Legendary CPU Architect Jim Keller Leaves AMD

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
I guess the title of this story should be "Jim Keller leaves AMD...again." Maybe he is going back to Apple.

Well known for his work during AMD's heyday, Keller (pictured) was involved in the creation of the original Athlon architecture, K7, and then served as a lead architect on K8. After playing an instrumental role in developing the world's first native x86-64 bit architecture, Keller later joined Apple and helped develop the company's A4 and A5 SoCs before rejoining AMD in 2012 to spearhead the firm's upcoming Zen architecture.
 
What i'm reading is "AMD executives think they can get us out of the RED by slashing R&D and optimizing costs".
 
Trepidati0n said:
What i'm reading is "AMD executives think they can get us out of the RED by slashing R&D and optimizing costs".
Click to expand...

Instead of what it should be, which is, "Let's hire new talent and keep existing good talent in order to hopefully design something that is actually competitive with Intel".
 
Oh oh, this doesn't bode well at all. I hope they do spin off the Graphics division, that way it won't be tied to the boat anchor that their CPU division is turning into.
 
Well, one nearly celebrity type person staying or going, even one that did some interesting stuff, probably won't really make a huge difference either way. If that were the case, AMD having had this guy around for so many years would have been a more successful company or at least have better processors. The problems with AMD don't have anything to do with and engineer a buncha people adore. They're all bigger business-y issues that have been persisting for a long time and have a lot to do with the whole merger thing.
 
I don't read this like you guys do. I kinda think he was finished on this chip and he's moving on (or taking a long vacation :D) to design something else.

It's possible that he thinks AMD is going down the tubes, but I doubt AMD let him go to save money, because he almost certainly has some sort of deal where he gets paid, for a lengthy period, if they let him go. I've had friends in high positions that get paid for as much as 2 years if they're fired/laid off.
 
B-b-b-but he was going to "look at Zen" and make AMD #1 again. lol
 
This is totally lame. AMD's competitive area was CPUs. Not necessarily at the top end, but Intel can't compete with the fx-6300 and fx-8320 at the same price. I hope this isn't the death knell for AMD.
 
CreepyUncleGoogle said:
Well, one nearly celebrity type person staying or going, even one that did some interesting stuff, probably won't really make a huge difference either way. If that were the case, AMD having had this guy around for so many years would have been a more successful company or at least have better processors. The problems with AMD don't have anything to do with and engineer a buncha people adore. They're all bigger business-y issues that have been persisting for a long time and have a lot to do with the whole merger thing.
Click to expand...

You need a captain with vision and skills to steer everyone else's creative yet different efforts.

One man can make a huge difference in creating a unified design goal and it getting done well.
 
It sounds like he left rather than got laid off, which could mean that he thinks Zen is a lead balloon, or AMD is doomed. Even if Zen is a huge improvement it really sounds like it's going to be way, way too late and too little to really be interesting.
 
*Sigh* :( Oh well, looks like Zen is dead and with it, AMD. Now I am definitely going to upgrade path of a 5820k but I will stick with the R9 290 for a while longer. The CEO and board do not have a clue what they are doing. Either that or they have a clue which means they intentions from the beginning was to get what they could well bringing AMD to an end.

No, there will be no more competition in the area that AMD leaves behind. Just glad that the AMD chips I have are nice and fast as well as the video card I have. However, there is nothing to look forward to from them anymore.
 
Just to possibly quiet the naysayers. Jim has done this before, he usually joins a company for a major cpu architecture design and bails afterwards.

K7/K8, left

A4/5, left

Zen, left

I think he doesn't enjoy the incremental stuff.
 
I think it just means that Zen is pretty much done from a design standpoint and Keller would rather do something different with his time than stick around for the tedious and boring completion process.

It is a little strange to basically abandon the chip before it hits the shelves though. There may be some bad blood with the inner politics at AMD and Keller, and he basically stuck it out it through til the design was finished then part ways again (either by his own accord or some pact).

Or it could be that was the deal to begin with when he came back to get Zen going in the first place. All we can do is speculate until Zen is on the shelves or some new information comes out.

I will say this, if AMD remains tight-lipped about Zen close to release as it was with Bulldozer then that's probably a really bad sign.
 
Mungojerrie said:
Just to possibly quiet the naysayers. Jim has done this before, he usually joins a company for a major cpu architecture design and bails afterwards.

K7/K8, left

A4/5, left

Zen, left

I think he doesn't enjoy the incremental stuff.
Click to expand...

Hey, I am an AMD fan and I think the writing on the wall is there for all to see now. You do not see the major architecture engineers leaving Intel, do you? Also, Zen out not until 2017 will be to late for me.
 
I think it's more telling that someone who is labeled a "legendary CPU architect", leaves companies after his "job" there is done instead of staying wtih a company to make more innovative and advanced products.

But based on history of jumping companies seemingly whenever he wants... maybe he just wants to do something else and we're reading too much into it. No doubt AMD has had a metric fuckton of bad press recently, I can't imagine it feels good to work under that either.
 
AMD is kind of a company that has a fascinating strategy. They're in the competitive tech business -- which necessarily relies on fresh and new products to generate sales -- and their strategy seems to be to not really release anything new. It would be like if Microsoft sold Windows 95 for ten years.
 
DigitalGriffin said:
You need a captain with vision and skills to steer everyone else's creative yet different efforts.

One man can make a huge difference in creating a unified design goal and it getting done well.
Click to expand...

True, but no one is completely un-replaceable. Someone else who's name we don't know could do the whole vision-y, steer-y, create-y junk just as well and we'd never know that.

I think it's like a million and three percent more smarter to wait to see what whatever products AMD tries to sell do before getting out a whole buncha shovels to dig the company a grave...or at least armchair claim the death of the company over the loss of one person. People tend to get like waaay to inflatuated over someone they go all "Le sigh, he/she's a celeb!" about and never bother to acknowledge that if stuff like that was true, companies that are currently super-duper successful that had their founder go away and get replaced wouldn't be super-duper successful and they'd all have gone out of business....human psychology is funny and dumbness like that sometimes.
 
Maybe he's just doing contract work? No need to stay around when the work is done, and you get to set your own price. Besides that, being as renowned as he is, he wouldn't have any trouble finding work.
 
Could this mean that Zen is actually done and he doesn't need to be there now?
 
When engineers are stepping down instead of an inept CEO, there's a problem.
 
If Jim Keller is as good as everyone says (certainly seems so) he can work in whatever he wants, wherever he wants. I'd say the guy finished Zen, and wants to do something else now :)
 
Doward said:
If Jim Keller is as good as everyone says (certainly seems so) he can work in whatever he wants, wherever he wants. I'd say the guy finished Zen, and wants to do something else now :)
Click to expand...

With modern CPU design you create the CPU circuits by using simple code. (pipeline stages, prefetch, # steps look-ahead for OOE, predictive branch algorithm, cache storage, etc...)The program then takes this code and transforms it into circuits. These circuits are then hand tweeked and placed to due with propagation, path, and noise issues. They then go to tape out, and then testing, and then more tweeking.

At this point for an early 2017 release, (and AMD has been missing targets) I would say the simple code logic is done. But that is a long cry from silicon.
 
BloodyIron said:
This is totally lame. AMD's competitive area was CPUs. Not necessarily at the top end, but Intel can't compete with the fx-6300 and fx-8320 at the same price. I hope this isn't the death knell for AMD.
Click to expand...

Uhh. yeah they can. It's called Celeron and i3.

Even the lowest desktop i3 is faster per core than the fastest FX CPU. Sure the FX chisp have more cores, and that helps in SOME tasks, but I'd still take fewer faster cores at the same price.
 
Mungojerrie said:
Just to possibly quiet the naysayers. Jim has done this before, he usually joins a company for a major cpu architecture design and bails afterwards.

K7/K8, left

A4/5, left

Zen, left

I think he doesn't enjoy the incremental stuff.
Click to expand...

If that exact pattern has really occurred, then I think you're right. I know a lot of hardware designers, (in many different electronic fields, even do a little myself) and this actually seems indicative of this type of person. Yes, that's a huge generalization and a bit stereotype(y) but I think if you're the type of person who can come up with new designs like this, you also look for new challenges, become a little restless and start feeling stagnate if you stick around doing one thing for too long.

I don't have any better a guess than anyone else on his reasoning (or AMD's) on this, but I have seen it with people who do this kind of work somewhat frequently.

I'm the same way to a degree. I would always prefer to be working on a new project than keep pushing at the same one once I'm happy with it. It just so happens that one of my most popular PCBs is one that I'm absolutely sick of, and wish people would stop buying in favor of some of my newer/cooler ones.
 
BloodyIron said:
This is totally lame. AMD's competitive area was CPUs. Not necessarily at the top end, but Intel can't compete with the fx-6300 and fx-8320 at the same price. I hope this isn't the death knell for AMD.
Click to expand...

Intel could easily lower prices if they really wanted the low margin market that AMD is stuck with.
 
DigitalGriffin said:
At this point for an early 2017 release, (and AMD has been missing targets) I would say the simple code logic is done. But that is a long cry from silicon.
Click to expand...

I thought they had moved Zen up in the queue in front of K12, targeting an October 2016 launch?
 
jwcalla said:
AMD is kind of a company that has a fascinating strategy. They're in the competitive tech business -- which necessarily relies on fresh and new products to generate sales -- and their strategy seems to be to not really release anything new. It would be like if Microsoft sold Windows 95 for ten years.
Click to expand...

Well, they did sell XP for about that long :p
 
"Looking ahead, Mark Papermaster, AMD's Chief Technology Officer, will step in as acting leader for Keller's team and will now help oversee one of the most important products in the company's history."

No pressure Mark. If this slips or doesn't live up to expectations, AMD is essentially done for.

Zen is a make or break project

Go execute!
 
dr.kevin said:
So with Zen, AMD will just be 5 years behind Intel ?
Still a shitty situation.
Click to expand...

True,

but it's not as if Intel has made any amazing strides in the last 5 years either.

If they can catch up with where Intel was 5 years ago, they will be immensely more competitive than they are today.
 
October 2016 is only just over a year away.

My 3930k will probably last that long.

If AMD can meet the following requirements, I'll probably buy one just on principle to support them.
  • Must be offered in 6+ core model.
  • Per core performance doesn't have to catch up with Intel performance-wise, but has to at least be faster than my 3930k, both stock-to-stock and overclocked-to-overclocked.
  • 40+ PCIe lanes.
 
More of the AMD corpse twitching before rigamortis sets in....

Lead of ZEN cuts bait and runs what could that possibly mean?
 
The price is there, but those CPUs aren't even close performance wise to the CPUs I mentioned. Check performance reviews and benchmarks. The proof is in the pudding.



Zarathustra[H];1041862848 said:
Uhh. yeah they can. It's called Celeron and i3.

Even the lowest desktop i3 is faster per core than the fastest FX CPU. Sure the FX chisp have more cores, and that helps in SOME tasks, but I'd still take fewer faster cores at the same price.
Click to expand...
 
So fucking what? That means nothing. That's not reality, that's just unfounded speculation. I can't in good faith recommend Intel processors comparable to the ones I stated as they are consistently $150-$200+ more expensive.

Why pay more and get the same?

I'm just talking about these particular CPUs.



aaronmac3232 said:
Intel could easily lower prices if they really wanted the low margin market that AMD is stuck with.
Click to expand...
 
The price is there, but those CPUs aren't even close performance wise to the CPUs I mentioned. Check performance reviews and benchmarks. The proof is in the pudding
Click to expand...

They are faster than the CPUs you mention in performance. At least that is what the reviews and benchmarks say..
 
You must log in or register to reply here.
Back
Top