CPU Whiz Jim Keller Leaves AMD

[btarunr]

Jim Keller, one of the lead architects of AMD's x86 CPU architectures, has left the company. He held the post of Chief Architect of Microprocessor Cores at AMD. With his association, AMD's launched some of its most successful CPU architectures, such as the original K7 (Athlon, Athlon XP, Duron); the very first 64-bit x86 architecture, and K8 (Athlon64). Keller then left AMD to join Apple, in its development of the A4 and A5 SoCs, before rejoining AMD in 2012 to begin work on the "Zen" architecture.

Keller's departure doesn't throw "Zen" in jeopardy. "Jim helped establish a strong leadership team that is well positioned for success as we enter the completion phase of the "Zen" core and associated system IP and SoCs," said AMD in a statement. "Zen" remains on-track for sampling in 2016, and its "first full year of revenue" in 2017, which would indicate a market launch some time in 2016. AMD CTO Mark Papermaster will take over as additional charge of Keller's position.



View at TechPowerUp Main Site

 

[FordGT90Concept]

Keller's job (architectural design) was done. Zen is in the hands of manufacturing now. AMD literally has nothing for him to do because they can't afford to put him to work on another architecture. Everything is riding on Zen.

 

[Xzibit]

inb4 anyone

CPU Wiz Jim Keller Leaves AMD

Guess he Wizzed on outah there...

The open question is, What's after Zen ?

 

[Steevo]

Or early samples of Zen suck balls and AMD is transforming into its smaller spinoff companies and throwing the debt load onto one company they will sink while the rest remain viable as a GPU maker, small CPU and integrated chip with IP and custom design option for SOC market, with hopes of getting bought out.


Face it, AMD is bleeding to death, and the only option is for them to break up and force Intel's hand on the monopoly/IP agreements to allow them to be bought out. Its not a question of if, but of when.

 

[Nejc]

of the two options.. FordGT90Concept´s seems more reasonable..

[URL='http://www.techpowerup.com/forums/members/steevo.19251/']Steevo: you have to take into account that when Keller was at AMD it was Intel that was bleeding in comparison.. I sure hope he worked his magic once more, AMD really needs this one[/URL]

 

[natr0n]

Keller's job (architectural design) was done. Zen is in the hands of manufacturing now. AMD literally has nothing for him to do because they can't afford to put him to work on another architecture. Everything is riding on Zen.

The way you say this is reassuring and makes sense too.

I was thinking oh man he's gone, so things must be going down hill.

 

[NC37]

Keller's job (architectural design) was done. Zen is in the hands of manufacturing now. AMD literally has nothing for him to do because they can't afford to put him to work on another architecture. Everything is riding on Zen.

Still, if they have no design plans for after Zen and no Keller around to make them...ehh. 40% IPC is not much. It would just put AMD back on par again. Plus there is also the ARM stuff AMD was looking into. Keller worked with ARM at Apple.

Its a blow to AMD no matter how it looks. Sure he could have been done with Zen but I doubt there would have been nothing to do. Perhaps it did just come down to being able to pay him. Makes one wonder how much they spent to coax him from Apple. Unless Apple was in on it. Guess we'll find out if he rejoins Apple.

 

[Steevo]

Considering the recent spin off of their graphics/integrated/APU (profitable) divisions into a profitable entity....


http://www.amd.com/en-us/press-releases/Pages/amd-demonstrates-2014oct01.aspx


Even the ending of their new press releases is pretty fucking bleak.



"
Cautionary Statement

This press release contains forward-looking statements concerning AMD, including the timing and features of AMD’s future products, which are made pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements are commonly identified by words such as "believes, "expects," "may," "will," "should," "seeks," "intends," "pro forma," "estimates," "anticipates," "plans," "projects," "would" and other terms with similar meaning. Investors are cautioned that the forward-looking statements in this release are based on current beliefs, assumptions and expectations, speak only as of the date of this release and involve risks and uncertainties that could cause actual results to differ materially from current expectations. Risks include the possibility that Intel Corporation's pricing, marketing and rebating programs, product bundling, standard setting, new product introductions or other activities may negatively impact AMD's plans; that AMD will require additional funding and may be unable to raise sufficient capital on favorable terms, or at all; that customers stop buying AMD's products or materially reduce their operations or demand for AMD's products; that AMD may be unable to develop, launch and ramp new products and technologies in the volumes that are required by the market at mature yields on a timely basis; that AMD's third-party foundry suppliers will be unable to transition its products to advanced manufacturing process technologies in a timely and effective way or to manufacture its products on a timely basis in sufficient quantities and using competitive process technologies; that AMD will be unable to obtain sufficient manufacturing capacity or components to meet demand for its products or will not fully utilize our projected manufacturing capacity needs at GF's microprocessor manufacturing facilities; that AMD's requirements for wafers will be less than the fixed number of wafers that it agreed to purchase from GLOBALFOUNDRIES INC. (GF) or GF encounters problems that significantly reduce the number of functional die AMD receives from each wafer; that AMD is unable to successfully implement its long-term business strategy; that AMD inaccurately estimates the quantity or type of products that its customers will want in the future or will ultimately end up purchasing, resulting in excess or obsolete inventory; that AMD is unable to manage the risks related to the use of its third-party distributors and add-in-board (AIB) partners or offer the appropriate incentives to focus them on the sale of its products; that AMD may be unable to maintain the level of investment in research and development that is required to remain competitive; that there may be unexpected variations in market growth and demand for AMD's products and technologies in light of the product mix that it may have available at any particular time; that global business and economic conditions will not improve or will worsen; that PC market conditions do not improve or will worsen; that demand for computers will be lower than currently expected; and the effect of political or economic instability, domestically or internationally, on AMD's sales or supply chain. Investors are urged to review in detail the risks and uncertainties in AMD's Securities and Exchange Commission filings, including but not limited to the Quarterly Report on Form 10-Q for the quarter ended June 28, 2014.
"



** Edit.


About the only thing that may be good to say is that I believe our Silicon dreams are at their end, and improvements to IPC will be limited here forward, depending more on better instruction sets, better handling of multiple cores and specialized hardware. If AMD can pull a rabbit out of their hat and get close to the process that Intel has had they may yet live, but still as a small version of what they were aiming for.

 

[HumanSmoke]

Or early samples of Zen suck balls and AMD is transforming into its smaller spinoff companies and throwing the debt load onto one company they will sink while the rest remain viable as a GPU maker, small CPU and integrated chip with IP and custom design option for SOC market, with hopes of getting bought out.

Possible, but my personal opinion is that he was bought on board for Zen and the SkyBridge interconnect to link Zen with K12. With SkyBridge being canned and K12 lacking priority, his job was reduced to the Zen architecture - which if I read it right was to leverage at least some of the previous architectures into a more competitive part. If Keller has already laid out the logic blocks for the architecture then the immediate job is done. The rest comes down to chip floorplan (which is what the deal with Synopsys was all about in reducing AMD's R&D load) and the fabrication process - which Keller (or anyone at AMD) cannot influence.

The real questions are the timetable ( Globalfoundries/Samsung's 14nm LPP seems to be suffering some issues leading to the revised 2017 date for Zen), and who is working on Zen's successor. Since AMD haven't announced a new chip architect which would be common practice, Keller leaving wasn't part of the original plan. If Suzanne Plummer was always destined to work on Zen's successor, why hire Keller in the first place?

Face it, AMD is bleeding to death, and the only option is for them to break up and force Intel's hand on the monopoly/IP agreements to allow them to be bought out. Its not a question of if, but of when.

One doesn't necessarily lead to the other. If Intel agrees to operate under a wider DoJ Decent Decree, then it can operate as a sanctioned monopoly. Intel already operates under Decent Decree (and has done so in the past - see the Intel / Raytheon case). Just as in IBM's case, so long as Intel does not abuse its monopoly position ( i.e. product and commodity prices remain largely in check) it remains fine. If Intel's sole business was x86 there would be more cause for concern for the DoJ, but Intel's business overlaps into areas where it does not have monopoly status - in the case of memory products, it is actually a minority player.

you have to take into account that when Keller was at AMD it was Intel that was bleeding in comparison.. I sure hope he worked his magic once more, AMD really needs this one

Different situation entirely. Intel backed a dead-end architecture (which is not the case presently), and AMD gained a measure of ascendancy due to being able to raid DEC for IP and personnel. Keller came from DEC (where he worked on EV6 bus which turned into HyperTransport). DEC also provided Dirk Meyer and the RISC architectural attributes for the K7 (and every modern CISC x86 design since), and Keller - along with David Cutler and Robert Short (also ex-DEC, and both employed by Microsoft) devised AMD64.

AMD doesn't have an ailing DEC packed to the gunnels with IP to kickstart them this time, and Intel isn't faltering....so not really the same situation, and expecting a repeat of the Athlon era is wishful thinking IMO.

 

[btarunr]

The open question is, What's after Zen ?

ARM SoCs.

 

[john_]

such as the original K7 (Athlon, Athlon XP, Duron); the very first 64-bit x86 architecture, and K8 (Athlon64)

Probably my English are trolling me? Reading this line makes me understand that you consider K7 as the first 64bit processor and K8 the second one. K8 was the first 64bit.

Keller left AMD in 1999. K8 and x86-64 came out in 2003. Probably that's what happening now. His job is done, he lives much earlier than the first Zen products hit the selves. If he is someone who doesn't like to sit and wait, it's completely logical. Why stay at AMD and wait to see if the company will have the opportunity to market Zen and make it a successful product? Why stay there and work on a Zen 2 or 3, when there is no guaranty that AMD will be alive in 2017 or not be bought before then from someone who is not interested in continuing making x86 products? If things don't go as hoped he will be just losing time. So, he leaves and if AMD is alive and kicking in 2018-2020 he might come back.

 

[Jermelescu]

They will do just fine. He did his job and left, just like back in the days and do wonders.
If AMD will be in trouble in the next 10 years he will be back, that is if he doesn't leave the game.

 

[Dieinafire]

He only went back out of pity in the first place. He saw the writing on the wall for amd as a company and got out because he didn't want to waste anymore of his time.

 

[Assimilator]

The question that people are asking regarding AMD's future is "will Zen be good?", but that's the wrong question. The correct question is "what comes after Zen?", and AMD doesn't have an answer for that, because they are terrible at long-term planning.

Meanwhile Intel is already working on not just the architecture to succeed Skylake/Kaby Lake/Cannon Lake, but the architecture after that one.

Intel isn't winning. They've won. The truth is, they won a long time ago.

 

[AsRock]

Keller's job (architectural design) was done. Zen is in the hands of manufacturing now. AMD literally has nothing for him to do because they can't afford to put him to work on another architecture. Everything is riding on Zen.

A truly nice thought, how ever if that was the case it would make more sense to tell the people that and not say shit.

 

[LiveOrDie]

Looks like his possessor needs a update going on that photo.

 

[MagnuTron]

Possible, but my personal opinion is that he was bought on board for Zen and the SkyBridge interconnect to link Zen with K12. With SkyBridge being canned and K12 lacking priority, his job was reduced to the Zen architecture - which if I read it right was to leverage at least some of the previous architectures into a more competitive part. If Keller has already laid out the logic blocks for the architecture then the immediate job is done. The rest comes down to chip floorplan (which is what the deal with Synopsys was all about in reducing AMD's R&D load) and the fabrication process - which Keller (or anyone at AMD) cannot influence.

The real questions are the timetable ( Globalfoundries/Samsung's 14nm LPP seems to be suffering some issues leading to the revised 2017 date for Zen), and who is working on Zen's successor. Since AMD haven't announced a new chip architect which would be common practice, Keller leaving wasn't part of the original plan. If Suzanne Plummer was always destined to work on Zen's successor, why hire Keller in the first place?

One doesn't necessarily lead to the other. If Intel agrees to operate under a wider DoJ Decent Decree, then it can operate as a sanctioned monopoly. Intel already operates under Decent Decree (and has done so in the past - see the Intel / Raytheon case). Just as in IBM's case, so long as Intel does not abuse its monopoly position ( i.e. product and commodity prices remain largely in check) it remains fine. If Intel's sole business was x86 there would be more cause for concern for the DoJ, but Intel's business overlaps into areas where it does not have monopoly status - in the case of memory products, it is actually a minority player.

Different situation entirely. Intel backed a dead-end architecture (which is not the case presently), and AMD gained a measure of ascendancy due to being able to raid DEC for IP and personnel. Keller came from DEC (where he worked on EV6 bus which turned into HyperTransport). DEC also provided Dirk Meyer and the RISC architectural attributes for the K7 (and every modern CISC x86 design since), and Keller - along with David Cutler and Robert Short (also ex-DEC, and both employed by Microsoft) devised AMD64.

AMD doesn't have an ailing DEC packed to the gunnels with IP to kickstart them this time, and Intel isn't faltering....so not really the same situation, and expecting a repeat of the Athlon era is wishful thinking IMO.

This

 

[R-T-B]

The open question is, What's after Zen ?

Hopefully profit.

 

[GhostRyder]

Well, its going to matter how much of an improvement that Zen actually is and where it falls. If the architecture meets its promises (Or at least close) it will put AMD's performance back into the game enough to make buying their CPU's (Beyond APU's for budget) reasonable again.

 

[hellowalkman]

If you guys remember Zen and K12 were originally slated for 2H 2015 ... I'm sure Keller's contract with amd came to an end.. and since the designing is done they let him go..

http://www.kitguru.net/components/c...h-performance-x86-micro-architecture-in-2015/

 

[john_]

Well, its going to matter how much of an improvement that Zen actually is and where it falls. If the architecture meets its promises (Or at least close) it will put AMD's performance back into the game enough to make buying their CPU's (Beyond APU's for budget) reasonable again.

Even if it is not good enough to equal the IPC of a Skylake or a Kaby Lake processor, at least it will be probably enough to NOT kill the performance of the integrated GPU, giving much better APUs. Because I believe that those Bulldozer modules in the APUs are a pure disaster not only when running programs that only need a good CPU, but also any program that uses the iGPU.

 

[Casecutter]

He's going back to Apple to start the transition from Intel to AMD.... he's that Good!

 

[Lionheart]

Hopefully profit.

That made me laugh, TY.

 

[SaltyFish]

The open question is, What's after Zen ?

Well, its going to matter how much of an improvement that Zen actually is and where it falls. If the architecture meets its promises (Or at least close) it will put AMD's performance back into the game enough to make buying their CPU's (Beyond APU's for budget) reasonable again.

Intel has been sitting on the Core-i architecture for how long now? If Zen does enough to close the gap, will Intel decide to finally move onto a brand new architecture? Is it worth it for Intel to do it though? Sure, AMD would have to bust its own rear end twice over to keep up if Intel does do it, but does the CPU market (general market, not the 1% of enthusiasts who scream for mainstream 8-core 4GHz CPUs) really crave it enough to keep up the R&D costs?

 

[R-T-B]

Intel has been sitting on the Core-i architecture for how long now? If Zen does enough to close the gap, will Intel decide to finally move onto a brand new architecture? Is it worth it for Intel to do it though? Sure, AMD would have to bust its own rear end twice over to keep up if Intel does do it, but does the CPU market (general market, not the 1% of enthusiasts who scream for mainstream 8-core 4GHz CPUs) really crave it enough to keep up the R&D costs?

The i-series comprises several architectures. Every "Tick" is a new microarchitecture in Intel's "Tick-Tock" lifecycle.