Limiting Framerate kills framerate - why?

[SteveW 1,792]

2 hours ago, joepoway said:

Hi Steve 

Perhaps I misunderstood the two other individuals within this post including Rob’s statement below:

No, in P3D when you switch from TFR = Unlimited (which is essentially operating the FFTF = 0.01) to 40 then you are now operating/using the FFTF = 0.33 (default unless you've modified it) ... with TFR = Unlimited the FFTF value defined in your Prepar3d.cfg is NOT used.  With with TFR = 40 and FFTF 0.33 is essentially saying I want 1/3 of your processing time to render autogen and terrain tiles.  From the numbers you've listed that's exactly what is happening.

 

I will say I have seen as many posts on this forum over the years and on LM forum as well that state otherwise that’s why I’m a bit confused as are others.

However you state unequivocally it works regardless of whether one uses unlimited or locked frames so it must be so end of discussion I presume.

Joe

 

The figures speak for themselves as far as I am concerned, they can be repeated endlessly, so you must make of them what you will and make your own mind up.

[SteveW 1,792]

1 hour ago, Rob Ainscough said:

Yes, default 0.33.

Oops, sorry, static FFTF. 

EDIT: but if FFTF can by dynamic then I would assume TFR could also be dynamic ... I don't know how these values are being modified dynamically since the intent of these values were to be static "during flight" ... if they can be modified via SimConnect or do they require locating offsets in one of LM's EXEs or DLLs? (I can probably discover that for myself, but got too many other things going on right now).

As far as I know (was told by Beau H from LM). TFR = Unlimited setting will ignore any FFTF value.

Cheers, Rob.

EDIT:  If this is working well for others, go for it, I'm not at suggesting this product doesn't work, it just didn't provide the results I wanted to see (my level of expectation).

Got you, thanks Rob. No unfortunately techniques can't be discussed on my part.

I think what Beau was referring to is that to get any mileage from FFTF there needs to be slack in the system, or overhead. And so using Unlimited no VSync will not present that slack. End of story for me.

[SteveW 1,792]

32 minutes ago, joepoway said:

Ed

I also seem to find the Dynamic tool improves my flight experience as well.

Regarding Steve I am also very thankful for his contributions but I felt he assumed I am not well versed in the findings regarding performance, settings, and interactions because I haven’t followed things or haven’t invested the time to do so. Quite the contrary I read virtually everything and am well versed in 5 flight sims as well as optimizations across all of them.

Also I still stand by my statement that there is still debate about how FFTF works in various scenarios even Rob restated his “upstanding for Beau at LM” stating the opposite of Steve’s statement.

I’m not trying to cause problems or state that I know one way another on this issue but there is not consensus as far as I can see. As a retired engineer and flight sim enthusiast I am just trying to fully understand things regarding our wonderful hobby. 

Ironically I believe FFTF does work with a setting of Unlimited at least from what the Dynamic Tool displays and my perception of how the sim performs and Steve’s data!

So Steve I also want to thank you for your ongoing contributions as well as Rob’s!

Joe

I don't pay much attention to that Chinese whispers stuff as it is wholly misinterpreted on a very large scale. We are talking about how DX D3D11 work and how it is implemented by P3D.

The effects of changing FFTF in the cfg or by the dynamic methods alter a value that's going to be there for rendering frames whatever way they are timed out of the GPU. Hopefully that helps with the understanding. FFTF can't work when there's no Peter to rob to pay Paul as someone cleverly put it recently. The graphs show the truth of the matter, and as I said before they can be repeated endlessly.

[SteveW 1,792]

5 hours ago, SteveW said:

Who's talking about:

Unlimited "Unlimited" with VSync=Off in P3D

Limited "Unlimited" with VSync=On in P3D

Limited with Fixed or Locked fps on the slider

Limited fps in NPI with Unlimited VSync=Off in P3D

Limited fps in NPI with Limited "Unlimited" with VSync=On in P3D

etc.

 

 

...this is what I'm talking about...

I see countless conversations about how someone has a set- up and how someone is rushing off to try the same set-up. And, I'm thinking I have no way of interpreting how exactly that set-up is configured.

So I don't share settings because they don't mean much to individual PCs. Instead I give you guys the information and tools to work it all out for yourselves.

[Dirk98 354]

On 8/15/2018 at 8:46 PM, SteveW said:

If we enable HT and don't disable one of the LPs of the main core - that is the core with the main  sim thread - that second LP, perhaps at 20%, means that the other LP at 100% is only around 80%.

Sorry, I can't get my head around this. Could you rephrase or explain it more? How do we want these LPs (that are currently at 80% and 20% load as per above) be loaded instead? What is the inference here?

Thanks.

[Dirk98 354]

On 8/15/2018 at 9:27 PM, American 833 Heavy said:

Pretty insane. On my 1950X without HT the number would have been 65535. But I saw no improvement with AM in the CFG on so I left it out.

You might see improvement by using an AM that actually limits the number of cores used by P3Dv4. So instead of 16 cores per your AM above try engaging 6-8 cores for P3Dv4 only. Also you might want to leave Core 0 for OS only as an experiment (but I think I recall somebody advised otherwise). 

 

[d.tsakiris 418]

On 8/20/2018 at 2:27 AM, Rob Ainscough said:

EDIT: but if FFTF can by dynamic then I would assume TFR could also be dynamic ... I don't know how these values are being modified dynamically since the intent of these values were to be static "during flight" ... if they can be modified via SimConnect or do they require locating offsets in one of LM's EXEs or DLLs?

I used the FSX version of the FSPS add-on, and sure enough, it would switch TFR from x to unlimited as soon as the framerate dropped below a pre-set minimum. Can't speak for the P3D version.

[jabloomf1230 2,445]

On 8/19/2018 at 8:27 PM, Rob Ainscough said:

EDIT: but if FFTF can by dynamic then I would assume TFR could also be dynamic ... I don't know how these values are being modified dynamically since the intent of these values were to be static "during flight" .

I wonder if it isn't necessary to attack TFR directly in order to make it dynamic. It's just as easy to modify other internal sim components  dynamically to achieve a target frame rate. AI aircraft apps like UTL do this as does the FSUIPC traffic limiter feature. They adjust the number of AI aircraft to acheived the TFR. The goes to show you that "intent" has no bearing on any of this. A sharp 3rd party programmer can achieve wonders with a bit of hacking.

[joepoway 241]

On 8/19/2018 at 7:18 PM, SteveW said:

I don't pay much attention to that Chinese whispers stuff as it is wholly misinterpreted on a very large scale. We are talking about how DX D3D11 work and how it is implemented by P3D.

The effects of changing FFTF in the cfg or by the dynamic methods alter a value that's going to be there for rendering frames whatever way they are timed out of the GPU. Hopefully that helps with the understanding. FFTF can't work when there's no Peter to rob to pay Paul as someone cleverly put it recently. The graphs show the truth of the matter, and as I said before they can be repeated endlessly.

Thanks again!

Joe

[SteveW 1,792]

9 hours ago, Dirk98 said:

Sorry, I can't get my head around this. Could you rephrase or explain it more? How do we want these LPs (that are currently at 80% and 20% load as per above) be loaded instead? What is the inference here?

Thanks.

Couldn't be more straightforward. The "TWO" LPs of the CORE are actually each a time slice of the core they are SHARING so it appears as if they are both processed simultaneously.

Edited August 21, 2018 by SteveW

[SteveW 1,792]

.And to cut a long story short - if one of those LPs is at 20% and you want 100% from the other LP, then you had better prevent that 20% from a-happenin'.

[Dirk98 354]

17 hours ago, SteveW said:

.And to cut a long story short - if one of those LPs is at 20% and you want 100% from the other LP, then you had better prevent that 20% from a-happenin'.

Ah! Ok, got it ))

Thx