AMD Quadcore sample appeared in Akiba!!!

[G.Foyle]

Say whatever you want, be whoever's fanboy, but you still can't deny it's a quad core chip IN SINGLE DIE. Intel doesn't have that (yet)

And to clarify I'm a fanboy of whoever has better CPU's (in terms of bang per buck ratio).

[MaxxxRacer]

thats one huge die. looks like a 8800GTX die.

[FghtinIrshNvrDi]

wow, I'm a little turned on.

Ryan

[itznfb]

that'll crunch some WUs

[freeloader]

that'll crunch some WUs

Do you think faster with SSE or straight FPU? Considering how much AMD has enhanced the FPU in the new core(s).

[ludeboy12]

damn cant wait to see some ES benches now.....

[informal]

Do you think faster with SSE or straight FPU? Considering how much AMD has enhanced the FPU in the new core(s).

SSE is vastly improved too! Doubled throughput compared to K8 core.

[largon]

thats one huge die. looks like a 8800GTX die.

It's big but not that big G80;
G80: 420mm²
Barcelona: 295-300mm² (closer to 300mm²)

[Sniper.nkc]

PS. The die looks kinda scratched, I guess to many people were fideling with the sample

It could be an "Engineering version." If it is, them engineers have some big nails on their fingers

[oijkue]

wow....it looks so sexy

[nn_step]

wow....it looks so sexy

somehow your sig and that phrase should not be shown together because some noob will probably think about something other than Quadcore if you know what I mean

[Nedjo]

It's big but not that big G80;
G80: 420mm²
Barcelona: 295-300mm² (closer to 300mm²)

G80 is 484 mm² (NVIO not included)!
Where did you get the number of ~300mm²?
No one is really sure about 90nm Windsor, and 65nm Brisbane die size.

Acording to Anand Windsor is 183mm².
According to Techreport Windsor is 220mm²
They agree about Brisbane's 126 mm².

The simple math says that Barcelona is 126x2+??mm² So, I guess 300 mm² is the minimum if you consider that H rev. core+L2 is bigger (probably >130mm²) then G rev., and L3 is quite "burden" on size!

Anyhow, 45nm is THE MUST for AMD, and it's ~18 months away.

[[XC] MarioMaster]

the old AthlonXP's had shiny dies like that

[dinos22]

yippie does it work lol hehe

[ownage]

They dont have to rush if you ask me, it only will look my e6400 old.
Rush isn't a good thing, look at nVidia.
I really hope K8L will be damn good, how more Good action, how better.

[pcfreak]

booking ticket for Barcelona

[largon]

Nedjo,
My bad, you'r correct about G80 - it is 484mm². Bad google search.

I got the ~300mm² value for Barcelona by comparing the available pictures of the bare die compared to the package dimensions.
It's definately 290-300mm².

The result should be valid as I just calculated the G80 die using 2 different low res pics of IHSless G80; the final value was "only" 1% off reality.

[zir_blazer]

Acording to Anand Windsor is 183mm².
According to Techreport Windsor is 220mm²
They agree about Brisbane's 126 mm².

Remember that there are two Cores called Windsors that are physically different, the BH-F2, that is a Dual Core with 512 KB Cache L2 per Core and the JH-F2, that is Dual Core with 1 MB Cache L2 per Core. Both could be correct.

[frankR]

OOOoooooooooo. A picture of a guy holding a CPU.

AMD doesn't show us anything, because they have nothing.

[Omastar]

I wonder what kinda heat this thing puts out on air. Not nearly as much as Intel's Q6xxx line I'm sure due to its monolithic nature.

[red]

http://www.theinquirer.net/default.aspx?article=37270

AMD measures TDP as maximum power draw for a CPU, Intel as maximum real-world draw. AMD is going to move to the Intel measuring scheme

Interesting. That means that tomorrow's AMD TDP of 125W is going to be more hot than today's 125W.

[kyosen]

Thank you largon and everyone who trusted me

These pictures was taken at Windows Vista event in Akiba lastnight.
Anyone who preordered Vista at a certain PC shop could take part in the event.
Surprisingly for me, any NDA was not required.

The black cap guy is a representative of Nihon AMD, Mr. Doi.
He didn't mention in details, but two notch of CPUon each side implies
that it's a sample chip for Socket F(or its successor)...Barcelona, apparently.

He said that AMD makes efforts for launching within 2007-1H.
To my question, "Possiblity in April-end?", He said "April?...too early".
Anyway I hope K8L will be shining on AMD and us within 2007-2Q

[arisythila]

You don't have to mark it that hard that youre a fanboy

HAHAHAHA.......

[Charles Wirth]

Thanks Kyosen-san,

Did they have dual quad demo's setup?

[lutjens]

I'd bet 10 bucks the chip is simply a deactivated example of the K8L. Built and then disabled by AMD for use at events like that one as a physical representation only. This way the chip could go to events like this one and there would be no risk of someone getting a hold of it and spilling the beans on how it performs...