Intel Announces Jim Keller's Departure, Other Leadership Changes

[phoronix]

Phoronix: Intel Announces Jim Keller's Departure, Other Leadership Changes

Legendary processor engineer Jim Keller has resigned from Intel just over two years since he joined the company to much fanfare...

http://www.phoronix.com/scan.php?page=news_item&px=Intel-June-2020-Leadership

[schmidtbag]

Interesting...
Back when Intel first poached him, I don't think they actually really wanted him for themselves, they just wanted him to stop helping AMD. They probably paid him to just bounce some ideas off of and sign a no-compete clause, but otherwise had no intention to actually use his skills; Intel clearly hasn't used him for any of the CPUs they've released in the past 2 years, and it seems their next overhaul of an architecture was pretty much already figured out by the time they got him.
So, now he's probably like "y'know what? I don't feel like sitting here and do nothing. I'm out of here. Let me know if you actually care about my opinion".

[igxqrrl]

Interesting...
Back when Intel first poached him, I don't think they actually really wanted him for themselves, they just wanted him to stop helping AMD. They probably paid him to just bounce some ideas off of and sign a no-compete clause, but otherwise had no intention to actually use his skills; Intel clearly hasn't used him for any of the CPUs they've released in the past 2 years, and it seems their next overhaul of architecture was already pretty much already figured out by the time they got him.
So, now he's probably like "y'know what? I don't feel like sitting here and do nothing. I'm out of here. Let me know if you actually care about my opinion".

Every sentence above is incorrect.

[schmidtbag]

Every sentence above is incorrect.

Proof? Not that I insist I'm right; it's all conjecture. But at the very least: what's 1 thing Keller did to improve Intel CPUs available today?

[torsionbar28]

Every sentence above is incorrect.

link?

[igxqrrl]

Proof? Not that I insist I'm right; it's all conjecture. But at the very least: what's 1 thing Keller did to improve Intel CPUs available today?

There's nothing here that one can "prove" without divulging inside information. Though there are plenty of articles you can google on the webs discussing the significant roles that Keller and Koduri were given. Not like "hey, can you optimize this branch predictor and improve die-to-die latency?" But more "Hey, you are now in charge of 20,000 employees. Improve our design cadence". Intel paid big money to get these guys (and a handful of others). It wasn't to keep them as figureheads or to keep them from helping other companies, it was a (desperate) attempt to revamp the IP and SOC development and manufacturing flow at Intel, which is extremely hard to do in what is a very insular culture.

2 years is too short of a timespan to see his influence on current chips. The development cycle is far longer than two years. I'll just say that my response wasn't conjecture, but nobody has to believe me! I no longer work at Intel but I can imagine with what I suspect is some accuracy exactly why he's leaving, and why a few other key hires may follow.

[angrypie]

I no longer work at Intel but I can imagine with what I suspect is some accuracy exactly why he's leaving, and why a few other key hires may follow.

Did he leave because Intel is beyond help?

Raja isn't doing much there either, if their dGPU can't even beat an RX 560.

[Jabberwocky]

Interesting...
Back when Intel first poached him, I don't think they actually really wanted him for themselves, they just wanted him to stop helping AMD. They probably paid him to just bounce some ideas off of and sign a no-compete clause, but otherwise had no intention to actually use his skills;

He went from Tesla to Intel.

In August 2012, Jim Keller returned to AMD
In January 2016, Keller joined Tesla, Inc
In April 2018, Keller joined Intel.

Intel clearly hasn't used him for any of the CPUs they've released in the past 2 years, and it seems their next overhaul of architecture was already pretty much already figured out by the time they got him.

This conclusion seems a bit farfetched. It would have been funny/stupid if it was the real reason for leaving.

So, now he's probably like "y'know what? I don't feel like sitting here and do nothing. I'm out of here. Let me know if you actually care about my opinion".

Working conditions is likely really bad at Intel at the moment, but a good engineer would not be too phased about that. I'm also wondering if they simply didn't listen to him. I'm sure he'd move along if that was the case.

Anyway I'd wish him all the best!

[wizard69]

Interesting...
Back when Intel first poached him, I don't think they actually really wanted him for themselves, they just wanted him to stop helping AMD. They probably paid him to just bounce some ideas off of and sign a no-compete clause, but otherwise had no intention to actually use his skills; Intel clearly hasn't used him for any of the CPUs they've released in the past 2 years, and it seems their next overhaul of an architecture was pretty much already figured out by the time they got him.
So, now he's probably like "y'know what? I don't feel like sitting here and do nothing. I'm out of here. Let me know if you actually care about my opinion".

Sadly people experience that in big corporations across the USA. I look at your comment and all I can say is that he hit the bullseye.

[wizard69]

There's nothing here that one can "prove" without divulging inside information. Though there are plenty of articles you can google on the webs discussing the significant roles that Keller and Koduri were given. Not like "hey, can you optimize this branch predictor and improve die-to-die latency?" But more "Hey, you are now in charge of 20,000 employees. Improve our design cadence". Intel paid big money to get these guys (and a handful of others). It wasn't to keep them as figureheads or to keep them from helping other companies, it was a (desperate) attempt to revamp the IP and SOC development and manufacturing flow at Intel, which is extremely hard to do in what is a very insular culture.

I don't want to dismiss your point of view out of hand but my experience with big corporation in entirely different industries by the way, mirrors what Schmidtbag has to say. In a nut shell: if somebody that is obviously smart is hired, you will have people (usually not too smart) undermining his every move. I've seen this multiple times so maybe my perspective is skewed. He might not have been hired to be a figure head but I expect that he was constantly bumping heads to get anything done.

2 years is too short of a timespan to see his influence on current chips. The development cycle is far longer than two years. I'll just say that my response wasn't conjecture, but nobody has to believe me! I no longer work at Intel but I can imagine with what I suspect is some accuracy exactly why he's leaving, and why a few other key hires may follow.

It would be interesting to hear more about why you believe people will be leaving Intel. Personally I'm expecting a bit of a public and very large reorganization soon. We seem to be seeing some minor changes already.