AMD RV770 refresh -> RV790

Arty said:
I'm not sure how we can compare the back of the PCB w.r.t to pin layouts of the die. Maybe fellix or silentguy can fill us in?
Click to expand...
It can be the same die with a different substrate.
It can be the same die with the same substrate with a different PCB layout (this is, after all, a multi-layer PCB)
It can be a different die with the same substrate with a different PCB layout.

IOW: there's no way to tell...
 
Well, assuming that the cooler retention bracket is the same, the grid of resistors (or whatever they are) is definitely smaller & its clearly a different number of components.
That could just be smaller, better resistors though couldn't it?
 
hoom said:
Well, assuming that the cooler retention bracket is the same, the grid of resistors (or whatever they are) is definitely smaller & its clearly a different number of components.
That could just be smaller, better resistors though couldn't it?
Click to expand...

I agree. How could it be that a chip with reportedly the same structure of RV770 becomes smaller only passing to a different version of the same production process? ;)

Who knows what is usually the purpose of the bigger pins around the chip profile, on the back of the vga? And the purpose of smaller ones?
 
Last edited by a moderator:
So it seems that real board are finally floating around in Taiwan. So where are the leaked benchmarks :D. Looking at the release timeline we should at least have a couple fake ones by now :p
 
I guess "more efficient shaders" mostly means better control flow around them - maybe the RV770 is a bit too upscaled. Do we have any efficiency comparison between the full rv770 and the 4830?
 
It was quite nebulous:

http://www.techreport.com/articles.x/13603/1

Of course, the obvious drawback to this move is a reduction in bandwidth, but we noted long ago that the R600 underperformed for a chip with its prodigious memory bandwidth. AMD says it has tweaked the RV670 to make better use of the bandwidth it does have by resizing on-chip buffers and caches and making other such provisions to help hide memory access latencies. On top of that, higher speed memories like GDDR4 should help offset some of the reduction in bus width.
Click to expand...
So, not "shaders" per se, but, erm ...

I don't know if there's anything to learn from the back of HD3870, but here's one anyway:

http://www.ixbt.com/video3/rv670-3-part2.shtml

Jawed
 
OK, so that's more of an architectural improvement for a specific feature rather than a general "shader" improvement. So not comparable to what Fudzilla is alleging.

I think shader core is prolly irrelevant though, except perhaps in a game like FEAR 2:

http://www.xbitlabs.com/articles/video/display/fear2-project-origin_7.html

Where it appears texture rate dominates (9800GTX+ and GTX260 have almost identical texture rates ) :LOL:

I suppose an increase in interpolator count might give a boost in per-clock performance.

Jawed
 
You must log in or register to reply here.
Back
Top