AMD or Intel?

[Apokalipse]

Yorkfield has many of the same advantages over Kentsfield that Deneb does over agena and it tends to 5-10% faster. I would expect to see a similar increase from Deneb vs agena which would put them behind Yorkfield not to mention Bloomfield.

You're still making assumptions about how much of an IPC Deneb is getting, and you don't know how well Deneb will improve in terms of overclocking.
I would expect overclocking alone to have a significant increase.

And again, that's just from initial Deneb release, which don't incorporate the SiGe transistors, or the high-K metal gates co-developed with IBM

[edit]Bloomfield will still be much more expensive, and not just the CPU's themselves[/edit]

 

[Puddle Jumper]

do you mean phenom over agena? i'm a little confused

The kentsfeild -> Yorkfield transition is very similar to the Agena -> Deneb one so I would expect to see a similar performance increase. (5-10%)

 

[Apokalipse]

do you mean phenom over agena? i'm a little confused

Deneb = 45nm Phenom
Agena = 65nm Phenom

The kentsfeild -> Yorkfield transition is very similar to the Agena -> Deneb one so I would expect to see a similar performance increase. (5-10%)

You say that without knowing what's actually improved in terms of the architecture

 

[Ventriac]

whew, the rere understands now

 

[Puddle Jumper]

You're still making assumptions about how much of an IPC Deneb is getting, and you don't know how well Deneb will improve in terms of overclocking.
I would expect overclocking alone to have a significant increase.

You are also making assumptions. I don't agree with recommending components based of of hardware that hasn't been released yet and has few benchmarks available. Many people started recommending sli boards before GT 200 was released because of some of the early info available on it but now that rv770 turned out better the people who bought those boards are stuck with nvidia cards.

 

[Apokalipse]

You are also making assumptions.

No I'm not. I haven't made any declaration as to how much of an IPC increase it's getting.

I don't agree with recommending components based of of hardware that hasn't been released yet and has few benchmarks available.

Even when there's credible evidence surrounding said products, not to mention the fact that they're cheaper?

 

[zmatt]

agena is the codename for the current core int he phenom. Both Agena and Deneb will be marketed as phenom so to differentiate you ahve to talk about the cores.

http://www.techist.com/forums/f76/amd-intel-future-149802/



Oh how so long ago. Those were some pretty big claims made back then Apok. Most of us were man enough to admit we were wrong and moved on.

What if Intel quad cores outperforms AMD "true" quad core ? Then which one you recommend ?

a voice from the past. And you said.

It's easy to ask "what if?"
but unless it actually equals reality, I don't bother.
Four cores on one chip is still better, irrespective.

Yes it is easy to talk what if. So what if Deneb is a repeat of Agena, which honestly is far more likely knowing what we known about K10. And no, 4 cores on one chip isn't always better. But that was proven a long time ago.

Even when there's credible evidence surrounding said products, not to mention the fact that they're cheaper?

All four cores will have 128-bit FPU's, a pipeline that retains its 12 stages (instead of C2D's 14 - lower is better), but at the same time increases the IPC to 4, as well as lowering latencies, improving virtualisation, branch prediction, and the speed that instructions can actually get to the cores (which the faster hypertransport bus also helps with).

Both the same person, both the same level of credibility. If you had a question of the credibility of the statements you should have said something. But I have a feeling anything that makes AMD look good, true or otherwise, you will toss out.

 

[Apokalipse]

Oh how so long ago. Those were some pretty big claims made back then Apok. Most of us were man enough to admit we were wrong and moved on.

I admitted to what I was wrong about. But there's plenty I wasn't.

Yes it is easy to talk what it. So what if Deneb is a repeat of Agena, which honestly is far more likely knowing what we known about K10.

I highly doubt that.
Especially considering that Agena -> Deneb, although there's more to it than you probably think, is a much simpler transition than K8 -> K10, and considering that, again, a monolithic core on 65nm is not exactly ideal, something Intel themselves have said.

Even if it were just a die shrink and nothing more (which it most definitely isn't), I would expect Phenom to do a lot better on 45nm.

 

[Puddle Jumper]

No I'm not. I haven't made any declaration as to how much of an IPC increase it's getting.
Even when there's credible evidence surrounding said products, not to mention the fact that they're cheaper?

You may not have specifically said how much of an ICP increase increase its getting but you did make assumptions.

I would expect overclocking alone to have a significant increase.

Sounds like an assumption to me.

 

[zmatt]

And jsut another thing to throw out there. Since we are on the topic of how good Deneb will be. A long long time ago Apok you had to say this

a die shrink, an instruction set and extra cache is not much in relation to architectural changes.

You cant have it both ways. Your talking about how much better Deneb will be with simple changes, but on the topic of the Core2 you said it dosent amount to much. So which way is it?