Tom's looks at Intel's 65nm P4

[KimB]

Linkage:
http://www.tomshardware.com/cpu/20051007/index.html

Nothing special as far as I can tell. The only difference is power consumption, which will in turn lead to somewhat higher clockspeeds down the road. It looks like this will put Intel's P4 back on par with AMD's Athlon64.

But once AMD shrinks down to 65nm as well, it'll be the exact same thing all over again. Intel needs to move on to its next-gen architecture quickly if it wants to have much of a chance.

Edit: Actually, comparing these power consumption graphs, this still may not be enough to bring the P4 to power consumption parity with the Athlon 64.

 

[wireframe]

I don't think Cedar Mill is intended to do much more than to alleviate some of the thermal issues the Prescott has had. That and the shrink should be good for some more money for Intel.

Did you read the part about heat and the fan now only being slightly audible when under full load, whereas the fan is silent during normal operation? That alone is a good improvement. It's not like the Pentium 4s are that slow.

 

[KimB]

Sure, but it still doesn't look like it'll approach AMD's power/performance ratios. It appears to be an improvement, but only marginally.

AMD, on the other hand, has lots of room to grow with its Athlon 64 line, even before another die shrink comes in.

 

[Slides]

But I wonder if AMD is really prepared for Intel's new architecture? What's the ETA on that?

 

[Jimmers]

Intel's shooting for 45nm for their next major architechture revision. 65nm is a stop-gap, and I really doubt it will be anything other than a superficial marketing point. For the consumer end, this may mean more competitive, cheaper dual-cores.

 

[wireframe]

But I wonder if AMD is really prepared for Intel's new architecture? What's the ETA on that?

Second half of 2006.

 

[Guden Oden]

Intel needs to move on to its next-gen architecture quickly if it wants to have much of a chance.

What are you talking about?

You must be living on some totally different planet than the rest of us humans.

 

[Sonic]

2nd half of 2006 is correct. The new architecture should take the strengths of P4 (if there are any) and of the Pentium M to put into the new CPU's. A dual core Pentium M with various improvements in the way of FPU and other things should be pretty good. That and support for 64-bit and Intel is on its way. I think there will be dual core versions of Merom and Conroe with 4 MB of cache.

 

[Rur0ni]

So Intel is delivering a new architecture on 65nm as well, besides the current P4 getting moved to 65nm. So it's not necessarily a stop-gap is it? But I do remember Intel I think claiming 45nm solves leakage problems, for the most part.

 

[Moloch]

Nice to see game benchmarks at a decent res so as to show more realistic results
As for mr 65nm P4- you're a bit late and you're still not good enough.

 

[KimB]

What are you talking about?

You must be living on some totally different planet than the rest of us humans.

Well, Intel is still has a near monpoly, it is true. But I mean in the minds of enthusiasts like us. I'm sure that Intel can keep floating on their strongarm tactics and marketting muscle to keep AMD in a tiny portion of the market until their next new architecture arrives.

But AMD has a huge lead among enthusiasts right now (just look at the ORB, for instance). That's what I was referring to.

 

[KimB]

2nd half of 2006 is correct. The new architecture should take the strengths of P4 (if there are any) and of the Pentium M to put into the new CPU's. A dual core Pentium M with various improvements in the way of FPU and other things should be pretty good. That and support for 64-bit and Intel is on its way. I think there will be dual core versions of Merom and Conroe with 4 MB of cache.

Well, there are plenty of good things about the Pentium4. Basically, everything about the Pentium4 is great except its primary design goal: to reach high clockspeeds. Having your primary design goal being just plain bad is pretty damning, however.

I'm sure Intel's next architecture will be good, but will it benefit at all from making use of the x86_64 instruction set, other than the ability to use more memory? This is all that Intel is currently marketting their "EMT64" setup as (God forbit they credit AMD with the instruction set), from what I can see on their website, and I haven't yet see any benchmarks that show a performance improvement from 64-bit operation on Intel hardware.

AMD, on the other hand, gains a good 20% performance in floating-point intensive applications, which will be a definite boon for games.

But it is true that if AMD wants to stay ahead, they need to put out someting dramatic by the time Intel releases their next architecture. It's just that there doesn't appear to be any prospect for their current P4 architecture to really compete with the Athlon 64 in performance, price, or power consumption.

 

[Guden Oden]

But AMD has a huge lead among enthusiasts right now (just look at the ORB, for instance). That's what I was referring to.

Intel isn't particulary interested in the ORB though, because as long as their overall marketshare is what it is, they don't need any enthusiasts schmentusiasts.

Even having the enthusiast mindshare hasn't helped AMD get out of the rut they're in...

 

[Tim]

Intel isn't particulary interested in the ORB though, because as long as their overall marketshare is what it is, they don't need any enthusiasts schmentusiasts.

Even having the enthusiast mindshare hasn't helped AMD get out of the rut they're in...

AMD has never made more money on their CPU buisness than they do now, AMD has never had bigger revennue in their CPU business as they have now, AMD has never had as high a ASP compared to Intel as they do now and AMD has never has as high a $-share as they do now. All in all AMD has never done better on the CPU side of their business.

 

[silence]

And as far as i know, they are also planning to release something new next year....
Dont forget their deal with IBM.

"We were already collaborating on 65-nanometer and 45-nanometer technology," said Chris Andrews, an IBM spokesman. "Now we've also added joint development on 32-nanometer technology to the agreement."

AMD has also now licensed IBM's C-4 "flip chip" packaging technology, which is used to connect the microprocessor to the circuit board on Big Blue's server processors, Andrews said.

and K10 is expected somewhere in 2007.... dont forget that AMD has been working in peace on K10, while Intel had to dump half of its line and redo all roadmaps and completlly change their priorities. So, IMO, K10 will be powerfull enough to wipe anything Intel throws out in 2006....

 

[phenix]

And as far as i know, they are also planning to release something new next year....
Dont forget their deal with IBM.



and K10 is expected somewhere in 2007.... dont forget that AMD has been working in peace on K10, while Intel had to dump half of its line and redo all roadmaps and completlly change their priorities. So, IMO, K10 will be powerfull enough to wipe anything Intel throws out in 2006....

K10? Whatever happened to K9?

 

[silence]

K10? Whatever happened to K9?

K9 became K10.... and what AMD designates as K9 are dual cores that you can buy already... they just shifted number by one, its still in development and i think with IBM giving them more and more R&D they gonna do monster....

from what i know it was decided that K8 are single cores, dual cores became K9 and next gen then became K10....

nothing really spectacular.

 

[hoom]

K10 because potentially in the eyes of the consumer:
K9 = Canine = Dog = poor performer = buy Intel

 

[Moloch]

K10 because potentially in the eyes of the consumer:
K9 = Canine = Dog = poor performer = buy Intel

How many consumers know the codename of their cpu?
Overall, not just enthus.
And if you read the thread you'd know dual core A64s are K9s.

 

[overclocked]

How long has AMD been working on their next generation of cpus, now K10?