Jim Keller leaves AMD

[Shehriazad]

The amount of lack of logic and the presence of huge amounts of doom-n-gloom is just mind boggling.


And I thought this forum was about facts n stuff.


Jimmyboy came to make Zen. Zen is now reaching the completion phase. Keller no longer has anything left to do because AMD will push Zen into every market they can. Keller is let go/leaves because there isn't really anything to do for him/his contract ended/AMD doesn't need him or doesn't want to pay for him currently.

And no...they also wouldn't need him back for Zen+ or anything. Iterating on an existing design is a lot easier than making up an entirely new one.

I'm pretty sure they only wanted Keller to have his name on Zen...he's almost a "Brand Name" himself.

And that AMD knows how to iterate on things they have shown even in the recent past.
Now let's ignore the fact that the Bulldozer family was a failure of an architecture...the iterations are actually very solid. Gen 3 +10%IPC, Gen 4 + 5%IPC....on high performance nodes they would've very easily won over gen 1 and 2.
But that obviously doesn't change the fact that the architecture itself was just not worth anymore "high performance" CPU parts...too much money to be lost/wasted there. Even if they had released a new FX 8 core that is 20-25% faster than 2nd gen...who would buy that if you can have +50% performance...other than hardcore AMD fans of course....because the pricing for sure would not have held up.


I'm not here to discuss the general stability of AMD as a company or their financial power or anything...but simply to say how very obvious it is that Jim Keller left after Zen entered the completion phase.

 

[cytg111]

The amount of lack of logic and the presence of huge amounts of doom-n-gloom is just mind boggling.


And I thought this forum was about facts n stuff.


Jimmyboy came to make Zen. Zen is now reaching the completion phase. Keller no longer has anything left to do because AMD will push Zen into every market they can. Keller is let go/leaves because there isn't really anything to do for him/his contract ended/AMD doesn't need him or doesn't want to pay for him currently.

And no...they also wouldn't need him back for Zen+ or anything. Iterating on an existing design is a lot easier than making up an entirely new one.

I'm pretty sure they only wanted Keller to have his name on Zen...he's almost a "Brand Name" himself.

And that AMD knows how to iterate on things they have shown even in the recent past.
Now let's ignore the fact that the Bulldozer family was a failure of an architecture...the iterations are actually very solid. Gen 3 +10%IPC, Gen 4 + 5%IPC....on high performance nodes they would've very easily won over gen 1 and 2.
But that obviously doesn't change the fact that the architecture itself was just not worth anymore "high performance" CPU parts...too much money to be lost/wasted there. Even if they had released a new FX 8 core that is 20-25% faster than 2nd gen...who would buy that if you can have +50% performance...other than hardcore AMD fans of course....because the pricing for sure would not have held up.


I'm not here to discuss the general stability of AMD as a company or their financial power or anything...but simply to say how very obvious it is that Jim Keller left after Zen entered the completion phase.

This is what puts a wrench in that theory for me (by IDC).
If you could, please counter these points .

Everything here points to this being a top-down decision. The decision makers could not coordinate the promotion of Keller's subordinate in advance of communicating to Keller that he was about to be departing the company, businesses cannot and do not function that way for a good reason.

So I'm with Phynaz on this one. I doubt Keller knew he was leaving the company when he woke up that morning. It was probably just as much news to him as it was to the rest of his organization. But it probably was not news to Mark Papermaster or Lisa Su as they were probably involved in the decision to boot Keller.

 

[dark zero]

Nintendo is not going to go to Intel or Nvida. They would charge too much. Going with an ARM solution or even dropping the console completely would be more likely.

I don't see MS or Sony every dealing with Nvidia again either after they both got burned in the past, so they'll look for another solution (on top of the fact that a 940M would be much slower than the PS4's GPU, and niether console would be able to play any games for the currently avaliable consoles).

That future of the consoles are the wet dream of any Intel fan... The reality will show that the consoles moves to ARM or Power PC again even if they lose developers in the process... But hell... They are already losing big names like Konami....

 

[Arachnotronic]

I am truly astonished that Keller's departure is being spun as positive by some. Mind. Blown.

 

[Azix]

This is what puts a wrench in that theory for me (by IDC).
If you could, please counter these points .

There's actually nothing to counter. The post assumes AMD would have replaced him with someone right away if he told them in advance. They do not have to. His role does not need to be filled, what they needed was someone to manage to roll out etc. Papermaster has the competence and experience in this.

 

[cytg111]

.................................................. Mind. Blown.

To be honest, that happend a long time ago, right? .. right?

 

[Arachnotronic]

There's actually nothing to counter. The post assumes AMD would have replaced him with someone right away if he told them in advance. They do not have to. His role does not need to be filled, what they needed was someone to manage to roll out etc. Papermaster has the competence and experience in this.

Are you implying that AMD is trying to save money by letting go of Keller and giving Papermaster added responsibilities?

 

[cytg111]

There's actually nothing to counter. The post assumes AMD would have replaced him with someone right away if he told them in advance. They do not have to. His role does not need to be filled, what they needed was someone to manage to roll out etc. Papermaster has the competence and experience in this.

So, is Papermaster getting a promotion, demotion or is he dual-wielding paychecks?

 

[Arachnotronic]

So, is Papermaster getting a promotion, demotion or is he dual-wielding paychecks?

Keller reported to Papermaster.

 

[mrmt]

Jimmyboy came to make Zen. Zen is now reaching the completion phase. Keller no longer has anything left to do because AMD will push Zen into every market they can. Keller is let go/leaves because there isn't really anything to do for him/his contract ended/AMD doesn't need him or doesn't want to pay for him currently.

There's actually nothing to counter. The post assumes AMD would have replaced him with someone right away if he told them in advance. They do not have to. His role does not need to be filled, what they needed was someone to manage to roll out etc. Papermaster has the competence and experience in this.

Now I see the light. Zen is the ultimate CPU design and there will be no improvements possible after it, so after completing the design Keller will just retire and AMD will user a new era of record shareholder returns built upon Zen products.

Guys, seriously, if you think "Jimmyboy" finished his mission and left in a planned fashion we wouldn't be seeing the CTO stepping into the shoes of the chief architect, that's damage control SOP, and that announcement would have come well before the departure time, in order to show the management team on top of the situation.

This departure, and the behavior of the AMD fanboys brings a strong deja vu, I can't but recall of Dirk Meyer and Rory Read, when both were fired the fanboy reaction was that the company was fine and nothing wrong was happening, but in hindsight we know that Dirk was fired because of the Bulldozer disaster and Rory was fired because of the spectacular failure of his business strategy. That said, why would you fire your chief architect if not for something related to a project under his watch?

 

[cytg111]

Keller reported to Papermaster.

(i know, it was rhetoric)

 

[cytg111]

Now I see the light. Zen is the ultimate CPU design and there will be no improvements possible after it, so after completing the design Keller will just retire and AMD will user a new era of record shareholder returns built upon Zen products.

Guys, seriously, if you think "Jimmyboy" finished his mission and left in a planned fashion we wouldn't be seeing the CTO stepping into the shoes of the chief architect, that's damage control SOP, and that announcement would have come well before the departure time, in order to show the management team on top of the situation.

This departure, and the behavior of the AMD fanboys brings a strong deja vu, I can't but recall of Dirk Meyer and Rory Read, when both were fired the fanboy reaction was that the company was fine and nothing wrong was happening, but in hindsight we know that Dirk was fired because of the Bulldozer disaster and Rory was fired because of the spectacular failure of his business strategy. That said, why would you fire your chief architect if not for something related to a project under his watch?

Eeverything you just stated is 100% redundant in both competing categories in this thread, deduction and trolldom.
Congratulations on winning the redundant-award. (cmon guys, give it up...)

 

[AtenRa]

I am truly astonished that Keller's departure is being spun as positive by some. Mind. Blown.

The only thing i have seen in this thread is doom for AMD now that Keller is going. I havent seen anyone here saying Keller leaving AMD is a positive thing. But some people understand that Keller leaving now will not effect ZEN in any dramatic way. But of course we have the usual people talking like AMD will not be able to make any new CPUs without Keller.

 

[DrMrLordX]

The amount of lack of logic and the presence of huge amounts of doom-n-gloom is just mind boggling.

Keller leaving isn't a good thing. The best-possible scenario here is that Zen is completely done from a design perspective, leaving Keller with "nothing else to do", which implies that AMD hasn't got the resources or foresight to pursue its successor well in advance. Either that, or they're planning to flog the design for a looooooong time which is essentially what they did with K8.

If Keller needs to settle down and move into retirement, doing so from a position in a company where he is well-respected and where he has done good work before would be highly-desirable. He could probably help to bring up a successor or three and guide AMD in the direction of whatever should be replacing Zen three years from now, assuming AMD is to remain a healthy competitor in the server + desktop markets. Now that he's gone, that isn't going to happen.

I don't see it as an indicator of impending doom. It's par for the course with Keller. It just isn't a particularly good thing, either.

I am truly astonished that Keller's departure is being spun as positive by some. Mind. Blown.

It isn't necessarily a bad thing either. It is some of both.

 

[guskline]

If Keller's leaving won't affect AMD's ability to produce new cpus, why was he hired by AMD in the first place?

Perhaps Keller was done and ready to move on. Perhaps. However, the nature of the departure sure raises eyebrows.

 

[jpiniero]

Should point out that Friday is the day that companies do layoffs in the US; I don't know if that's the case in Europe or other countries. Plus the timing is good... They are now in a quiet period so they don't have to answer questions on it until the results (?). If Keller was leaving on happy terms they wouldn't have tried to bury it or would have announced it before the quiet period started.

I am also taking into account the earlier announcement by AMD to separate out Radeon for likely spinoff or sale. You really can't seperate the two at this point unless you are planning on shutting down the CPU business.

 

[Arachnotronic]

Right now, AMD should be getting back first silicon samples of Zen if the design taped out 3-4 months ago considering fab cycle time is 90-120 days.

I wonder if first silicon came back and things don't look so good ("yay, we hit +40 IPC over XV, but oh crap, it doesn't clock high enough to be competitive"), leading to Keller's termination?

 

[Fjodor2001]

Or, yay, first silicon looked great with good IPC and high clocks. We can finish the rest on our own and don't need Jim K for that phase, so he's free to leave for some other cool project.


Lot's of speculation around here... the reasons for Jim leaving could be anything...

 

[frozentundra123456]

I am truly astonished that Keller's departure is being spun as positive by some. Mind. Blown.

Denial is strong in some circles. OTOH, I think it is being blown out of proportion by both sides. Just because Keller made some great designs earlier when CPU design was much less advanced than it is now does not mean he will (has) done the same with Zen on a shoestring budget. The easy gains have been made long ago in x86. Still, I don't think his leaving AMD is a disaster. They have much more serious problems than the presence or absence of one engineer, no matter how talented.

 

[jj109]

Or, yay, first silicon looked great with good IPC and high clocks. We can finish the rest on our own and don't need Jim K for that phase, so he's free to leave for some other cool project.


Lot's of speculation around here... the reasons for Jim leaving could be anything...

Yes, first silicon looks great so let's quietly push the schedule to 2017.

lol.

 

[NTMBK]

Either Zen is a dud and Keller was let go, or AMD wasn't offering him enough compensation/future prospects to keep him on board.

 

[Fjodor2001]

Yes, first silicon looks great so let's quietly push the schedule to 2017.

Huh?

 

[Fjodor2001]

The easy gains have been made long ago in x86.

Yup for Intel, but AMD still has plenty of low hanging fruit to pick. Now AMD just has to copy Intel's uArch changes from 3-6 years ago years. After that, not much performance improvement has been done by Intel, so the gap will be small.

 

[Racan]

Keller's previous stint at AMD was even shorter (1998-1999), and as we all know K8 turned out to be a huge failure, oh wait...

Seriously now, it's very likely that Zen won't be enough to save AMD, but I don't see this departure as any proof of Zen's failure.

 

[frozentundra123456]

Yup for Intel, but AMD still has plenty of low hanging fruit to pick. Now AMD just has to copy Intel's uArch changes from 3-6 years ago years. After that, not much performance improvement has been done by Intel, so the gap will be small.

Yea, they hired this magical CPU designer to reverse engineer Intel's CPUs. That really makes a lot of sense. It is the antithesis of everything Keller is supposed to do.