CPU Guru Jim Keller leaves AMD
Click here to post a comment for CPU Guru Jim Keller leaves AMD on our message forum
Aura89
This guy is so flip floppy when it comes to which place has his employment that its not really surprising nor do i really care.
Asgardi
The interesting question is obviously "why". Did this guy see that "we are done" and leaves before inevitable failure, or did someone buy him off? Or some other issue like health?
Ryu5uzaku
That was like his longest time he spent at AMD, he does not stay long anywhere if you look at his history. Earlier he was at AMD for 1-year to make the best AMD cpus. 😀
Should not affect anything if it regards first gen Zen.
d_mouse
I might be confused but hasn't he only been at 3 companies in the 20 odd years?
Loobyluggs
Jim leaves.
Jim sell shares.
Microsoft place offer to buy AMD.
Immediate share price increase means Jim's shares would have been worth several hundred million more than he sold them for.
Ryrynz
Meh. The important news is that Zen unlikely until to Q4.. WTF.
Dazz
The CPU has been done for a few months now, June or was it July they finished it, it's now being taped out. Global foundries are still not capable of doing 14nm in sufficient volumes. From Jims history i would say he is a contractor they do their work test it make sure it's sufficient to the clients requirements then move on, they are not an employee and they do not hang around.
Noisiv
Jim was responsible for overseeing the teams defining the roadmaps for AMD’s CPU cores, systems IP and server and client SoCs.
Jim’s departure is not expected to impact our public product or technology roadmaps, and we remain on track for “Zen” sampling in 2016 with first full year of revenue in 2017.
Can you notice the evolution of wording? Now it's sampling in 2016.
Means mass production 9-12 months later.
I'd be very very surprised to see Zen in stores in 2016.
Ryu5uzaku
So Intel is getting my money once again next year since I feel the need to upgrade my 3770k. The link AMD Joe posted does talk about on track for 2017 so ok AMD needs to survive 2016 without any new cpu products + some new gpus holy what..
Noisiv
Why?
3770k should hold its own in 2016.
Problem with Zen is they have set their target in stone well ahead of launch, and are hoping for the best.
As if Intel can't do nothing about its own products.
What if Intel perf. leaps?
What if Zen does well against Skylake (and that is certainly an "IF"), but Intel decides to cache in its superior perf/W into pure perf?
AMD is like a chess player who thinks only about its own moves. And then they ***up even that.
oh so now it's on track for 2017...
Ryu5uzaku
Yeah it should but I had planned a new build and giving this build to my gf so 😀 I was waiting on Zen but rather not wait 1½ years for it. If Zen does good vs Kaby Lake because that is what they are up against then AMD might have a winner if it does good vs Skylake it will sell. If it does worse then either of those it will leave AMD in a dire spot.
But yeah we will see how future bodes for AMD.
Noisiv
Well yes... being 2017 product it seems like Zen will, at least partially, have to go against 10nm Kaby (2H'17).
But everyone is having problems with 14nm, and I can't see Jim leaving at this point in time being part of well-thought process.
Something clearly went awry.
Ryu5uzaku
Kaby Lake seems to be 14nm part now a refresh of skylake and Cannonlake 2017 10nm part in most recent intel roadmaps and press releases.
I think it is more that Jim finished what he set out to do and left for another challenge, that has been his way of doing things it seems. But eh we can only guess why he left.
Noisiv
Right... having to go against Cannonlake.
10nm, 2H'17
I cant agree with this "he finished what he set out".
Because that assumes all AMD needed was a little bit of Jim.
Then why let him go, and how is that not a bad thing - if he's that indispensable, and he turns to gold everything he touches .
Another fight is on the horizon: assuming all goes well with Zen. They will have... what... 3Q before Cannonlake shows up.
Denial
They didn't let him go, he just left. He was at Apple prior -- I'm sure if Apple wanted to keep him they would have too, as he designed the highly successful A6 including Apple's custom Swift core. He left there the same way, finished the design and left before it even shipped in a product. He kind of just floats around solving engineering problems for various companies.
fantaskarsef
Sounds like a true engineer would, what he's interested in might be solving problems, rather than making a brand big. But in the end, it doesn't change anything atm I guess.
Kaarme
CPU market and development has been so stagnant precisely because Intel has dominated it, AMD being unable to challenge anything but the very cheapest (no profit) part of it. Intel hasn't bothered to do anything about the performance, nor the number of cores (aside from specialty processors). It's clear as day Intel would have the know-how and technology to release amazing multi-core CPUs, but instead we have been living with similar quad cores for a decade because there's been no incentive to do anything about it, in lack of competition. Intel makes more money than a small country for its owners, which of course is the ultimately purpose of a commercial enterprise, but it still sucks for consumers.
So, if Intel suddenly took AMD and Zen seriously, it would be a win-win for the whole world. I don't see any problems.
H83
Noisiv, from what i´ve read in other sites, Zen is going to be "sampled" in the first months of 2016, so it can be released in mid 2016 while in 2017 it´s going to be the first year when Zen is going to be sold for a complete year, leading to a full year of revenue selling Zen´s cpus.
I think they´ve used the word "sampling" wrongly or something but i can be wrong about this.
Noisiv
They've used "sampling" before in public press (Buldozer). Meaning first working samples which are sent to clients.
07/16/2010
Advanced Micro Devices, the world's second largest supplier of microprocessors, said that it had taped out the first central processing units (CPUs) based on the code-named Bulldozer core.
The company hopes to begin sampling of the chips with customers shortly and initiate mass production sometimes in 2011.
It ended up like this:
Tape out Q3'10, sampling begun Q4'10, and final product was released Q4'11
I.e. 12 months between sampling and release.
So far they have NOT made any statement about Zen being taped out. They did mention few FinFET designs tape out.
But if that had been Zen tape out, they would have said so. No reason not to.
And if Zen did tape out in July 2015, why announce sampling as late as sometime in 2016.
Noisiv
With titles like Q4 2016 at the earliest and similar floating around... just forget about 2016.
From AMDJoe's link:
Now AMD finds itself in another difficult place — and company leaders think they know the name of the potential escape hatch.
It is called Zen — a microprocessor design project whose results are expected to start showing up in a family of new products starting in 2017.
