I remember reading the Larrabee white papers, press material, etc. for an article[1] I wrote some five years ago. I think the reason that Larrabee was pitched f

I remember reading the Larrabee white papers, press material, etc. for an article[1] I wrote some five years ago. I think the reason that Larrabee was pitched for graphics initially was just to limit the domain of "things" (tessellation, texture sampling, etc.) that had to be catered to in silicon. I've no doubt that had it come to market, and had it worked well, we would be looking at a radically different CPU architecture than Haswell on the market today. Of course, there's always the outside chance that the experiment worked so well that Intel is keeping an architecture inspired by the Larrabee research project it in reserves for after it's gone as far as it can shrinking transistors and needs something new to sell. [1] http://www.trustedreviews.com/opinions/intel-larrabee-an-introduction http://www.trustedreviews.com/opinions/intel-larrabee-an-int...