Intel Discrete GPU Boss Raja Koduri Explains Why He Left AMD
Following the launch of Vega, a whirlwind of events happened, involving former Radeon Technologies Group boss Raja Koduri. What started off as a sabbatical from AMD turned into the head of graphics leaving the company, and then joining Intel to spearhead the Santa Clara chipmaker's own foray into discrete GPUs, something it has not done since Larrabee. So, why exactly did Koduri leave AMD and join Intel?
Koduri somewhat addressed that question during a recent interview with Barron's. What it boiled down to was evaluating the next decade and seeing where his talents would be best utilized.
"When I took a break from AMD my thinking was, What am I going to do for the next 10 years? What is the problem statement for the world for the next 10 years? Where I landed was, we’re in the middle of a data explosion. The amount of data that is being generated in the world is way more than our ability to analyze, understand, and process," Koduri said.
He also thought about the technologies that would be required to keep pace with the data explosion.
"As I was thinking through that and the elements that are required, the core pieces of technology required, and which company has these assets, these people, these resources, the only company that checked my list...was Intel," Koduri added.
It's an interesting answer, albeit a little vague and perhaps not entirely forthcoming. Koduri did not explain what exactly Intel brought to the table that was not available at AMD, where he led the development of Vega and put in place a graphics roadmap that extends into at least one future GPU architecture.
Now the world waits as Intel embarks on its graphics Odyssey. The marketing machine is in overdrive these days, with Intel generating a ton of hype. Whether it can live up to it all remains to be seen, and we will get our first real look in 2020, when Intel is scheduled to deliver its first modern discrete GPU. In the meantime, Koduri is exuding confidence that Intel will be competitive from top to bottom in its discrete GPU stack.
"I want to set the record straight that Intel has a world class design team sitting here. What I’m doing is helping them figure out how to build products that scale up from the low power, mobile domain up to petaflops—the big data center GPUs," Koduri added.
He also points out that "Intel has been doing graphics for a long, long time," albeit mostly at the integrated level. In other words, this is not entirely brand new territory for Intel, though certainly what the company is attempting to do is far more ambitious than past efforts.
In short, everyone at Intel is saying all the right things. Now we just have to wait and see how the company executes its strategy.