Innovation from Nvidia? Where?

Whereas Nehalem has at least finally made some large changes to x86 architecture with integrated memory controllers, improved hardware threading, more cores on chip, better SSE etc what exactly has Nvidia done?

As far as I can tell, their cards are a bit faster than before, with some improved power management. Other than that their cards are still too hot and power hungry, their drivers are still buggy, they don't care about fixing their drivers quickly, they don't support DirectX 10.1 and the emphasis is on a few extra fps for gamers rather than rock solid 2D that will be used 99.999% of the time. About the only innovation they've introduced is PhysX, and that was bought in.

Not that ATI is much better with even more excessive power consumption and cards that run far too hot.

@Peter Kay

Maybe that should be "Innovation from Intel, where?" then.

Integrated memory controllers? What, like AMD have had since Opteron shipped some years ago now? Better late than never I suppose.

Improved hardware threading? What, you mean the resurrection of HyperThreading from their old NetBurst architecture?

More cores on chip? Sun, IBM, AMD, blah, blah. I won't get into the argument about whether a GPU shader is or is not a "core", other than to say that, if it is, Nvidia and ATI have something of a lead here .

As for Nvidia, buying PhysX is definately not innovative, using it to provide HW physics on CUDA is IMHO.

@TeeCee

Yup, pretty much all you say is spot on. It's innovation for Intel, not the industry as a whole. The difference , of course, is that Intel's execution has been in general much better than AMD's. They're much better at making tradeoffs that work, even if on paper the competing solution should be technically superior.

Sticking PhysX on CUDA wasn't innovative, but was an excellent business decision. It was inevitable hardware physics would be implemented on graphics hardware - Nvidia and AMD had themselves been saying things in that direction for months/years.

At a time when Nvidia was being spanked by AMD(ATI) PhysX kept them in the game. Now the performance at the high end is pretty much equivalent and AMD possibly win in the ever changing middle ground PhsyX introduces enough uncertainty for some people to choose Nvidia over AMD.. The question is what AMD will do about this; if enough A list titles are suitably enhanced by physics they'll be forced to officially support PhysX

Mr

Did I miss something or did AMD have the controlLer first as well as the so called quick path.

Shouldn't amd have some say as to whether Intel can use either controller or Hypertransport er quick path

What happens when u have a monopoly

"ALL WAYS ARE INTEL'S WAYS!"

@Stiles

So exactly why should an interface that Intel created to tie their CPUs to their chipsets be public domain? Just because you think it should be???

It has been a LONG time since CPU interfaces where standardized across multiple vendors (aka: socket compatible).

This wasn't created as an open interface for the industry (like PCI or PCI Express), but was created as a closed interface between their CPUs and their chipsets. If you want to use that interface, then you have to license it.

Whether the existing license covers this new interface will be determined by the courts.

Eventually...