Using Loading Times to determine Affinity Mask

[SteveW 1,792]

Oh and by the way - all those sims XP as well all respond to the same treatment and care of configuration - just like aircraft.

[westman 257]

i use AM 510 , core 0 and 9 for other apps fine tuned the cpu for 5.1ghz synced all 10cores work really good. tested 1022, 511 and some otherwhat you think

Steve is it some other combos you can think of. I dont follow Rob and Gerards line with different core clocks , the I9 is more tricky to tune then the 8700k.

the loadtimes is relally good it get worse with Ht enabled lose atleast 300mhz and loadtimes 2sec worse

 my 8700K run with HT on same AM as Kevin that one rocks , 8700k with ht off  is no go for me 

 

Edited August 27, 2018 by westman

[SteveW 1,792]

31 minutes ago, J van E said:

So... all this talk about AM is just to get a 6% improvement in... what? Fps? Loading times? Somehow I never understood all this AM talk and this topic isn't making it any easier LOL I just can't understand what I am reading. I am surprised there isn't a utility out there by now that takes care of all this calculating and tests your hardware with a single click. 😉 Imho we shouldn't have to deal with all this stuff... Luckily I mainly use AFS2 now, which doesn't need any of this AM stuff, but recently I installed P3D again (for my ATC controlled flights with real time weather in a study level sim) and AFS2 really spoils the fun of P3D when it comes to performance. Which is why I tried to make sense of AM once more by reading this topic. It's all in vain though. I guess I'm too stupid to get it. 😉 But anyway, now I know you have to go through all of this just for a 6% improvement I think I'll stick with what I've got. Until that utility is there or someone has written a AM-guide for total newbies. 😋

Also - don't forget performance gains are only one thing These guys are interested in maintaining their systems correctly, understanding how to use HT and AMs effectively ensures they can overclock and use addons with confidence. Your post serves no purpose whatsoever other that detract from your friends and collages enjoyment of their systems.. LMAO.

We get more than enough from nay-sayers wasting everyone's time.

[JoeFackel 760]

We could castrate the content of P3D down to ASF2's amount ..... THAT would give us a huge amount of fps 😉

 

[Guest]

38 minutes ago, SteveW said:

Also - don't forget performance gains are only one thing These guys are interested in maintaining their systems correctly, understanding how to use HT and AMs effectively ensures they can overclock and use addons with confidence. Your post serves no purpose whatsoever other that detract from your friends and collages enjoyment of their systems.. LMAO.

We get more than enough from nay-sayers wasting everyone's time.

I think you didn't quite get my point... I wasn't being sarcastic or anything. I was honestly surprised by that 6% and I am honestly lost when it comes to AM. I thought my tongue in cheek remarks were clear but apparently not. I didn't want to waste anyone's time. Sorry you didn't get it. It was actually a cry for help more than anything else. 😉 

27 minutes ago, JoeFackel said:

We could castrate the content of P3D down to ASF2's amount ..... THAT would give us a huge amount of fps 😉

 

I had that coming. 😉 

[newtie 533]

38 minutes ago, J van E said:

It's all in vain though. I guess I'm too stupid to get it.

Hi Jeroen-

No, you're not. It's not that difficult and does result in a better flight experience. LM did something in it's last iteration(s) of P3D where a tested AM does make a difference. I've got the hours in to prove it.

Anyway, forget the graphs, percentages and commentary; its a distraction. After a long testing period on the latest P3D 4.3, and AM of 213 and my other addons on the open cores via batch file. I'll give you a few simple resources to work with to accomplish this.

One last note that I'm sure you're aware of. There isn't a standard PC platform out in the ether. There are going to be a lot different variables associated with each machine. Those variables change depending on where the aircraft is located and how it's being used. My point is that even after testing an AM at the same location, airport, settings, as I did, it's going to behave differently at alternate locations. My AM results in the best smoothness at the places I fly with the settings I use. Here you go..

For AM configuration:

http://www.gatwick-fsg.org.uk/affinitymask.aspx?SubMenuItem=utilties

or

https://www.rapidtables.com/convert/number/binary-to-decimal.html?x=01010101

For batch creation:

https://www.sevenforums.com/tutorials/83632-processor-affinity-set-applications.html

Hex converter for batch file creation:

https://www.rapidtables.com/convert/number/binary-to-hex.html

Have at it!

Cheers,

Mark

 

 

 

[JoeFackel 760]

25 minutes ago, newtie said:

For AM configuration:

I really like THAT, even dorks like me understands this one: 😉

http://store.fsxtimes.com/cal-cam.php

 

Edited August 27, 2018 by JoeFackel

[newtie 533]

9 minutes ago, JoeFackel said:

I really like THAT one, even dorks like me understands this one:

http://store.fsxtimes.com/cal-cam.php

 

🤣🤣🤣

[Cruachan 705]

2 hours ago, J van E said:

Somehow I never understood all this AM talk and this topic isn't making it any easier.

Hi Jeroen,

Believe me, I too shared your pain until, that is, the arrival of the moment when something clicked and the journey towards enlightenment began!

I suggest you take another look at my first post and also Supplemental (1) a few posts below that. There is enough info there to get you started. Three things to keep in mind:

1. When displaying Logical Processors (2 per Core when Hyperthreading is enabled) start reading from the RIGHT. The first Logical Processor is Core 0 containing LP0 and LP1, Core 1 contains LP2 and LP3....and so on, moving along leftwards from core to core. For example, in a 4-Cored CPU the last Logical Processor in Core 3 is LP7. 

2. An Affinity Mask added to Prepar3D.cfg is telling the sim which Cores or Logical Processors are to be assigned for the exclusive use of Prepar3D. This is represented in Binary Code where each 1 (masked/assigned) and 0 (unmasked/unassigned) forms part of that Affinity Mask. Again, Binary Code is written from right to left. This Affinity Mask (Binary) must be converted to Decimal before it can be employed in Prepar3D.cfg.

3. When viewing Core or Logical Processor Activity in Windows Task Manager you start from the LEFT (Windows 7) or TOP LEFT (Windows 10) with Core 0 or LP0.

Hope this helps to whet your appetite for more!

I realise you are probably familiar with at least some of these basic concepts, but there may be others coming to this afresh who might derive some benefit.

Regards,

Mike

 

 

 

 

[Guest]

3 hours ago, newtie said:

Anyway, forget the graphs, percentages and commentary; its a distraction. After a long testing period on the latest P3D 4.3, and AM of 213 and my other addons on the open cores via batch file. I'll give you a few simple resources to work with to accomplish this.

Thanks, I'll have a look at those!

2 hours ago, JoeFackel said:

I really like THAT, even dorks like me understands this one: 😉

http://store.fsxtimes.com/cal-cam.php

Wow, that is a VERY easy calculator! 🙂

2 hours ago, Cruachan said:

Believe me, I too shared your pain until, that is, the arrival of the moment when something clicked and the journey towards enlightenment began! 

Thanks for your explanation! Yes, I had the same with understanding VOR: it never clicked. 😉 But suddenly it did and it no time I was writing my own VOR tutorials LOL I am looking forward to my AM-click. 😉

Concerning your test: I should note how much time it takes for P3D to go from start to showing the loading screen...? Why not note the time it takes to end up on the cockpit? And btw there is no real need to check the CPU activity in this case because all that seems to matter is the time? I also wonder why I should disable all addons... If I disable them all and find the best AM, do I still have to figure out which cores to use for addons...? Maybe I will just start with testing loading times. 😉

EDIT
Anyone knows how to temporay disable UTL...?

EDIT 2
Okay, so (for starters) I simply noted the time it took for P3D to go from start up to being in the cockpit (with UTL still enabled, nothing else). I have an i7 4790K so 4 cores and HT enabled. Here are my results:

AM 1 (00000001) = 68 seconds
AM 5 (00000101) = 58 seconds
AM 21 (00010101) = 50 seconds
AM 85 (01010101) = 41 seconds
AM 93 (01011101) = 40 seconds
AM 125 (01111101) = 40 seconds
AM 253 (11111101) = 37 seconds

At first I was seriously surprised to see how much time is being saved with AM 253 compared to AM 1! But then I thought: what are the times without AM...? So I tested that and it was 37 seconds... (and a few hundreds less that AM 253).

In short: I get the best loading times without an AM. So er... what should I conclude from all this so far...?

Edited August 27, 2018 by Guest

[GSalden 4,656]

Without AM you will have RAW power for max fps . However the fluctuation in min - max fps is larger than with the correct AM.  And there might be more long frames.

[newtie 533]

2 hours ago, J van E said:

In short: I get the best loading times without an AM. So er... what should I conclude from all this so far...?

Forget the loading times. As I've already said, you're going to have to put in some work. Just because P3D loads fast, does not mean or guarantee a smooth flight experience. I went from no AM, to 85,116 and 245; those are the ones I remember 😀 For some unexplained reason* for what I run and how I run it, 213 hit it. I spent the time to go back and check again. 

Good luck!

Mark

*Actually it's quite explainable. The AM I'm using matches the data flow I need to support the experience I want.

[Guest]

35 minutes ago, newtie said:

For some unexplained reason* for what I run and how I run it, 213 hit it. I spent the time to go back and check again. 

Okay... so it's 'just' a matter of using each setting for a while and pick the one that seesms (subjectively) best... What's the use then of all this testing, I wonder.

53 minutes ago, GSalden said:

Without AM you will have RAW power for max fps . However the fluctuation in min - max fps is larger than with the correct AM.  And there might be more long frames. 

Ah, ok, understood.

Edited August 27, 2018 by Guest

[kevinfirth 1,393]

51 minutes ago, GSalden said:

Without AM you will have RAW power for max fps . However the fluctuation in min - max fps is larger than with the correct AM.  And there might be more long frames.

Noooooooo!

No AM in the above case with a 4790 means running with an AM effectively of 11,11,11,11,11,11,11,11 - this means the main sim thread will HAVE to share a physical core with another thread.  The result to some extent then HAS to be that the main rendering thread will be impeded to a degree, which means reduced performance.

AM 253 may be marginally slower at loading ( but tell me you'll really notice a few milliseconds!!), whereas I bet a significant amount it will perform noticeably better in sim....

[newtie 533]

18 minutes ago, J van E said:

pick the one that seesms (subjectively) best

Not subjective, is. You will know and feel it when you see it.

I have been flying for quite some time now with 213, unlimited w/vsync and a few other small adjustments. Every once in a while, I get the compulsion that there is something better than perfect. That was ~ a week ago.

Back to perfect. I still can't believe how well P3D runs, looks and feels.

I just fly these days which is great fun, and a relief all at the same time..

Cheers,

Mark