Nvidia G71 - rumours, questions and whatnot

Ah ok, now it makes sense, was looking at different diagrams and some said TMU's and others had ALU/TMU's

thx

 

Well atleast you woke up from your coma, a lot of people don't.
Infact one of our very own has been in a coma for some time

 

Who would that be? (I wouldnt mind the link and the name via PM )

 

YGPM.

 

http://www.channelregister.co.uk/2006/01/12/nvidia_g71_march_launch/


This is really wierd, goes against just about all other indictors that we have heard about this core well I don't know if they did a typo or what, but 48 pipes and lower clocks then what Hexus's info was. So most likely not a typo.

 

I would say its a typo/confusion with R580 configuration, given the referenced 32 pipelines in relation to G71 in the second paragraph.

 

Oh ya know think they are talking about the x1900 man that was some tough reading

 

Yeah that last paragraph is about the X1900.

 

Anyone , are G71 useing 90 nm with Low-k Or without it ?

 

There are no 90nm process without Low-k.

 

TSMC don't offer a non low-k option at 90, as far as I'm aware. G, GT and LP, at all I/O steps, are low-k.

 

Non-low-k 90nm would had been mighty convenient for NV though I guess

 

Why?

 

Presuming Ailuros is talking about low-k allowing for a higher clock, ATI's new GPUs wouldn't have clocked so high without it, lowering their comparative performance.

 

But that's only a temporary situation, one which nVidia handled pretty well in any case.

 

Ugh, I assumed he meant from the point of view of libraries (which is how I've always viewed it).

Though I still don't understand what it is, exactly, about low-K that burnt NVidia's fingers all those years ago.

Jawed

 

No I was rather thinking of the relativity between different libraries. The transition from 130 to 110nm must have times easier than from 110nm to low-k 90nm.

 

That all depends upon how good the libraries are to begin with.