P3D V3.2 Performance SLI vs non-SLI and other stuff...

[SteveW 1,792]

Would you still suggest 1360 ?

Well, it's time better spent than with testing no AM, that's for sure.

 

No AM will fill all your cores with sim jobs, the activity of these climbs and falls with the sim load. As the sim loads up this activity restricts the throughput of other processes on those cores squeezing them out. However, sub-processes of the sim run in the affinity of the sim, and so these also load up the cores running the sim jobs, squeezing them out.

 

It's a balancing act that needs testing per system, but this generally leans toward restricting the sim to four cores, because ultimately the sim will be loaded up heavily.

 

I noticed that Mike suggested his setup worked better across a broader range of situations. This is the real purpose of tuning the correct AM. Final fps is too heavily governed by factors other than CPU throughput, and is not a good way to compare the characteristics of the sim.

 

Hi Claude, I hope that helps answer the question to some extent. I've given away everything I can.

[Claude Troncy 4]

 

 

Hi Claude, I hope that helps answer the question to some extent. I've given away everything I can.

Thanks

Claude

[supersym 52]

Hello !

I would want to know if you get the same average between SLI vs Non SLI with the latest drivers ?

Despite the side effects It seems for me that the difference decreases.