AMD VP gives ballpark Llano gpu performance [Fudzilla]

[jvroig]

IMO the downside of Llano is the claim about 90% of 2010 performance, SB will be waaaay much better in this (will be more than 2010 performance).

What makes you think that? You can't just say "2010 performance" and not identify exactly what kind of performance you mean. Do you mean 2010 IGP performance? 2010 low-end performance? 2010 mid-range? 2010 high-end?

And as far as I remember, that 90% quote (in full, actually "90% of mainstream performance in half the die size") quote was applied to Bobcat (a CPU core, nothing to do with GPU), which was meant to be taken as "better than Atom while competing in effectively the same market".

Otherwise, if you insist that you are correct and that I am mistaken, then show me a link to where this info has been presented.

Here's mine: [Bobcat] delivers 90% of today’s mainstream performance in less than half of the silicon area.

This has been known for quite some time, so I believe you have just been a little confused with all the Ontario / Zacate / Llano talk and benchmarks floating around. The only real estimate of performance has come from the AMD VP (topic of this thread) saying "400-500 GFlops", which doesn't really amount to much until real benchmarks have been done.

 

[Aetherius]

Yes, I'm sorry. What I mean is, 90% (today) mainstream (processor) performance (not graphic). Actually it decent enough performance for everyday task, but it isn't Llano but Bobcat? Well, my mistakes.
About the Intel HD Graphic quality, maybe you could try playing games using IGP of intel, nvidia, and amd. Set the quality to high, let your eye see it. I don't know whether it is driver or hardware. Last time I'm playing Heroes of Newerth to compare between my GT 8400M with Intel. I set both to high quality. Intel has good framerate, but a big 'meh' on the quality (bad texture, no lightning, no reflection, nothing).
So, I'm not sure 2-3x increase in Intel IGP will translate to increase in graphic quality.

 

[Dresdenboy]

BTW with all the changes in the Llano core, more power efficiency, new process and maybe some unannounced functionality (still there is some new logic in the FPU and the L/S unit), I'm expecting it to show a significant step up from AMD's current 45nm designs.

Llano also seems to have APM Boost. At least Hiroshige Goto included this feature for Llano in his diagrams. This improved turbo mode could be about pushing voltage and frequency until actual power consumption hits the TDP minus a safety margin. Not sure, if it will go even higher eating up previously "saved" TDP headroom, as SB does.

 

[Martimus]

Is that a driver issue or a hardware issue, because according to Fudzilla here it looks like Intel is planning to double graphics yet again in 2011/2012 with Ivy Bridge.

If that happens, Intel will have successively doubled its GMA HD graphics performance in every year from 2010 to 2012. (45mm GPU in 2010, 32nm GPU in 2011 and 22nm GPU in 2012).

This is pretty standard in the discrete market, so seeing it in the IGP market makes sense if they are finally updating the IGP with each node shrink. At the same time, I don't see this as catching up, since AMD will be doubling IGP performance with every node shrink as well.