AMD 65 nanometre chips available

[Gizmo]

The wait for AMD 65 nanometre CPUs is over. At least, this is what on-line stores indicate today.

Read more on the Front Page:
http://www.aoaforums.com/frontpage/content/view/1389/2/

[keithwalton]

It would appear that they're not really avaliable yet.

http://www.theinquirer.net/default.aspx?article=35732

[cloasters]

Looks like Newegg doesn't have them. I wonder if 65 nanometers will really be all that?

[keithwalton]

For amd's sake they need to clock very well. They're only just reaching 3GHz again on a chip, they need to be hitting mid 3GHz's when overclocked however to keep up with the core 2 chips. Intel are ready and waiting to answer with higher clocked core 2's and well there 45nm process isnt far away now. 1Q 2007 i believe shipments are due to begin

[skool h8r]

So about the same time as AMD quads then? As for them needing to reach mid 3Ghz when OC'd, if you take a 3200 at stock, it's at 2.0Ghz (or 1.8Ghz, i forget which). But, it's rating is 3.2Ghz. That's not the actual speed, it's the rating speed. It's kinda hard to explain.

I just pray that AMD have something good planned with these chips. If they market them right, it could turn out very well.

OT: From logon to usable desktop, it takes less than 5 seconds and fully loaded in less than 10. This includes the diamondback drivers, the XGear gamepad drivers, AVG, Sound Drivers, Nvidia drivers, MSN Messenger and alcohol 52%. It's pretty awesome to be able to start playing a game in less than 1 and a half minutes (from power up to game loaded).

[Gizmo]

Originally Posted by skool h8r

But, it's rating is 3.2Ghz. That's not the actual speed, it's the rating speed. It's kinda hard to explain.

You are referring to the PR rating. That was relevant when we were dealing with the old P4 cores, which did much less work per clock cycle. That rating no longer applies when comparing to the Conroe. Clock for clock, the Conroe does just as much, and in some cases just slightly more work than the AMD chip, IIRC.

Originally Posted by skool h8r

I just pray that AMD have something good planned with these chips. If they market them right, it could turn out very well.

Not really. At best AMD will maintain parity. Intel will be running on the 45nM process, where AMD will be at 65 nM. This means that while AMD may have a manufacturing efficiency advantage (as they have been boasting), Intel will have a process size advantage, and will probably continue to have a significant performance advantage, if for no other reason than the fact that Intel's 45 nM process should be able to clock higher.

[keithwalton]

Indeed the PR ratings are now a useless comparison thanks to the core 2 duo chips, as gizmo has said clock for clock the conroe chips are just has fast as the athlons, and infact due to the shared l2 cache architecture they are better at single and multithread apps. The 2.8GHz Fx62 is pretty much beaten in everything by the 2.93GHz X6800 often significantly faster, but the 2.66GHz E6700 alot of the time is quicker than the Faster clocked Fx 62 as well.

Then it comes to head room, the Fx can do just over 3GHz and thats it, where as some 1.86GHz core 2's have been clocking into the low 3GHz as well and other chips hit way into the 3GHz heading on for the 4GHz mark.

The Fx72 (3GHz) from amd on the 65nm process is going to have to overclock to 3.5GHz + to even be worth considering.

[cloasters]

Gad, how the mighty have fallen. I wonder if Intel's dual cores would be as good if their Israeli division never designed the mobile processor with the enlarged L-1 cache? Dang it, I can't remember the proper name for the proc, sorry.

[Gizmo]

Originally Posted by cloasters

Gad, how the mighty have fallen. I wonder if Intel's dual cores would be as good if their Israeli division never designed the mobile processor with the enlarged L-1 cache? Dang it, I can't remember the proper name for the proc, sorry.

I believe you are referring to the Dothan? In any case, you are right. If the Israel facility hadn't convinced the higher-ups at Intel that they needed a separate, purpose-built mobile chip, Intel would be up the creek right now.

[keithwalton]

Banias was there first chip known as the Pentium-M and often branded with a wireless chip as centrino.

[Gizmo]

Ah, that's right. Thanks Keith!

[cloasters]

Dothan, that's it, thanks Gizmo! I still can't believe that Intel was stupid enough to keep building larger L-2 caches--while neglecting to address the performance crippling tiny L-1 cache in generation after generation of their new processors. Nuts, I tell ya!

[Wolf2000me]

I don't know cloasters. As long as the out of order micro ops can be processed without additional delays there really is no problem.
I could be mistaken of course