious about phase change memory
Chicago (IL) – The chip manufacturer will be dedicating more resources to the development of a new memory technology that is believed to be able to replace flash memory down the road.
According to industry sources, Intel will be setting up a joint-venture with Numonyx to co-develop phase change memory (PCM) technologies. Numonyx, recently created from Intel's former flash memory group and resources from ST Microelectronics, focuses on the development of non volatile memory technologies.
While Intel said that it has been working with ST Microelectronics on PCM since 2003 - a collaboration that resulted in the demonstration of the 128 Mb "Alverstone" PCM chip at the Fall IDF in 2006 – the new R&D effort will combine staff and other resources from Numonyx and Intel.
Sources noted that Intel is playing with the idea that PCM may become the successor of flash the industry has been looking for over the past years, improving not only the write cycles of current flash (an estimated 100 million write cycles compared to only 1 million) as well as increasing the responsiveness and data transfer performance. Possible PCM applications indicated by Intel are future solid state disk (SSD) drives that are fast enough to prevent "multi-core CPU-based systems from choking on I/O limitations."
PCM devices are a type of so called "non volatile" storage that is able to keep the contents in its memory intact when there is no power supplied. Other than flash memory, which stores electrical charge in a silicon or silicon nitride gate, PCM uses a chalcogenide-based technology, similar to what is used today in rewriteable CDs. Heat provided by a laser converts the chalcogenide alloy between its crystalline (conductive) and amorphous (resistive) phases to store data.
PCM is believed to have some drawbacks that still need to be overcome, such as its extreme sensitivity to temperatures. However, what makes technology interesting is the fact that it already has reached impressive densities, indicating that PCM could become economically viable and emerge as a successor for flash memory. In contrast, Magneto-resistive RAM, short MRAM, has also been discussed as a potential successor of flash, but mass-produced chips are only available in 4 Mbit capacities at this time.