So I've been doing a quick bit of testing with FRAPS benchmark and FRAFs for reading the graphs.

 

My Hardware:

 

i7 4790k OC @ 4.7GHZ

8 GB 1866mhz Corsair ram

EVGA GTX 970 4GB GPU

Win 10 on SSD

Normal HDD's for everything else including P3D

 

Software

Win 10

Prepar3D V3

ORBX Global, vector, EU England and EGKA Shoreham were used including EU landclass

 

NVI 4xSGSS and 4xMSA in P3D

 

So I saved a scenario in the baron at EGKA on the active runway, 'clear skies' weather theme, AI and road traffic etc turned off as to keep variables to a minimum. Each benchmark consisted of immediate take off and climb to 2000 feet at 1000fpm and maintaining runway heading this was done for 4 minutes each time.

 

Its getting late so I only tested a few different combinations.

 

HT=ON AM=84 FPS locked at 31

 

HT=ON AM=84 FPS unlimited

 

HT=OFF NO AM 31FPS Locked

 

HT=OFF NO AM Unlimited FPS

 

These graphs don't mean a lot to me, So i'll let the experts take a look I'd imagine the smoothest looking graph is performing the best? Tomorrow I may try some more combinations maybe HT=ON with no AM for instance. I've always felt that HT=OFF and unlimited FPS felt the smoothest in Prepar3D. If people want to see my settings I'm happy to upload screenshots of them too. Please note the Vsync and triple buffering were not used in any of these tests.

 

Also I tried AM 212 with HT on and although performance seemed good, after a couple of minutes of flight ground textures were really blurry. On my rig it seems if I mask the first physical core and P3D uses the second core as the main core I get blurry textures and missing autogen.

 

Interested to hear peoples thoughts on these graphs

 

Just realised that FRAFS sets its own scale on the graph for each file so I've made each graph the same scale now oops.

 

HT=ON AM=84 31FPS Locked

 

HT=ON AM=84 Unlimited FPS

 

HT=OFF No AM 31FPS Locked

 

HT=OFF No AM Unlocked FPS

 

That makes it a bit clearer!

 

Also Tried

 

HT=on No AM 31FPS Locked

HT=on No AM unlimited FPS

 

The long frames you are seeing at the end of each graph is where I'm approaching another airport and its loading in.

I tried AM 212 with HT on and although performance seemed good, after a couple of minutes of flight ground textures were really blurry. On my rig it seems if I mask the first physical core and P3D uses the second core as the main core I get blurry textures and missing autogen.

212=(11,01,01,00), which reduces the number of cores available to the sim. So although fps remains OK, your data collecting may need more horsepower. If you don't need to avoid core 0 then don't, try 85 or 170.

...Looking at that case a little closer;

 

84=(01,01,01,00)

212=(11,01,01,00)

 

Both provide three cores to the sim, but 212 presents what looks like four cores, but in fact is only four "Logical processors", the last two both occupy the same physical core. So the sim splits its major threads over 4 cores, knowing nothing of the three actual physical cores.

 

With 212, If either of the last two data collecting threads (sharing the last core) equal demand over 100% throughput, the Turbo step may increase if it can or the main threads starts to wait for data.

 

Could be worth a try compared to 212; 116=(01,11,01,00) to avoid core zero and keep to three cores. This alters data collection slightly, moves the third threads back to core 2 to share with the threads on the second LP, leaving the fourth LP threads to a free core.

4 cores HT enabled. With core zero saved for other tasks, and 4 LPs allocated with AM=116. The sim spreads across four processors, dividing out into four parts, importantly, three of which, 1, 3, and 4, utilise maximum throughput when loading terrain. Since we're really only running on three cores, it makes more sense to allocate the four threads in this way if we have intense terrain loading. During other times the threads 2 and 3 combined are not maxing the core (2). There are some processes on core 0 to avoid, but can be seen getting maximum throughput with terrain loading:

 

 

These graphs don't mean a lot to me, So i'll let the experts take a look I'd imagine the smoothest looking graph is performing the best? Tomorrow I may try some more combinations maybe HT=ON with no AM for instance. I've always felt that HT=OFF and unlimited FPS felt the smoothest in Prepar3D. If people want to see my settings I'm happy to upload screenshots of them too. Please note the Vsync and triple buffering were not used in any of these tests.

 

 

Well HT=OFF No AM 31FPS Locked looks pretty good going by your charts.

Do you still think HT=OFF and unlimited FPS feels smother subjectively?

 

I don't see the point of using 31 as a locked frame rate if you are not using Vsync and maybe not even then.

30 would be more logical.

Would be good to see a run at HT=OFF No AM 30FPS Locked and Vsync on with triple buffering.

 

gb.

Thanks steve I'll give your suggestion a try.

 

Gboz

I don't know why I opted for 31 fps. I vaguely remember reading something about locking at 1 or 2 fps above your target. My reasoning against using vsync and triple buffering is that my monitor is 60hz, my understanding is that vsync works with your monitors refresh rate which would mean running at 60fps to match the 60hz however this is not really possible in heavy scenery areas with ad dons such as pmdg. When able to reach 60fps with vsync on it is buttery smooth. But at 30fps with vsync and triple buffering on, panning in the vc doesn't seem smooth at all to me and neither does the sim. Correct me if I'm wrong.

 

Yes I still think ht off and unlimited fps feels the smoothest even though frames locked on the graphs looks smoothest. I don't quite understand why!

Thanks steve I'll give your suggestion a try.

 

Gboz

I don't know why I opted for 31 fps. I vaguely remember reading something about locking at 1 or 2 fps above your target. My reasoning against using vsync and triple buffering is that my monitor is 60hz, my understanding is that vsync works with your monitors refresh rate which would mean running at 60fps to match the 60hz however this is not really possible in heavy scenery areas with ad dons such as pmdg. When able to reach 60fps with vsync on it is buttery smooth. But at 30fps with vsync and triple buffering on, panning in the vc doesn't seem smooth at all to me and neither does the sim. Correct me if I'm wrong.

 

Yes I still think ht off and unlimited fps feels the smoothest even though frames locked on the graphs looks smoothest. I don't quite understand why!

 

Fair enough. Can't argue with what you are seeing in front of you.

I just thought vsync at 30 FPS might of given smother results.

 

gb.

Tried AM 116 thismorning and here are the results

 

HT=ON AM 116 30FPS

 

HT=ON AM 116 unlimited FPS

 

This AM did not help, I noticed the delayed texture loading when approaching the airport near the end.

Walders, it appears locked frames is the way to go.

Tried AM 116 thismorning and here are the results 

This AM did not help, I noticed the delayed texture loading when approaching the airport near the end.

If you don't need to avoid core 0 try 85 or 170.

If you don't need to avoid core 0 try 85 or 170.

since you have mentioned an AM=116, I've been trying it out today and getting a good smooth performance. The widows task manager shows a similar graph you posted, except I have a little more activity on the first two cores as I have assigned am for ASN and TrackIR on the first two cores.

I running an iiyama monitor at 2560x1440 and settings are fairly high. testing using FTX Norway, ORBX hd trees., with base GEPD and UT Europe installed.

Frames at the moment locked at 60 as it seems to give me the best result.( have been using 30 for awhile but decided to run 60.

 

bob

 

edit I7@4770 o/c 4.4 980ti

Good work Bob.

What about using this program to set cores; Multi Core Advanced Environment ?

http://www.avsim.com/topic/287839-multicore-utility-fsps-multicore-environment/

 

It allows you to automatically set affinity and priority for multiple apps. This will not influence the rest of your apps and services which will continue to use any core that the OS task scheduler assigns for them.