5870 + Samsung 2233rz

[Kanpai]

It wouldn't bottleneck my Q6700 3.33GHZ would it? 1680*1050 res

 

[appletree]

It wouldn't bottleneck my Q6700 3.33GHZ would it? 1680*1050 res

Depends on your target fps, target quality settings and target games. For example the mentioned HD 6850 is of similar performance to 460 but I find it difficult to play some of the more demanding games with 16xAAF and 4xAA at 1920x1080. There are games where CPU becomes bottleneck much earlier than card of GTX 460 performance level and vice versa. But I consider majority of todays games still to be much more GPU bound than CPU bound. So in other words combination of 3.33 GHz quad + GTX 580 would provide better framerates for majority of the games than highly overclocked hexacore from intel + GTX 460. Some could say that GTX 580 is overkill for 1680x1050 but it really comes to the question what is the minimum fps you find tolerable.

 

[appletree]

I'd help with any graph you need btw if i know how

Hello, no need to worry, I can create the graphs easily from raw fraps file, for example this is how Rroff's data compare to mine on HD 6850 when put into graphs.
http://img3.imageshack.us/img3/1307/msgraphs2.png
However I think what's really needed is if someone could measure it for ATI and nvidia cards on the same system to exclude any other possibility / error.

I found actually that the problem was greatly enhanced when my 5870 was overclocked which i found baffling too.

Could be that with higher clocks - the card had just enough power e.g. to manage the game with lower clocks most of the time = there were fewer frequency switches which resulted in smoother gameplay.

 

[Rroff]

BTW you labelled the graph GTX460, not that it would make any odds with quakelive but my frametimes are from a GTX470.

It still looks like powerplay/adaptive power management is kicking in on your 6850 plot even with locked clocks? - every time you have an increase in time between frames its almost always followed by a decrease on the next, whereas on mine its fairly variable but within a tight range of 8ms.

 

[sadbuttrue]

appletree - I assume your cpu is running with power saving features disabled?

 

[appletree]

Rroff: afterburner reports 775 MHz core and 1000MHz memory all the time. And it feels MUCH better than without the tweak. When I am focused on the game it doesn't bother me that much (though sometimes I notice a minor stutter for a fraction of a second etc. If I focus on smoothness instead of the game itself then probably it bothers me slightly more - unless eyes are playing tricks with me.

But to me the fact that that I am unable to do the switching (forced clocks in 3D, default clocks in 2D) automatically or that I have to install third party beta software just to enable correct 120Hz experience, or spend time analyzing on problems which should work right out of the box is much more annoying then the actual QL smoothness after the tweak. If I didn't feel like I need a more powerful card for other games I probably wouldn't be considering nvidia just because of the Quake live @ 120Hz
However because of this the decision whether to get high end ATI/AMD or nvidia is much easier.

sadbuttrue: yes, "with all power savings disabled in bios and windows"