When’s A RAM Not A RAM?
It was interesting to meet the CEO of Numonyx, Brian Harrison, at IEF 2009 earlier this week. Why? Because Numonyx is bringing out a Phase Change RAM and Phase Change RAM is, like wafer scale integration and the universal memory, a bit of an industry joke.
Intel co-founder Gordon Moore worked on Phase Change 40 years ago and, in 1970, wrote a joint paper with Ron Neale on the subject.
I ran into Ron in Sutton High Street a month or so ago and he said the technology was as much a non-starter now, as it was in 1970.
Intel transferred its Phase Change development to Numonyx which suggests Intel shares that view.
But when I asked Harrison, earlier this month, if anything had happened, technologically, to make Phase Change a stable, volume product, he replied: : “The best proof is it’s in production.”
Well, it’s not really the best proof. The best proof would be if anyone used it. He said he’s got 40 customers. But who wants a 128Mbit part when 32Gbit floating gate flash is around?
“People who want a memory with more than 100,000 cycles and more than ten years of data retention,” replied Harrison.
OK. Maybe there’s some of those, but it’s all rather unconvincing.
And look at the lamentable Samsung record on Phase Change. The first press release it put out announcing the part was in 2006.
http://www.eetimes.com/news/semi/showArticle.jhtml?articleID=192700709 2006 announcement saying production will be in 2008.
http://techon.nikkeibp.co.jp/english/NEWS_EN/20070226/128173/ – 2007 announcement says shipments have started, mass production planned 2008.
http://www.i4u.com/article6531.html - also saying mass production in 2008.
http://www.samsung.com/global/business/semiconductor/newsView.do?news_id=1042 - September 2009 announcement that it had started production.
So don’t blame me for sniggering.