And the Bulldozer die size is……

[halofanman]

How big is AMD’s Bulldozer/Orochi core? That one is easy, 315mm^2 according to AMD’s slide deck that fell in to SemiAccurate’s hands last week.

Source

[Origin_Unknown]

I'm not quite sure why but that looks odd

[kuroikenshi]

Nice!!!!!

[Mats]

How come there's so much empty space between the L3's on the left side? That is, IF it really is empty.

[-Boris-]

How come there's so much empty space between the L3's on the left side? That is, IF it really is empty.

What should they use the space for? They can't make the chip much smaller since the northbridge uses space on the right side. Might put a fifth core in that space in future revisions, but that would make a similiar sized empty space on the right side.

[zanzabar]

I'm not quite sure why but that looks odd

it is set up like a gpu using unified shaders physically, and each pair cores shares a corner and l2 cashe

[Mats]

What should they use the space for? They can't make the chip much smaller since the northbridge uses space on the right side.

Don't you agree that there's unusually much empty space? How about rearranging cache and NB, and make it smaller?
I guess the cache have to be somewhat symmetrical, I just thought they could do better than that.

[Tenknics]

315mm^2? ouchie..

4core sandyb with igp is 216 mm^2. 6 core westies are 239 mm^2 die...

not good for amd.

[flyck]

315mm^2? ouchie..

4core sandyb with igp is 216 mm^2. 6 core westies are 239 mm^2 die...

not good for amd.

Doesn't matter. Their current lineup is even bigger and not competitive at all.

They probably saved R&D and drawing costs to keep it like this... and those saved costs probably are alot more profitable then the volume in which BD will ship.

[kuroikenshi]

Don't you agree that there's unusually much empty space? How about rearranging cache and NB, and make it smaller?
I guess the cache have to be somewhat symmetrical, I just thought they could do better than that.

the cores require that space on the die to play pong

[v0dka]

315mm^2? ouchie..

4core sandyb with igp is 216 mm^2. 6 core westies are 239 mm^2 die...

not good for amd.

Apples and oranges. 4 BD modules are not the same as 4 SB cores. You could say that this BD die is an 8 core design.

[bamtan2]

How come there's so much empty space between the L3's on the left side? That is, IF it really is empty.

there is no way to know what is there. images like this are not accurate. think they give away everything in an image?

[Hornet331]

Apples and oranges. 4 BD modules are not the same as 4 SB cores. You could say that this BD die is an 8 core design.

In the end thats also pretty much irrelevant and its performance what counts

[AliG]

In the end thats also pretty much irrelevant and its performance what counts

Agreed, and even still I think Bulldozer will only be competitive in servers and low power machines (HTPC and laptops) from what AMD has suggested so far. It seems that they are simply conceding the single threaded performance crown to Intel in favor of being able to put many smaller, higher clocked integer cores on a single die. Which is fine by me if it brings them back to consistent profitability (which is good for everyone), but I for one got tired of waiting and just got an i5 2500k. At best we're looking Bulldozer being 10% slower than Sandybridge clock:clock in terms of ipc, their biggest advantage over Intel right now seems to be llano (which simply just isn't meant to be a high end component).

I had been waiting since 2008 to do a legitimate new AMD ugprade and frankly when my am2 cpu+mobo died and I saw how cheap it was to build a top notch Intel system I really couldn't wait any longer

[Pantsu]

I wonder how much an 8-core SB would take, without IGP. Then again BD isn't full 8-core, so it's hard to compare the die sizes. In any case performance is what matters for the customer. If AMD needs a 315 mm^2 to compete with a 216 mm^2 chip including graphics portion it doesn't bode well.

[AliG]

I wonder how much an 8-core SB would take, without IGP. Then again BD isn't full 8-core, so it's hard to compare the die sizes. In any case performance is what matters for the customer. If AMD needs a 315 mm^2 to compete with a 216 mm^2 chip including graphics portion it doesn't bode well.

Yeah I'm having a tough time seeing how they can afford to compete in price:performance if the product they're riding their company on is 150% larger and likely 5-10% slower across the board

[Hornet331]

I wonder how much an 8-core SB would take, without IGP. Then again BD isn't full 8-core, so it's hard to compare the die sizes. In any case performance is what matters for the customer. If AMD needs a 315 mm^2 to compete with a 216 mm^2 chip including graphics portion it doesn't bode well.

Well you see how this turns out /financial wise) in the margins.

[Lanek]

In the end thats also pretty much irrelevant and its performance what counts

+ the TDP.. if really they are 95-125W .. it's ok.

[AliG]

+ the TDP.. if really they are 95-125W .. it's ok.

How does that compare to a 95-125W Intel processor (legitimate question)? I believe they use different methods calculating TDP

[flyck]

Yeah I'm having a tough time seeing how they can afford to compete in price:performance if the product they're riding their company on is 150% larger and likely 5-10% slower across the board

You think the cpu will perform 5-10% less then its competitor based on nothing? if the FX8 8150 performs 5-10% slower then the FX6 and FX4 have no meaning to exist. since they will be slower then their current AM3+ lineup.

and you say the die size is 150% larger or = 550mm2.... the die size seems to be ~45% bigger.

Prize isn't determined by die size. Prize is determined by 1001 factors and only one of them is die size.

Depending on the volume, die size might even be completely irrelvant given the product...

If i have a volume of 1000 chips which all need 1 wafer but the R&D took 5 years with a team of 100 people, the die sizecost is completely overhauled by the other costs.

Same goes with BD. BD has a relativly short live in his current form, has low volume but broad market space (client -> server). If they need to make 8 different maks to optimize die spaces, the cpu price will be determined by the masks more then it will by the die size.
This is ofcourse different for intel because they have such huge volumes that the die size becomes a real factor in the total cpu costs.

For trinity (replacer of llano) die size will be alot more important for the cost. (high volume, big market space)

[Mats]

there is no way to know what is there. images like this are not accurate. think they give away everything in an image?

Of course not, but that can be said about every CPU die.
I don't think I've seen so much empty space in a long time, I mean, space that seems to be empty.

[Lanek]

How does that compare to a 95-125W Intel processor (legitimate question)? I believe they use different methods calculating TDP

good question, but i m sure you can find an article on xbitlab or Anandtech who expose thoses difference.

[M.Beier]

AMD is pulling a FERMI....
Talk and talk and no product untill way too late... I remember how we were discussing if it would arrive before Gulftown on this forum, and many others, now we are wondering if it will arrive before or after SNB-E....

[AliG]

You think the cpu will perform 5-10% less then its competitor based on nothing? if the FX8 8150 performs 5-10% slower then the FX6 and FX4 have no meaning to exist. since they will be slower then their current AM3+ lineup.

and you say the die size is 150% larger or = 550mm2.... the die size seems to be ~45% bigger.

Prize isn't determined by die size. Prize is determined by 1001 factors and only one of them is die size.

Sigh, this is why the news threads get locked up so quickly. Seriously people need to stop getting so upset, I really don't understand all the fanboyism. Sure its cool to have your brand of preference, but to get insulted as if you are an employee? That's just too far.

Typo on the size, I meant to say 150% of the size, which means 50% larger. As for performance, no I'm not blindly estimating this. In case you haven't noticed I have a few thousand posts on this forum, many of which have been contributing towards discussions regarding up coming cpu performance. It's been predicted that Bulldozer will excel at multithreaded performance, but still be a decent amount behind sandybridge in single threaded performance - and there a lot of reasons for how we have reached this conclusion.

[Manicdan]

that empty space has been known for almost a year now, and sofar the only thing i herd about its purpose is that it helps with higher clocks or something.
(<-not an expert)