CMO performance on latest CPU/GPU

[Twistedpretzel]

I am interested in this, hopefully I pick up one of the new Ryzen processors soon and I'll report benchmarks.

Just out of curiosity, what is that metric you are using and how do I see it in my game? (53:1 or 37:1) Thanks!

[thewood1]

I just timed it on a stopwatch. Its not hard.

[Twistedpretzel]

Definitely not hard to do! Thank you for clarifying your testing methodology.

[Rory Noonan]

ORIGINAL: stoltz

I have fixed my problem. I cleaned out some programs that had snuck into my start up. That has fixed the memory leak or whatever problem I was having. Now scenario Bass Drum runs very well. Thanks for the tips.

Glad to hear !

[p1t1o]

ORIGINAL: Twistedpretzel

Definitely not hard to do! Thank you for clarifying your testing methodology.

[:D][:D]

[Twistedpretzel]

ORIGINAL: thewood1

I just timed it on a stopwatch. Its not hard.

Hope I am not beating a dead horse, but I am just trying to establish a solid testing parameter that is repeatable with similar results for each user/setup.

I would assume for benchmark and testing purposes you turn off all messages that would pause the game and run it at a fixed speed until the time runs out, without actually playing the scenario like launching aircraft or giving any human input? Which speed do you recommend using?

[thewood1]

I built a version of Op Bass Drum with missions for the blue side. Of course no pauses. Its not fixed speed. Its as fast as it can run.

My suggestion is to run a mission, any mission, on the old PC and the new one.

[Twistedpretzel]

I had too many variations with timing to make areal determination. I have a 5800x and compared to an older i7-7700k the anecdotal evidence is that CMO runs much smoother during large scenarios, with a lower pulse time at all speed settings.

I did notice of the 16 threads on this CPU one is used heavily, another 3 are used moderately, and the other 12 don't seem to be used for CMO at all. I enabled multithread in the command.ini file in the config folder and after short testing it did not appear to have any noticeable impact on thread usage. For no other reason than curiosity and more power always being better, I wonder what kind of performance would be garnered if all 16 threads were used! It is nice having plenty of processing power in the backend not dedicated to CMO though, it also keeps temps relatively low and doesn't tax the system too much while still providing excellent performance.

That is my super anecdotal feedback.

[thewood1]

Weird that you have so many differences in timing. All of the testing I've done has shown very consistent times with the same rig and scenario.