AMD Prepares ‘Bristol Ridge’ APU (Carrizo) For Desktop

[KamiKatze]

2016, codename Zen, 14nm FinFET+, SMT, competes with intel, 8 core parts.

 

"8 cores" or 8 cores?

[LukaP]

"8 cores" or 8 cores?

SMT, so 8 true cores. atleast thats all that we know for now. I truly cant see them going CMT again, the market really isnt ready for it yet...

[Immort]

Sounds illogical to me, especially since 14nm, 8 core Zen CPUs are coming by 2016 Q3. Why even bother with excavator and 28nm if you'll have Zen ?

[Guest]

Maybe it IS Zen O_O

 

 

 

[patrickjp93]

Sounds illogical to me, especially since 14nm, 8 core Zen CPUs are coming by 2016 Q3. Why even bother with excavator and 28nm if you'll have Zen ?

Budget parts for office PCs and such.

[patrickjp93]

SMT, so 8 true cores. atleast thats all that we know for now. I truly cant see them going CMT again, the market really isnt ready for it yet...

I believe current word is they'll implement both SMT and CMT.

[LukaP]

I believe current word is they'll implement both SMT and CMT.

That is

1. impossible
2. counterintuitive

CMT and SMT are two completly different things... Its like saying they are making a pizza that will be both baked and unbaked at the same time.

[patrickjp93]

That is

• impossible
• counterintuitive

CMT and SMT are two completly different things... Its like saying they are making a pizza that will be both baked and unbaked at the same time.

@Opcode, would you like to do the honors?

1) not impossible

2) not really. Depending on workloads either one is advantageous over the other.

[LukaP]

@Opcode, would you like to do the honors?

1) not impossible

2) not really. Depending on workloads either one is advantageous over the other.

Yes, but you cant have both at the same time, in one core/module. thats my point. yes one cpu could have both implemented, or even have different cpus with different designs, but you cant have both at the same time

[Misanthrope]

Q3 2016? Is that right? A year and a half away? That's an awful long time to wait.

 

Aye, by then I expect intel to be as fast on igpu anyway

[patrickjp93]

Yes, but you cant have both at the same time, in one core/module. thats my point. yes one cpu could have both implemented, or even have different cpus with different designs, but you cant have both at the same time

One core can have both implemented. You just cannot use both simultaneously.

[patrickjp93]

Aye, by then I expect intel to be as fast on igpu anyway

It might already be with HD 6000 and Iris 6100. I'm eagerly awaiting benchmarks and performance reviews.

[HippY]

That is

1. impossible
2. counterintuitive

CMT and SMT are two completly different things... Its like saying they are making a pizza that will be both baked and unbaked at the same time.

You should see me baking or cooking

 

Impossible and counter-intuitive are 2 words that are meaningless when I bake or cook as I can create everything. Even half baked pizza. burnt pasta. 

 

But I can make beautiful tea, coffee!

[Nielips]

Don't know why people are bashing on it being 28 nm, as Intel have shown with devils canyon release. Die shrinkage doesn't play a dramatic difference at the sizes we are now at, unlike they once did, architecture seems to be a lot more important.

[Sharkyx1]

Alternatively it's AMD's chip for Nintendo's late 2015 console. If Keller is stupid enough to recycle Excavator for Zen, great improvement that it is, then no one will have faith in AMD any more.

Nintendo releasing a new console this year

not likely

[patrickjp93]

Nintendo releasing a new console this year

not likely

Have you seen the contract news between Nintendo and AMD? It sounds like Christmas 2015.

[Sharkyx1]

Have you seen the contract news between Nintendo and AMD? It sounds like Christmas 2015.

if you say so,

 

But i don't think this would help ninty much if its true, unless they can get some third party action and even then, i doubt people will believe its as powerful as its competitors

[SkilledRebuilds]

Woohoo DDR4 driven HTPC/Encoder/Decoder/LightGaming Box!!!

Would be nice to see if DDR4 APU's kept scaling with memory bandwidth or if they level out in performance before utilizing the extreme bandwidth available...

 

Reviews are one thing...but I'd be keen to test one, before buying one...

[vm'N]

It's every workload from compute to transactions to processing.

No, its power-efficiency comes when you actually have software that can leverage on the highly parallized processing power of the powerarchitecture.

At 5GHz the 12-core Power 8 flagship consumes only 80 watts and has double the performance of the IB flagship E7 Xeon in almost all of these areas.

Where do you get 80 watts from?

Also, Intel makes the same x86 architecture span every segment, and for all the hype about the custom ARM cores, I sorely doubt we're going to see such a massive performance jump as to threaten both Intel and Power 8/9 by the end of 2016.

Intel actually makes two x86 architectures. However Intels core architecture have been designed to cover more markets.

This does have the impact that the architecture wont be tailored to a specific market.

[vm'N]

You can probably Google about a million of them. K12 is being designed to take over the x86 server ecosystem or at least AMD hopes.

K12 is not been designed to take over the x86 server market.

There have been no signs to indicate that.

[LukaP]

Don't know why people are bashing on it being 28 nm, as Intel have shown with devils canyon release. Die shrinkage doesn't play a dramatic difference at the sizes we are now at, unlike they once did, architecture seems to be a lot more important.

What the hell are you on about?

1. 28nm process when intel is at 10nm is madness, thats 3x size reduction
2. DC shrunk nothing
3. TRANSISTOR shrinkage plays a BIGASS fucking difference, thats why AMD have such a high TDP for what their IPC is
4. Yes architecture is important, but you cant just stay at a process that is now 4 years old and call it a day

[vm'N]

That is

• impossible
• counterintuitive

CMT and SMT are two completly different things... Its like saying they are making a pizza that will be both baked and unbaked at the same time.

If I remember correctly, SPARC already have a architecture featuring SMT and CMT.

[LukaP]

If I remember correctly, SPARC already have a architecture featuring SMT and CMT.

Yes ive read some more on it, and apparently its possible to have both, but its not ideal, and you cant leverage both at the same time

[vm'N]

Yes ive read some more on it, and apparently its possible to have both, but its not ideal, and you cant leverage both at the same time

It is only ideal if you go full-costum.

[LukaP]

It is only ideal if you go full-costum.

Do you think amd will try to modify x86 into a frankenstein enough design to make CMT/SMT be better than SMT? not with their money they wont