Hmm not so sure now. Just had another test and so long as I set AF=15 eventually all cores start working and I get even load. This merely helps to kick those other cores into action faster it seems. I'll test some more but maybe not so incredible after all.

Wow, looking forward to trying this.  I just wish there was a way to automate the process.

Hmm not so sure now. Just had another test and so long as I set AF=15 eventually all cores start working and I get even load. This merely helps to kick those other cores into action faster it seems. I'll test some more but maybe not so incredible after all.

Yes, think it needs more testing by other users. I certainly had a smoother flight, and GSX did not stutter. But, could be placebo. FSX can take you round the houses. But different CPU's, different overclocks etc could make a difference. I am no expert with this stuff.

BTW, I use AFM=14

Wow, looking forward to trying this.  I just wish there was a way to automate the process.

Post your results Mark. I am still on the fence with this.

Ok, will try this tomorrow.

Would it have a similar affect on FSX that's still DX-9?

Oh yes, is it what!! BTW, your banned :LMAO: :LMAO:

 

Anyway, just departed Dubai in the T7, using GSX for pushback, I used to get stutters during the PB, but no more. Not sure what core couati exe uses, but it does not stutter the sim any more.

Also my sim is smooth as silk.

Will need to fly out of a more heavier area to prove this is not placebo. We shall see.

 

Don't tease us now, is this some kind of cruel lure into false euphoria? Seriously now, I will give it a go and report back with hopes of great success.

Have to check this out in a few hours - thanks for the heads up

Don't tease us now, is this some kind of cruel lure into false euphoria? Seriously now, I will give it a go and report back with hopes of great success.

Hope so Ric, couldn't handle the humiliation.

Would it have a similar affect on FSX that's still DX-9?

Be the same for DX9 I would assume.

Hmmm... this is interesting. Before the 'tweak', my cores (4 of them) were being pretty efficiently used, but core 0 was almost always pegged at or near 100%, where as cores 1-3 varied depending on the demands of the scene, pretty much in sync with each other, but often at or near 100% for a demanding scene.

 

After the tweak, all 4 cores seemed to exhibit a very similar usage line, which seemed to me seems like a more efficient utilization of resources.

 

I can not say that I noticed superior performance (test was relatively brief), but I did feel that response to my TrackIR head movements MAY have been compromised with the tweak. (I load TrackIR with an elevated priority)

 

Follow-up:

 

I ran a 'scientific' benchmark (using a saved flight and FS Recorder saved flight, frames set to unlimited, TrackIR disabled and FRAPS benchmarking) to test this. I actually got an FPS LOSS of 0.5 for a 180 second test flight. I also noted some nasty pauses/stuttering with the tweak (which I had initially thought might be TrackIR related).

 

So, this tweak is a non-starter for me.

 

Sigh...

What this tweak does, and why it probably is not the best way to run FSX:

 

When FSX starts up it checks its CPU environment as to the execution cores which are available; the default is all "primary" cores, and the .cfg affinity setting overides the default.  Once the available cores are known, it assigns execution threads to these cores in a manner that best assures their mutual cooperation with each other.  Once a thread is assigned to a core, it will exclusively run there.  What happens when you change the affinity to a running process via windows is that you have now just overridden the program's tailored assignment of threads to cores and have told windows to load level all its threads as best as possible across all the cores which you have selected.  While given that FSX CPU resources are now equally utilized for all its execution threads (which windows is doing its best to do so), it is really the case that those threads can now potentially suffer 'thread collision', leading to stutters and less than optimal performance.  I remember doing this exact tweak with my dual core athalon (4200+ X2, oc'ed of course) when SP2 came out, and though the CPU utilization seemed better the performance did not.

 

Edit: I have no idea what P3D does under the same situation.

Thanks for your summary Rod. The only benefit i saw was GSX not stuttering fsx during push back. During flight I saw no benefit, so have not bothered with it.

After much more testing I would say this tweak is a no go, and probably mostly placebo. I tried a number of long flights (some using the tweak, some not) and I really couldn't tell the difference other than if I checked task manager.

 

On the plus side though, this tweak prompted me to change my affinity mask from 14 to 15, so that all cores were utilized. Doing this has definitely improved performance by roughly 30-40% in heavy areas (roughly a 3-5fps gain). I understand that the original reason to set 14 was to leave core 0 for Windows tasks, but I found losing that entire core cost me more in terms of performance. It may also be a legacy tweak from the days of XP, Vista, and 7. I use WIndows 8.1 which may process CPU instructions differently.

Thanks for report Sox. Will change to =15. I totally agree about the placebo effect. As I posted above, I came to the same conclusion.