'Normalized' framerate target GPU power consumption testing

[Maudima]

It's no secret that high framerate Gsync/Freesync displays are a gamer's monitor of choice nowadays. To optimize the gaming and viewing experience with these displays comes framerate capping and technologies like Nvidia's ultra low latency. This optimizes games to run smoothly on these types of displays. In practice this means enabling Vsync and running games at a framerate that's slightly below the monitors maximum refresh rate to focus on smoothness and prevent tearing. To do this it means using settings that won't have the GPU running at 100%. 

 

Assuming gamers are following best practices and not running GPUs at 100% I would be pretty interested to see how for example an RTX3080 would compare to an RTX4080 (or any model honestly) in power usage when targeting a realistic framerate. For a title like Cyberpunk this would probably mean 60fps but for competitive games like Apex or Overwatch this would mean 144/140hz. With rumours flying around that the RTX 40 series might be even more power hungry than the RTX 30 series I was wondering if anything like normalized power consumption testing for specific framerate targets has been done by anyone? Of course the headlines will be dominated by 'it's power hungry' but in day to day gaming I actually don't max out my GPU because I try to follow best practices making any claims about 'maximum' power usage useless for most of my gaming. I would be very interested to see these realistic efficiency numbers. 

Thoughts? 

[Stahlmann]

The point of VRR is to keep screen tearing from happening while BELOW your monitor's refresh rate and while it's fluctuating. In my case even with a 3080 and a 3440x1440 monitor it means my GPU will still be at 100% load in most games.

 

If you're just gonna lock your framerate then VRR won't be much of a benefit for you.

 

I get your point about fps normalized power consumption being useful data to some, but it has nothing to do with how VRR is used.