Yes folks, the cloud and parallel processing need killer apps
Microsoft's research chief Craig Mundie outlined his view of where computing is headed over the next decade or so and here is a big takeaway: The cloud and multiprocessor computing could use a few good killer applications.Mundie was speaking at the Technology Review Emerging Technology conference.
Microsoft's research chief Craig Mundie outlined his view of where computing is headed over the next decade or so and here is a big takeaway: The cloud and multiprocessor computing could use a few good killer applications.
Many consider Software + Services (S+S) to be Microsoft’s way of keeping its PC-software money-making machine afloat while the cloud-computing waves come rushing in. But that view ignores the reality that it actually does make sense to run some applications and/or pieces of applications locally, and others off-premise in remote data centers.
In other words, Microsoft thinks the cloud and client will merge and enable new applications. Sounds OK, but is that a killer app?
News.com chief Dan Farber highlights Mundie's talk, which also revolved around spatial computing as the next big thing. Think 3D technology that adapts to you and gives you a feeling of being there. Naturally, this vision will require parallel programming and more processing cores than you can imagine.
Here's a look:
Much of Mundie's talk rehashed familiar themes from Microsoft Research. Mike Ricciuti, who is riding shotgun with Dan, also highlights how Mundie thinks cloud computing needs a killer application too. Mundie's entry into the killer app derby revolves around a mix of the cloud, robotics and sensors --and, naturally, client software (cause you have to use Windows somehow).
While Mundie's chat is just swell, the larger question for me is this: Can you predict a killer app? Historically, killer apps aren't preplanned. And they usually catch folks by surprise.
Spatial computing in 3D, robotics, the cloud and sensors have been on Microsoft's drawing board forever. It's interesting stuff. But none of Mundie's entries are likely to be killer apps.
I do wish we'd find one soon, though. It's particularly worrisome to me that we have all of these cores in chips and there isn't a lot of software taking advantage of them. The mystery killer app will emerge once software vendors figure out how to best take advantage of parallel processing.