Slides surface on Bulldozer innards

[red1776]

More on Bulldozer Architecture here from WCCF Tech....



http://wccftech.com/amd-slides-detail-bulldozer-zambezifx-am3-platform-architecture/

 

[Route44]

G, in the past 8 months we have read a ton of rumor with a ton of supposed genuine slides only to find out so much is bogus. X-Bit Labs website in particular comes to mind

And you understand this a lot better than most people I know. So this NOT directed at you. I just like to know what you think and how reliable is this website?

 

[dividebyzero]

@Route44
If you check out the post I put up yesterday...you'll see both my level of absolute trust in Chuckie D's articles, but more importantly, the 315mm² die size meshes pretty damn close to what was posited a couple of years back (see links in post #23 -posted January. so Chucky's "exclusive" isn't quite as earth-shattering as I'm sure he's portraying it)

As every man+ dog expected, the die is pretty big (considering a 2500K/2600K measures 216mm² and a Gulftown hex core at 239mm²), but it's still a step in the right direction considering the dinner-plate sized Thuban.

 

[red1776]

G, in the past 8 months we have read a ton of rumor with a ton of supposed genuine slides only to find out so much is bogus. X-Bit Labs website in particular comes to mind

And you understand this a lot better than most people I know. So this NOT directed at you. I just like to know what you think and how reliable is this website?

i am in line with Chefs thinking on this. Some of it is on, some not so much. Charlie is either spot on...or on a different planet...and no way to tell until after its over. I have a different idea on the large die size though. I think that it is part of AMD's CPU strategy of going modular. As far as what the performance will be, I am watching for the prices. I think that will tell us exactly where they will be against Intel.

As far as WCCF reliability goes, they like to get the bleeding edge jump on stories/rumors, so they are hit or miss. I do think they have good intentions (unlike some others) but they do get the wool pulled over occasionally...okay a lot.

 

[Route44]

Okay guys, thanks. Again both of you know a ton more about cpu architecture than I do so Z (thanks for the link; by the way, nice ocing on those Thubans ) and red what is the bottom line concerning possible performance from this article? I know until we have real benches and user experiences it is all conjecture, but I'll take conjecture from you two any day.

 

[red1776]

Okay guys, thanks. Again both of you know a ton more about cpu architecture than I do so Z (thanks for the link; by the way, nice ocing on those Thubans ) and red what is the bottom line concerning possible performance from this article? I know until we have real benches and user experiences it is all conjecture, but I'll take conjecture from you two any day.

Arg...I can only speculate that AMD has found the answers to making their version of 'netburst' successful. It was a disaster when Intel tried it with the P68 'Willamette' in 2000. They are claiming that they have solved the branch prediction challenge. The higher clocks would seem to be evidence of this as well. If they have indeed figured it out, multi threaded performance might be phenomenal.