Bulldozers Weak/Strong points?

[jimbo75]

That said...the big unknown here is GloFo's 32nm process tech. That determines clockspeeds, yields, and power-consumption. So we have no way of knowing whether GloFo is going to do good by AMD or if they are going to be the achilles heel in the equation.

Unfortunately we do know GloFo went gate-first integration for HKMG which is known (as in "science" known, process technology and all that) to enable denser chips (same design by smaller die size) but at the expense of having lower drive currents (less clockspeed potential) compared to a gate-last integration.

I've yet to see anything from GF stating that they have lower drive current than intel at 32nm.

 

[Idontcare]

I've yet to see anything from GF stating that they have lower drive current than intel at 32nm.

Seriously? Every one in the entire freaking industry trails Intel in drive currents at every node, its been that way for a decade if not longer.

Given that Intel's 32nm driver currents intrinsically benefit from gate-last integration, even Intel would be hard-pressed to craft a gate-first integration flow that would best their own gate-last enabled drive currents at 32nm.

It is possible that GloFo did the seemingly impossible and matched Intel's drive currents despite the unavoidable handicap that comes with gate-first integration, but the published science and engineering to date most certainly suggests it is not probable.

 

[podspi]

Hopefully this will be somewhat mitigated by the design. Isn't GF (and the rest of the alliance) going gate-last next time around?

 

[jimbo75]

Seriously? Every one in the entire freaking industry trails Intel in drive currents at every node, its been that way for a decade if not longer.

Given that Intel's 32nm driver currents intrinsically benefit from gate-last integration, even Intel would be hard-pressed to craft a gate-first integration flow that would best their own gate-last enabled drive currents at 32nm.

It is possible that GloFo did the seemingly impossible and matched Intel's drive currents despite the unavoidable handicap that comes with gate-first integration, but the published science and engineering to date most certainly suggests it is not probable.

http://www.globalfoundries.com/eBooks/dac_hkmg/

I also remember reading over a year ago that GF matched intel at 32nm. Obviously the yield situation wasn't so good.

 

[LiuKangBakinPie]

Plus points
SSE4.1 and SSE4.2
.
AVX (Advanced Vector Extensions) with two additional subsets, called XOP and FMA4
.
AES (Advanced Encryption Standard)
.
LWP (Light Weight Profiling)

Weak points
A dual core model ain't really a dual core and a quad ain't really a quad core.

 

[gdansk]

Plus points
A dual core model ain't really a dual core and a quad ain't really a quad core.

How not? It'll have 4 (2x2) FPUs and 4 integer cores. Two modules, for all intensive purposes, is a quad core.

 

[Meghan54]

Two modules, for all intensive purposes, is a quad core.

That's Intents and Purposes...not intensive purposes. Derp....

 

[podspi]

^ Actually, I think it works both ways ...

Hopefully GF's 32nm is good, since it's going up against (at least for AMD) Intel's 22nm...

 

[gdansk]

That's Intents and Purposes...not intensive purposes. Derp....

Ah well, you knew exactly what I meant. Of course, I didn't, but you did. The phrase being a common "eggcorn", I hope you accept my apologies for desecrating this long established English idiom.

 

[Riek]

Plus points
SSE4.1 and SSE4.2
.
AVX (Advanced Vector Extensions) with two additional subsets, called XOP and FMA4
.
AES (Advanced Encryption Standard)
.
LWP (Light Weight Profiling)

Weak points
A dual core model ain't really a dual core and a quad ain't really a quad core.

You can also say a module is close to a dual core without the need to have complete duplicated cores. Or having close to the performance of 2cores while using the surface slightly above a single core. e.g. you can say it as an advantage and not a disadvantage.

Possibly a big drawback of BD will be the turboboost. They will turboboost against TDP. So if the running applications are using less resources BD will clock higher and perform better.... but this means that BD will be relativly closer against its TDP compared to its competitor or predecesors. So it will be using alot more power on average with the same TDP as other cpu's. (intel on the other hand has a similar problem where the cpu's can go over TDP for a short periods)

 

[Idontcare]

Ah well, you knew exactly what I meant. Of course, I didn't, but you did. The phrase being a common "eggcorn", I hope you accept my apologies for desecrating this long established English idiom.

If this were VC&G this would have been my response to the post you were addressing:
http://forums.anandtech.com/showpost.php?p=31517838&postcount=57

edit: making it more clear as to who I was directing my post towards.

 

[mosox]

Most of all no DRM. If only for this I will buy AMD insted of Intel.

 

[IntelUser2000]

http://www.globalfoundries.com/eBooks/dac_hkmg/

I also remember reading over a year ago that GF matched intel at 32nm. Obviously the yield situation wasn't so good.

Yield is important to performance too. Reaching high performance with 1 transistor isn't important as reaching that same performance with a circuit block, and far less important than reaching that using a full microprocessor die.

Also I think the comparison they were doing was with an older version of Intel 32nm process which was shown before products were ready. So the differences are even greater.

 

[jimbo75]

That brochure first appeared in July last year so it was compared to intels 7 month old process.

You are right in saying that it could be worse (or intel could have improved theirs), but it could also have got better since then.

 

[Phynaz]

I won't challenge your recollection of history as I freely admit we are pushing up against the hairy edge of the limits of my trustable memory (phenom was what, 2007? 2008?) but the mental "note to self" I have filed away in the back of my mind was that the "hype" by AMD was limited to that of Barcelona and the server aspects of the chip.

IIRC, which is a long-shot in this case, Phenom the desktop consumer processor was not hyped by AMD, they were quite mum about it.

You forgot about the 3GHz Phenom demo in the case made out of fans.

http://www.anandtech.com/show/2287/10
http://forums.anandtech.com/showthread.php?t=66823

 

[Idontcare]

http://www.anandtech.com/show/2287/10

I read this in that link, so prophetic: (queue TLB bug)

In a demonstration designed to prove that Phenom isn't broken, AMD featured a quad-core Phenom X4 processor, with standard cooling, running at 3.0GHz.

But they did cover their ass with this:

While Phenom won't be anywhere near that clock speed when it launches at the end of this year, AMD expects to be at 3GHz within the first half of 2008.

 

[Idontcare]

http://forums.anandtech.com/showthread.php?t=66823

^ ZOMG we sure were a forum full of a bunch of hateful trollish asshats back then! D: (myself included ) Yikes! Not pretty, please don't link to threads that go back that far in the past ever again, there should be a statute of limitations on forum thread links. :|

 

[exar333]

^ ZOMG we sure were a forum full of a bunch of hateful trollish asshats back then! D: (myself included ) Yikes! Not pretty, please don't link to threads that go back that far in the past ever again, there should be a statute of limitations on forum thread links. :|

Man were you a jerk IDC! J/K :awe:

Anyway, it's funny that BD vs. Nehalem was mentioned in that thread 4 years ago! Wow! Crazy how time flies and we are essentially speaking to the same thing now, but more along the lines of BD vs. SB.

 

[formulav8]

If you are incapable of figuring out why this is the case, or like to assume you do know but no one else can possibly be so clever as to figure it out as well, then you need to give up now for your life will be forever a torment as you continue to perceive yourself to be surrounded by ignorant masses from now till death.

I know this is off topic but I couldn't help myself :sneaky: :biggrin: A mean one you were.

And you were right it was specifically Barcy that they were saying is 40% faster and such.

 

[Idontcare]

^ Pretty sure that was me on a nice day back then too. Mr friendly, that's what they use to call me then

 

[Abwx]

I know this is off topic but I couldn't help myself :sneaky: :biggrin: A mean one you were.

And you were right it was specifically Barcy that they were saying is 40% faster and such.

In Fluent, according to an Intel s slide , a 2.2ghz Barcelona
roughly matched a 3.0ghz Xeon 5472...
Why it didn t do well overall is another question to wich
only AMD could give, retrospectively , an accurate response....

 

[Phynaz]

In Fluent, according to an Intel s slide , a 2.2ghz Barcelona
roughly matched a 3.0ghz Xeon 5472...

Somebody lied to you.

 

[Abwx]

Somebody lied to you.

No, it was pointed by anand , and i checked
the slide on intel s site, and that was real...

Indeed, it was a 2ghz barcelona..

From anand " Intel s own marketing seems to
admit that a xeon E5472 with 800mhz memory
is just as fast as a AMD s quad core at 2ghz"
(edit : in Fluent).
It dId very well also on LS-DYNA....
Check by yourself...
http://www.anandtech.com/show/2386/11

https://h30406.www3.hp.com/campaigns/2008/events/caesymposium/images/ANSYS-CFD.pdf

Sure that it s somewhat annoying in respect
of your motto about Randy Allen....

 

[OCGuy]

In Fluent, according to an Intel s slide , a 2.2ghz Barcelona
roughly matched a 3.0ghz Xeon 5472...
Why it didn t do well overall is another question to wich
only AMD could give, retrospectively , an accurate response....

What flavor is AMD kool-aid?

 

[Abwx]

What flavor is AMD kool-aid?

Trolling as a raison d' etre ?....
Or pehaps it s tiring to think a little..?..