Ex-AMD Engineer explains Bulldozer fiasco

[Idontcare]

I think Dmens may be talking about the multi-year delay, not the last few months.

If you mean the 45nm bulldozer project that mostly bears no resemblance to todays bulldozer save for a recycling of the internal code name and the same design team, then that 45nm bulldozer was scrapped because it was going to be even more akin to "copious amounts of tiny weak cores" and internal projections showed it was going to unacceptably suck with conventional workloads expected at the time of launch.

That had nothing to do with 45nm or the design tools, the architecture itself was deemed untenable given the projected software landscape.

(I think I am remembering that correctly, its been so long ago now)

 

[dmens]

If you mean the 45nm bulldozer project that mostly bears no resemblance to todays bulldozer save for a recycling of the internal code name and the same design team, then that 45nm bulldozer was scrapped because it was going to be even more akin to "copious amounts of tiny weak cores" and internal projections showed it was going to unacceptably suck with conventional workloads expected at the time of launch.

That had nothing to do with 45nm or the design tools, the architecture itself was deemed untenable given the projected software landscape.

(I think I am remembering that correctly, its been so long ago now)

Hey, I understand the ranting and raving about design tools and methodologies, I do it all the time.

But if the architecture itself is unable to compete in single thread performance, then AMD management f*ed up big time. Even Banias/Dothan/Yonah was showing signs of what it could do in single thread and that was well before BD was in actual development.

And if there was a design restart with the 32nm process, they should have normalized single thread perf projections to the C2D line, but it doesn't look like they did.

 

[Deanodarlo]

If they weren't hand crafting, what were they doing for the 4 long years of development?

I guess the fact is Intel do have a massive amount more in terms of resources for this kind of optimization, however Bulldozer does give the impression that it's not running at its potential and is way too large for the performance it offers.

 

[Tikerz]

I'm talking about this >>> Ex AMD designer: Bulldozer to disappoint at XtremeSystems. This was last year.

If everyone would have just accepted what he said over a year ago, the number of strokes people are giving themselves in all of this BD discussion would be greatly reduced.

 

[EightySix Four]

"hand-crafting various performance-critical parts of its chips and rely completely on automatic tools. "

This sounds like a disgruntled former employee who was replaced by a robot.

Except he was right...

 

[Magic Carpet]

It is quite obvious, the people behind the original FX processor, are no longer with AMD. I blame the leadership though. It must go, simple as.

 

[pm]

Thanks for the post.

I don't completely buy the engineer claim. With the amount we know, we might as well even claim he's saying that to get the spotlight on himself. Admit it, everyone wants it.

Tech is extremely complicated nowadays. So many limitations, and engineers approach from ALL angles to solve the problems. Not like 1990s where you get one "macro" feature and the performance skyrocketed. It's always easy to fault it on one thing.

Maybe the engineer worked on the part he complains about.

I know CTho personally. He is a very competent engineer and is extremely capable. I dislike the insinuations you are making about someone you have never met and presumably know very little about. If CTho said it, then I'm sure he believes it based on data that he's collected and his experience.


Patrick Mahoney
Senior Design Engineer
Intel Corp.

* Not speaking for Intel Corp. *

 

[Stoneburner]

I know CTho personally. He is a very competent engineer and is extremely capable. I dislike the insinuations you are making about someone you have never met and presumably know very little about. If CTho said it, then I'm sure he believes it based on data that he's collected and his experience.


Patrick Mahoney
Senior Design Engineer
Intel Corp.

* Not speaking for Intel Corp. *

Intel Design Engineer? God you guys must be having a great time.

 

[Idontcare]

I know CTho personally. He is a very competent engineer and is extremely capable. I dislike the insinuations you are making about someone you have never met and presumably know very little about. If CTho said it, then I'm sure he believes it based on data that he's collected and his experience.

pm, I think inteluser was referring to the engineer that had left amd years ago. Yeah he quoted someone mentioning Ctho9305 but I don't think that was who he was directing his comments towards (otherwise I'd agree with you, Ctho knows what he's talking about)

 

[IntelUser2000]

pm, I think inteluser was referring to the engineer that had left amd years ago. Yeah he quoted someone mentioning Ctho9305 but I don't think that was who he was directing his comments towards (otherwise I'd agree with you, Ctho knows what he's talking about)

Yep, thanks.

You are right, Ctho knows what he's talking about.

BTW, if it was directed at Ctho, I would have said Ctho not "engineer".

 

[Idontcare]

Intel Design Engineer? God you guys must be having a great time.

If Intel is ran anything like TI, then actually no its not great.

When we thumped Moto in the late nineties in the race to own the mobile phone market, when the smoke cleared and we had succeeded, our reward internally was all our budgets were scaled back, projects cancelled, etc...management wanted to slow things down and soak up the profits for a while. As an engineer it was actually quite frustrating to be idled after getting use to the sprint for so long.

But maybe Intel is still in "only the paranoid survive" mode and the project budgets cup will continue to runneth over?

 

[Phynaz]

Intel Design Engineer? God you guys must be having a great time.

Notice the Fort Collins location?

It's former Intel Fort Collins people that are behind AMD's turbo core design if I remember correctly.

PM most likely has friends that are taking a lot of heat over BD.
(Pun not intended)

 

[IntelUser2000]

Notice the Fort Collins location?

It's former Intel Fort Collins people that are behind AMD's turbo core design if I remember correctly.

PM most likely has friends that are taking a lot of heat over BD.
(Pun not intended)

Damn, just damn. I'm sorry I tried to ever make a little joke, which IMO is being indirectly done ALL OVER the CPU thread. Honestly, I think the talks about Bulldozer sucking or whatnot became old after a day or so, but it rages on.

Sheesh.

 

[dma0991]

I'm talking about this >>> Ex AMD designer: Bulldozer to disappoint at XtremeSystems. This was last year.

The past has came back to haunt the future! Guess we should have listened to terrace215 when he said IPC was decreasing.

We're doing a 'postmortem' for Bulldozer. We already know it sucked now we want to know why it sucked.

 

[aigomorla]

But maybe Intel is still in "only the paranoid survive" mode and the project budgets cup will continue to runneth over?

no friends at intel tell me they always have projects running.

Only like 30% of what actually gets done do we see.

They do lots of experimentation with die material, i heard they are even playing with graphitine as well.

They build mock cpu's about the size of cd's which work and can do about 1 terraflops of calculations...

They try to get insane IO speeds with SSD's and other combinations of sorts...
They put active servers in cooking ovens to see how long it takes for the entire system to go POP.

They do lots of funny stuff according to my friend, stuff we dont really hear about.

 

[SunnyD]

Disgruntled (Ex-)Employees with axe to grind. Blog at 11.

This seems to be the new fad.

 

[soccerballtux]

If Intel is ran anything like TI, then actually no its not great.

When we thumped Moto in the late nineties in the race to own the mobile phone market, when the smoke cleared and we had succeeded, our reward internally was all our budgets were scaled back, projects cancelled, etc...management wanted to slow things down and soak up the profits for a while. As an engineer it was actually quite frustrating to be idled after getting use to the sprint for so long.

But maybe Intel is still in "only the paranoid survive" mode and the project budgets cup will continue to runneth over?

their shareholders will have something to say about that. Personally I hope Intel slows down a bit and gives AMD some breathing room. My Ph2 965 @4Ghz is plenty fast for the next 4 years

 

[soccerballtux]

no friends at intel tell me they always have projects running.

Only like 30% of what actually gets done do we see.

They do lots of experimentation with die material, i heard they are even playing with graphitine as well.

They build mock cpu's about the size of cd's which work and can do about 1 terraflops of calculations...

They try to get insane IO speeds with SSD's and other combinations of sorts...
They put active servers in cooking ovens to see how long it takes for the entire system to go POP.

They do lots of funny stuff according to my friend, stuff we dont really hear about.

lolwut
you mean graphene?

 

[aigomorla]

their shareholders will have something to say about that. Personally I hope Intel slows down a bit and gives AMD some breathing room. My Ph2 965 @4Ghz is plenty fast for the next 4 years

lulz...

personally i wont miss Sandy-E.

I want IVY-B and Haswell NOW! 22nm!! GOGOGO!!

lolwut
you mean graphene?

eyehaychoo?

lol they been playing with modified coal.. hows that?

 

[NostaSeronx]

315mm² x 80% => 252mm²
5.99 pts / 80% => 7.49 pts in Cinebench

If only if it were true, I would have reacted like:

Instead. I reacted like

 

[aigomorla]

Instead. I reacted like

But now were seeing a lot of these:

 

[quest55720]

Does anyone know if the bobcat was auto designed or hand tuned? Because if it was auto designed he is dead wrong and making excuses that it was not the engineers but the bean counters.

 

[alyarb]

I'm sure the engineer's math is a bit skewed toward the worst case. I can understand using synth tools as a point of departure and tightening up personally in the few areas where it matters to do so. My question is with Anton's math. 213m per module + 400m for L3 leaves 800 million (almost the size of thuban) transistors for memory and I/O. It can't be from the ballooning front-end because the 200 million per module incorporates that. It's the goddamn thuban-sized uncore that I want to understand, but mostly I'm with bryan.

Does anyone know if the bobcat was auto designed or hand tuned? Because if it was auto designed he is dead wrong and making excuses that it was not the engineers but the bean counters.

If you divide 2 billion transistors by 315 mm^2, you get about 6.35. If you divide 450 million transistors by 75, you get about 6. So Zacate is actually less dense than Zambezi (TSMC vs GloFo). This means if this "20% bigger" accusation turns out to be remotely true, then it will have some implications for Zacate and presumably Llano as well.

 

[bryanW1995]

I've heard this story before. I have nothing against x-bit passing this of as "news" since BD failure is a hot item now, but Cliff Maier mentioned this quite some time ago already.

So it's actually the other way around if I remember correctly. Cliff left before BD was anywhere near completion, but he said (in so many words) that BD will be a flop due to reliance on automated tools. 20% bigger, 20% slower.

Most people dismissed him as your average disgruntled employee, yadda yadda yadda.

Now that Bulldozer flopped epically, he now gets his much deserved vindication.

Cliff had a vested interest in stating that doing it "by hand" was more efficient because he would (and did) lose his job if the automated process was better. I don't know anything about AMD's culture, but if it was that cut and dried that doing it by hand was somehow empirically "better", wouldn't nvidia and intel be doing it that way as well? Or is maybe this 20% bigger, 20% slower statistic for such a small part of the cpu that everybody (amd included) is now better off having it automated? BD certainly sucks, and it's very easy to pile on right now, but this thing smacks a bit of sour grapes to me.

 

[woolfe9999]

I'm surprised no one mentioned this:


LOL, management. As if they know any better.

AMD went straight downhill after Jerry left. When you get people like Hector Ruiz (more like Hector Ruinz), I can't say I'm not surprised how they're turning out. In fact, it won't be news to me if they get bought soon.

http://www.forbes.com/sites/ericsavitz/2011/10/13/amd-now-verging-on-irrelvancy-analyst-says/

On the topic of AMD being bought, I found it interesting, when looking over the thread containing Cliff Maier's comments about AMD's general suckitude, was Maier saying that if AMD gets bought out it loses its x86 license, which is non-transferrable. Unless I misunderstood what he was saying but I don't think I did.