FSX-MS Is AffinityMask Important??

[lownslo 592]

9 hours ago, electricman said:

Well then you have superb eyes since the resting periods of the physical core are in the order of milliseconds and that you can't see in taskmanager

or do you claim that a quad core with HT on is the same as an octo core? then maybe you should take this discussion on to Intel and their engineers.

regards, Herman

Not surprisingly, you've done everything here but present supporting evidence.

[electricman 5]

Google is your friend and you find all the evidence  you need about what exactly HT is and how it works.

[lownslo 592]

Yeah, I didn't think so.

[butwanna 0]

+Capt Spears Sorry, I didn't mean to mislead anyone with my post. I was quoting another websites tip (Kostas flight simulator world) on Affinity Mask. I probably should have kept my mouth shut, I was just learning myself and I still am, I prematurely took Kostas tip as gospel I guess. I didn't know (according to a post here) that FSX SE already has FSX using cores 1, 2 and 3? (I'm using computer lingo 0=core 1, 1=core 2 etc.) I wonder if that's true because. I don't use AM in my fsx.cfg. I was thinking FSX was using 0 and 1 and so I was moving other running applications off of 0 and 1, forcing them to use 2 and 3. If FSX SE is already using cores 1,2 and 3 I have been hurting my performance, not helping it. Now I am all confused.

[Sagoland 22]

https://steamcommunity.com/app/314160/discussions/0/496881136926977562/

Look under stability updates.

[butwanna 0]

41 minutes ago, Sagoland said:

https://steamcommunity.com/app/314160/discussions/0/496881136926977562/

Thank you Sagoland, That article states that for FSX steam edition the correct Affinity Mask value is already calculated and put in the fsx.cfg "for i3, i5 and i7 cpus. Unfortunately I still have a core2quad q9650 and there is no such parameter in my fsx.cfg at all so I still have to figure out what it does without one or with one. I know that my cpu doesn't have HT and I believe the correct value for the CPU is 14 but I still would like to confirm which cores FSX SE uses with no AM (default) and what cores it will use if I insert an AM of 14. And thus far I keep getting conflicting answers. But that is a great link. thx again 

[electricman 5]

FSX SE has an automated AM build in hardcore, it isn't in the cfg and there is no need for it neither, with your CPU you are better off without it and giving FSX all the CPU power it can get. A lot of the AM stories are no more then snake oil, put it in and you get more stutters, take it back out and you find yourselfs at the edge of a completely new install.

You want some real gain? then in NVidia inspector or you NVidia control panel change the "prerenderd frames" from 3 to 1 and try to run FSX on high priority if your system can handle it ( download freeware Prio and install, then start FSX and in the taskmanager tab Processes right click and set the priority to high, then close FSX and restart your system to let that take effect, and enjoy FSX ( gained an average of 7 frames in KSEA PNW sceneries ( almost on constant 30 frames with the iFly 737 NGX) 

[TechguyMaxC 516]

43 minutes ago, electricman said:

FSX SE has an automated AM build in hardcore, it isn't in the cfg and there is no need for it neither, with your CPU you are better off without it and giving FSX all the CPU power it can get. A lot of the AM stories are no more then snake oil, put it in and you get more stutters, take it back out and you find yourselfs at the edge of a completely new install.

You want some real gain? then in NVidia inspector or you NVidia control panel change the "prerenderd frames" from 3 to 1 and try to run FSX on high priority if your system can handle it ( download freeware Prio and install, then start FSX and in the taskmanager tab Processes right click and set the priority to high, then close FSX and restart your system to let that take effect, and enjoy FSX ( gained an average of 7 frames in KSEA PNW sceneries ( almost on constant 30 frames with the iFly 737 NGX) 

 

There is copious data on this subject, including frame time analysis of the affects of affinity mask on various hardware and settings configurations.  The data was gathered in P3D, but P3D still uses ESP as its engine and the scenery engine has not been entirely re-written, as the graphics engine has.  In other words: the data should be largely comparable to FSX/SE.  

[electricman 5]

For those non believers that want prove and those want to test AM for themselves without screwing up the cfg file; start your flightsim ( doesn' t matter which one you have and beginning screen is good ), open taskmanager, tab processes and find fsx.exe or p3d.exe, right click on it and and then click " set affinity", uncheck " all cores" then check the cores you want to work and minimize the taskmanager ( don't switch it off or it will revert back to all cores ) ( AM14 is cores 1,2 and 3 checked  and core 0 unchecked, for those already have the AM in the cfg they should see here the same as the cores they set in the cfg, if not then your entry in the cfg isn't correct and unchecked means in contradiction to some believers that the core is switched off for your sim, if you don't believe that then AM is no more then a hoax ), now go flying a steady pattern which you can repeat and over some denser scenery and especially watch out for microstutters and stability in framerate and bluries, not a gain in framerate, repeat this with only 2 cores and 1 core to see the difference and see what happens in sim.If no improvement against all 4 cores then switch all cores back on and close task manager and forget about AM, don't look at the graphs, they don't tell you what your sim is using on CPU power, any activity left on a core that is switched of for your sim comes from your OS system, services running in the background ( 50+ services are still running ), your AV still roaring a bit un the back, your weather engine connecting to the internet and downloading and installing weather, maybe like I've seen some guys listening to streaming music while flying, others on a second screen checking the stock market to see if they have made enough money to actually buy the real plane etc, etc,.

For those still thinking AM brings some improvement; if you have a 4 core with HT on you will see 8 cores in the set affinity tab, now for every core you have switched on put a "1" on a peace of paper ( lowest core on the left, highest core on the right), for every core switched off a "0", that gives you the binarie code for your AM, convert that with the many binarie to decimal converters on the net and you have your number to put in the AM.

I hope this helps people to determine for themselves wether or not AM will help them and don't have to follow sites long discussions any more about AM from people only out for a discussion instead off helping people with real problems.

[Capt Speirs 52]

10 hours ago, electricman said:

For those non believers that want prove and those want to test AM for themselves without screwing up the cfg file; start your flightsim ( doesn' t matter which one you have and beginning screen is good ), open taskmanager, tab processes and find fsx.exe or p3d.exe, right click on it and and then click " set affinity", uncheck " all cores" then check the cores you want to work and minimize the taskmanager ( don't switch it off or it will revert back to all cores ) ( AM14 is cores 1,2 and 3 checked  and core 0 unchecked, for those already have the AM in the cfg they should see here the same as the cores they set in the cfg, if not then your entry in the cfg isn't correct and unchecked means in contradiction to some believers that the core is switched off for your sim, if you don't believe that then AM is no more then a hoax ), now go flying a steady pattern which you can repeat and over some denser scenery and especially watch out for microstutters and stability in framerate and bluries, not a gain in framerate, repeat this with only 2 cores and 1 core to see the difference and see what happens in sim.If no improvement against all 4 cores then switch all cores back on and close task manager and forget about AM, don't look at the graphs, they don't tell you what your sim is using on CPU power, any activity left on a core that is switched of for your sim comes from your OS system, services running in the background ( 50+ services are still running ), your AV still roaring a bit un the back, your weather engine connecting to the internet and downloading and installing weather, maybe like I've seen some guys listening to streaming music while flying, others on a second screen checking the stock market to see if they have made enough money to actually buy the real plane etc, etc,.

For those still thinking AM brings some improvement; if you have a 4 core with HT on you will see 8 cores in the set affinity tab, now for every core you have switched on put a "1" on a peace of paper ( lowest core on the left, highest core on the right), for every core switched off a "0", that gives you the binarie code for your AM, convert that with the many binarie to decimal converters on the net and you have your number to put in the AM.

I hope this helps people to determine for themselves wether or not AM will help them and don't have to follow sites long discussions any more about AM from people only out for a discussion instead off helping people with real problems.

 still waiting for a 64 bit install and fly sim. I fear that by the time something like that comes out and with all the time spent on the ground trying to make this dinosaur fly, I will have forgotten how to fly. 

gotta admit it is fun to mess with FSX and all the good people you get to meet, you normally wouldn't if all was running well.

[Zatoichi 45]

I tried a heavily edited config versus one generated by FSX.

The FSX one got roughly twice the frame rate as the edited one.

50 fps versus 100 fps in the opening with the ultralight.

 

I still get between 10 and 50 (using the default config) when using a Carenado, though.

[TechguyMaxC 516]

On 5/2/2017 at 1:53 AM, electricman said:

For those non believers that want prove and those want to test AM for themselves without screwing up the cfg file; start your flightsim ( doesn' t matter which one you have and beginning screen is good ), open taskmanager, tab processes and find fsx.exe or p3d.exe, right click on it and and then click " set affinity", uncheck " all cores" then check the cores you want to work and minimize the taskmanager ( don't switch it off or it will revert back to all cores ) ( AM14 is cores 1,2 and 3 checked  and core 0 unchecked, for those already have the AM in the cfg they should see here the same as the cores they set in the cfg, if not then your entry in the cfg isn't correct and unchecked means in contradiction to some believers that the core is switched off for your sim, if you don't believe that then AM is no more then a hoax ), now go flying a steady pattern which you can repeat and over some denser scenery and especially watch out for microstutters and stability in framerate and bluries, not a gain in framerate, repeat this with only 2 cores and 1 core to see the difference and see what happens in sim.If no improvement against all 4 cores then switch all cores back on and close task manager and forget about AM, don't look at the graphs, they don't tell you what your sim is using on CPU power, any activity left on a core that is switched of for your sim comes from your OS system, services running in the background ( 50+ services are still running ), your AV still roaring a bit un the back, your weather engine connecting to the internet and downloading and installing weather, maybe like I've seen some guys listening to streaming music while flying, others on a second screen checking the stock market to see if they have made enough money to actually buy the real plane etc, etc,.

For those still thinking AM brings some improvement; if you have a 4 core with HT on you will see 8 cores in the set affinity tab, now for every core you have switched on put a "1" on a peace of paper ( lowest core on the left, highest core on the right), for every core switched off a "0", that gives you the binarie code for your AM, convert that with the many binarie to decimal converters on the net and you have your number to put in the AM.

I hope this helps people to determine for themselves wether or not AM will help them and don't have to follow sites long discussions any more about AM from people only out for a discussion instead off helping people with real problems.

You can't argue with raw data, which I provided in the link in the post above yours.  The frame time data demonstrates there is a difference in performance between various AM settings.  The argument you have presented thus far provides not an iota of evidence, and merely states an opinion, which is demonstrably false.  

[lownslo 592]

On 5/1/2017 at 8:22 AM, electricman said:

You want some real gain? then in NVidia inspector or you NVidia control panel change the "prerenderd frames" from 3 to 1...

This is not definitive, and indeed a correct Maximum Pre-rendered Frames setting is dependent upon a number of variables.  A search of just the Avsim forums reveals a number of threads... all of which come to a different conclusion.

More info here: http://www.tweakguides.com/NVFORCE_7.html

and our own SteveW's perspective here: http://www.codelegend.com/forums/viewtopic.php?f=2&t=500

[electricman 5]

Hyper-threading on or off makes no difference, as long as an appropriate AF is applied.

I see no advantage in using 3 cores to free up one processor for background applications. It's best if you just just limit background applications, and run on all four cores.

Use the internal frame limiter! Frame stability depends on the internal frame lock. Unlocking the frames creates horrendous micro-stuttering. An external frame limiter does not alleviate the problem.

Overclock your CPU. The increases in frame stability and average frame rates are quantifiable.

If you still see some stutters,  then temporarily lower your LOD_Radius until you are in a more suitable area to return it to the max value. Higher LOD values seem to be responsible for the longest frames, and it's probably due to the CPU just being swamped.

 

I only read here exactly the same as I stated, don't know how you interpreted that info, first off all this is for P3D which was changed by LM, no AM is the same as an AM four cores and the guy states that OC'ing helped, he didn't test the OC with no AM which is a pitty which probably would have given the same results.

I only gave an easy way for everyone to test the AM without disrupting there cfg ( I tested AM a long time ago and upon removing it out of the fsx.cfg I ended up with a slideshow and enormous stutters leading up to a complete uninstall and reinstall of FSX, I think the AM put in the cfg messes the automated hardcore programmed AM up completely giving this result, therefore this easy and save way, no more, no less and up to people whether to use it or not, like I said in my post, I try to help people with simming problems, not to attack other people on there point of view and not helping somebody in no way so take it or leave it I couldn't care less, this is the last thing I said about this topic

 

[HighBypass 2,339]

I've now unticked AM from the DX10 fixer after finding out that SE has built-in AM for i7 CPUs. Performance is certainly no worse! In fact I may get a little more come to think of it.

Freeware landclass for Japan with a plenty of autogen around Tokyo despite my having set the slider to sparse.Default Cessna on the active at Haneda. Global AI ship traffic meant several large vessels underway around the harbour. I was running FSOpen Clouds for real weather. FPS went from 11 or so up to around 16. Most places I'm smooth with an internally locked 25 fps. I don't have any complex PMDG aircraft.

Win7 64 bit ultimate

i7 950 CPU at 3 and a bit GHz

msi nvidia GTX960 2Gb 

24Gb RAM