2 hours ago, American 833 Heavy said:

Sounds good.

So after a brief test I locked 30 fps with the 30 fps limiter and used an eye candy Corporate Jet addon that is notoriously hard on FPS. That FFTF setting really did the trick with the aforementioned 64 thread AM setting. Think I'll stick with this new config as opposed to unlimited, no FFTF or AM.

 

I'll also have to play around with the AM and see how few threads I can get away with and still lock 30 fps.

 

But the thing with these AMD TR chips is they are primarily geared towards multithreading which is where they shine on CPU intensive tasks. On the 2990WX the single threaded performance is comparable to the 18 Core i9 7980XE, Multithreaded more than double the perf. But due to the architecture of the TR chips they split all that work across the cores. Its really 2 Ryzen CPUs fused into one so it may be beneficial to actually leave all the theads enabled on AM for me. On an i7/i9 its a different ball game. As both these top dog CPUs get the job done but get there in different ways.

 

Also I have always wondered is there a particular part of the P3D cfg that is ideal for placing the Job Sceduler header and the solo AM setting for it? I actually placed it this time right after the MAIN section. I have also put it at the very top and bottom of the cfg in the past when using it as well.

It seems to be working really well after that test so I'll just leave it after the MAIN section I guess....

So long as the cfg section items appear under the correct section header, it does not matter the order of sections.

The effect of HT is that the core is utilised more fully by the same quantity of threads and that those threads perform better because roughly half the context switching time is saved and also the waiting states are utilised by the other LP. Threads are swapped in the core to mimic simultaneous behaviour, but is only time sliced.

Edited August 15, 20187 yr by SteveW

3 hours ago, joepoway said:

Rob

I want to make sure I understand something regarding FFTF and the Dynamic Tool which I use. Here’s the question:

Does the Dynamic FFTF tool work if you run Unlimited or must you set a frame limit for it to work?

I recall from previous posts FFTF settings in the config file seem to only work when not running unlimited.

Please advise, thanks in advance!

Joe

Having same doubts as well since the explanation by the p3d dev is otherwise which is in line with what I recall from fsx, see post #2: http://www.prepar3d.net/forum/viewtopic.php?f=6312&p=139706&t=120133

6 hours ago, joepoway said:

Rob

I want to make sure I understand something regarding FFTF and the Dynamic Tool which I use. Here’s the question:

Does the Dynamic FFTF tool work if you run Unlimited or must you set a frame limit for it to work?

I recall from previous posts FFTF settings in the config file seem to only work when not running unlimited.

Please advise, thanks in advance!

Joe

Works with unlimited as well - it is mentioned in the product description as well

I have them unlimited in the game and limited to 30.5 in NVI only as this gets me the best performance and FFTF does the job right for me

11 hours ago, BrettT said:

...Is nvidia inspector an option and if so does anyone know if this works with a VR headset?

 

 

In native VR everything behaves as you would expect. I don't use flyinside anymore because LM nailed VR in 4.3 the performance is at least as good possibly better in my experience.

VR rift vsync is in multiples of 90/45/30 instead of the usual monitor refresh multiples of 60/30/20. External limiters work but not nvidia inspector unless I was using an old version.

I have used dxtory as the external limiter in VR because it's free, you can adjust it on the fly and I thought it did a slightly better job than NI inspector at holding the software vsync steady anyway.

Thanks Glider!  I appreciate it and will have to give it a look

18 hours ago, Rob Ainscough said:

I installed the Dynamic FFTF product last night and did a quick test flight out of LIPZ with FSL A-320 (latest update) ... came up with some settings that seemed to work well ... good performance and no disappearing AG.  BUT, I really need to do more testing as this is subjective ... unfortunately my two other Test PCs are being used for "something else" so I can only test on my main FS PC.

Cheers, Rob.

Can't wait you share your TFR, VSYNC settings along with Dynamic FFTF settings that seem to work well already.

Thanks!

9 hours ago, Shomron said:

Works with unlimited as well - it is mentioned in the product description as well

I have them unlimited in the game and limited to 30.5 in NVI only as this gets me the best performance and FFTF does the job right for me

I have used the same method and indeed the tool appears to be functioning with P3D being set to Unlimited as we can see the FFTF settings changing with altitude (the method I choose) but I can’t wait for Rob to verify this!

Joe

On 8/16/2018 at 8:47 PM, BrettT said:

Thanks Glider!  I appreciate it and will have to give it a look

I gave the free trial of dxtory a go in native VR again brett. Problem is that it can't hold a software vsync firmly enough so if you set dxtory to 30fps for example but the unlimited fps are really wanting to gravitate to the real hardware vsync of 45fps (90/2), there is an underlying fight going on between dxtory wanting to keep it at 30fps but the sim wanting it at 45fps which reduces smoothness.

I'm finding in native VR that using vsync with tripple buffer unlimited works well enough. The internal vsync keeps frames locked at ratios of 90 so depending on hardware in rural areas it locks to ~45fps (90/2), in regional city it will lock at ~30fps (90/3) and in big cities it locks to ~22.5fps (90/4). The only downside is when the internal vsync can't decide what multiple to lock to which can reduce smoothness. For example if you are in a 45 degree bank in a rural area it will lock to ~45fps for a period through the turn. When orbx scenery comes into view it suddenly locks to ~30fps then back to ~45fps really quickly which reduces some smoothness but it is good enough.

In native VR I tend not to use internal FPS lock because it is too expensive on FPS (it worked better in flyinside) but locking it to vsync/unlimited in native VR with appropriate settings works really well Lockheed Martin really made good improvements to VR performance improvements in 4.3.

Edited August 19, 20187 yr by glider1

On 8/13/2018 at 10:18 PM, Michael Moe said:

I bought Dynamic FFTF and set it to 4500 feet 0,4 and 0feet at 0,01 and the blurry testing KLAX area are gone. I run UNLIMITED@25HZ VSync on

Where to find this one?. It's the FSPS product or a different one?.

Cheers, Ed

 

6 hours ago, edpatino said:

Where to find this one?. It's the FSPS product or a different one?.

Cheers, Ed

 

Same one

14 hours ago, glider1 said:

I gave the free trial of dxtory a go in native VR again brett. Problem is that it can't hold a software vsync firmly enough so if you set dxtory to 30fps for example but the unlimited fps are really wanting to gravitate to the real hardware vsync of 45fps (90/2), there is an underlying fight going on between dxtory wanting to keep it at 30fps but the sim wanting it at 45fps which reduces smoothness.

I'm finding in native VR that using vsync with tripple buffer unlimited works well enough. The internal vsync keeps frames locked at ratios of 90 so depending on hardware in rural areas it locks to ~45fps (90/2), in regional city it will lock at ~30fps (90/3) and in big cities it locks to ~22.5fps (90/4). The only downside is when the internal vsync can't decide what multiple to lock to which can reduce smoothness. For example if you are in a 45 degree bank in a rural area it will lock to ~45fps for a period through the turn. When orbx scenery comes into view it suddenly locks to ~30fps then back to ~45fps really quickly which reduces some smoothness but it is good enough.

In native VR I tend not to use internal FPS lock because it is too expensive on FPS (it worked better in flyinside) but locking it to vsync/unlimited in native VR with appropriate settings works really well Lockheed Martin really made good improvements to VR performance improvements in 4.3.

Regarding the use of Vsync and triple buffering in VR I have a question.

I use an Oculus an Rift and will at times monitor frame rates with the Oculus   debug tool using the heads up display. I have always disabled Vsync and Triple Buffering but do run Unlimited. Even with Vsync disabled I see the same results where the Rift automatically seeks out multiples of 90 as you say with Vsync enabled.

I clearly could be wrong but I was under the impression that when in VR mode at least with my Rift it automatically Vysyncs to the headset desired multiples of 90? I also see the same behavior in other sims and any VR games I play in my Oculus library when I monitor actual frames in the headset using the Oculus debug tool.

Now I do see a difference in performance when I tried to lock frames but not with Vsync or TB. 

Additionally I use the debug tool to set my Pixel Density to 1.7 to improve the clarity. Also I use the Dynamic FFTF tool as discussed in this thread but am still curious why it works with an unlimited setting and I hope Rob will confirm with his testing.

Joe

Joe

On 8/16/2018 at 11:33 AM, joepoway said:

I have used the same method and indeed the tool appears to be functioning with P3D being set to Unlimited as we can see the FFTF settings changing with altitude (the method I choose) but I can’t wait for Rob to verify this!

Joe

I've just bought FFTF Dynamic and it seems to me that frames has to be limited in-sim when using the AGL height method. To my understanding, with FFTF set to 0,01 on the ground (AGL=0) is like your frames are unlimited, but when you're above your AGL max height, and your FFTF gets a higher value (depending on what you set, but usually close to 0,33) it will not matter if you have your frames unlimited, P3D simply will react as if they were limited.

I'm also interested to hear about Rob's tests and further comments.

Cheers, Ed

21 minutes ago, edpatino said:

I've just bought FFTF Dynamic and it seems to me that frames has to be limited in-sim when using the AGL height method. To my understanding, with FFTF set to 0,01 on the ground (AGL=0) is like your frames are unlimited, but when you're above your AGL max height, and your FFTF gets a higher value (depending on what you set, but usually close to 0,33) it will not matter if you have your frames unlimited, P3D simply will react as if they were limited.

I'm also interested to hear about Rob's tests and further comments.

Cheers, Ed

Exactly the benefits from those two worlds. No need to run limited 

Works really well on my system. 

Did a 12hours flight CYVR - VHHH with FT VHHX enabled with PMDG 77W and no issues. Highly recommended piece of software. 

Thanks Michael Moe 

Edited August 19, 20187 yr by Michael Moe

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.

 

 

55 minutes ago, edpatino said:

I've just bought FFTF Dynamic and it seems to me that frames has to be limited in-sim when using the AGL height method. To my understanding, with FFTF set to 0,01 on the ground (AGL=0) is like your frames are unlimited, but when you're above your AGL max height, and your FFTF gets a higher value (depending on what you set, but usually close to 0,33) it will not matter if you have your frames unlimited, P3D simply will react as if they were limited.

I'm also interested to hear about Rob's tests and further comments.

Cheers, Ed

I’m a little confused by your reply Ed.

Are you saying you think it does work even when frames are set to unlimited?

My belief is it does function even with frames set to unlimited which is not the case when you don’t use this dynamic tool.

Joe