@Rob_Ainscough, many thanks for that Rob. Much to digest and far more than I imagined.

That's a 3.5hr video. Are there any time points that would help?

@sonny147

Hi

my MB is a  ROG_STRIX_Z390-F_GAMING

should i use xmp1 or xmp2.

Ram: RAM HyperX Predator DDR4-3200
32GB K2 SW

 

Thanks Frank from EDDM

 

@Pete Dowson posted me a couple of links to YT videos and I watched the short one which described how to install and run AI Suite III.

I bit the bullet and ran it using the 5 step option so it optimised fans, memory and GPU as well as the CPU. I never saw any BSODs or anything untoward and the process completed in around 10 mins.

I setup a test at EGLL slewed 500ft above T5 looking east towards the whole airport and London in the distance. No Ai were loaded and I used Cold Front weather for consistency. VSync was turned off and I set my display to 60Hz.

The first test was with my current settings - 4.7Ghz and HT on. fps were around 50-51

Next test was same clock speed but with HT off. fps went to 54-55.

The final test was after the AI Suite overclocking. HT off and clock speed in Task Manager showing 4.9 / 5.0. fps = 58-60 which surprised me. I had a chat with Pete and during it I noticed the fps dropped back to mid 50s.

So a modest improvement and on a flight from London City to Heathrow 30fps was maintained except for a few seconds around 2nm out from 27L when scenery or something is being loaded and it dropped briefly to 25 before recovering to 30 for the landing and roll-out.

Looking at my BIOS it shows cores 1-3 at 5.2Ghz and 4-6 at 5.0Ghz. A modest improvement for no personal risk. The software does it all for you.

3 hours ago, Ray Proudfoot said:

@Pete Dowson posted me a couple of links to YT videos and I watched the short one which described how to install and run AI Suite III.

I bit the bullet and ran it using the 5 step option so it optimised fans, memory and GPU as well as the CPU. I never saw any BSODs or anything untoward and the process completed in around 10 mins.

I setup a test at EGLL slewed 500ft above T5 looking east towards the whole airport and London in the distance. No Ai were loaded and I used Cold Front weather for consistency. VSync was turned off and I set my display to 60Hz.

The first test was with my current settings - 4.7Ghz and HT on. fps were around 50-51

Next test was same clock speed but with HT off. fps went to 54-55.

The final test was after the AI Suite overclocking. HT off and clock speed in Task Manager showing 4.9 / 5.0. fps = 58-60 which surprised me. I had a chat with Pete and during it I noticed the fps dropped back to mid 50s.

So a modest improvement and on a flight from London City to Heathrow 30fps was maintained except for a few seconds around 2nm out from 27L when scenery or something is being loaded and it dropped briefly to 25 before recovering to 30 for the landing and roll-out.

Looking at my BIOS it shows cores 1-3 at 5.2Ghz and 4-6 at 5.0Ghz. A modest improvement for no personal risk. The software does it all for you.

I'd be interested to see what core voltage it applied to get your o/c? Glad you found a no risk improvement and my 8700k is safe 😉

2 minutes ago, kevinfirth said:

I'd be interested to see what core voltage it applied to get your o/c? Glad you found a no risk improvement and my 8700k is safe 😉

There’s an advanced mode which I’ll try tomorrow. The easy option today didn’t impact on RAM speed and probably didn’t change voltage which is the same as before. 1.35 I think.

On 7/30/2021 at 4:19 PM, Ray Proudfoot said:

The first test was with my current settings - 4.7Ghz and HT on. fps were around 50-51

Next test was same clock speed but with HT off. fps went to 54-55.

I am assuming your first test was with your original setup with the proper affinity set on the addon exe apps and P3D as you previously showed us, and then when going to HT=Off you reset your affinity on the exe apps and P3D to appropriate values.

Since you can't get something out of nothing you might be wondering where the small fps increase comes from? This is due to the background tasks finishing later and the main task has less to do, similar to reducing your FFTF value where that reduces the scenery load, or reducing the objects loaded.

When doing lots of testing with a glider in the Grand Canyon I could visually see the end of the canyon was building up slower with lower FFTF or less work on the background tasks (less cores allocated). These settings gave me a larger variance in fps overall.

It would be interesting to see how much 'hotter' the system gets with HT enabled.

2 hours ago, SteveW said:

I am assuming your first test was with your original setup with the proper affinity set on the addon exe apps and P3D as you previously showed us, and then when going to HT=Off you reset your affinity on the exe apps and P3D to appropriate values.

Yes, CPU was mildly overclocked on all cores to 4.7. Additional EXEs were shared on two separate cores not used by P3D.

After HT=off I assigned them to the two cores in least use.

2 hours ago, SteveW said:

It would be interesting to see how much 'hotter' the system gets with HT enabled.

I haven’t monitored the temps yet but when P3D is loading there is a clear increase in fan movement. It reduces when the scenario screen is shown. No issues whatsoever with several flights since Friday. Approach to EGLL is clearly better with 30fps maintained bar a couple of seconds for loading something, scenery I assume.

I’ll check the FFTF setting tomorrow.

6 minutes ago, Ray Proudfoot said:

I haven’t monitored the temps yet but when P3D is loading there is a clear increase in fan movement. It reduces when the scenario screen is shown.

With HT enabled, no matter how many Logical Processors you allocate in the P3D AffinityMask, all the LPs are used by the system resources (networking, disk etc.) which are highly threaded, so you should see an increase in heat as the same amount of system work completes in less time than with HT disabled, especially when loading the scenario.

14 minutes ago, Ray Proudfoot said:

I’ll check the FFTF setting tomorrow.

When you use a lower FFTF you should see more fps as the scenery loading is slowed down same with reduced cores or disabled HT. That's why some use the FFTF Dynamic program which you can set up to reduce FFTF when you are low to the ground as you can see less distance to the horizon, and increase the scenery loading when you gain altitude by increasing FFTF. That's only effective when you have throughput in hand as you have to rob Peter to pay Paul and so often systems with heavy display settings won't give results there.

Edited August 2, 20214 yr by SteveW

@SteveW, as I have turned off HT that no longer applies. I’m aware loading will be longer but that doesn’t appear to be noticeable at present.

1 minute ago, SteveW said:

When you use a lower FFTF you should see more fps...

But I have fps limited to 30 via my monitor setting and VSync. Will FFTF make any difference?

That's obvious, so you should rather think of it as higher fps available capacity.

So the point is, if the fps drops below your VSYNC the capacity of more fps would save that dropping so far.

When the sim starts up and loads the scenario that will usually be the most intense amount of work during the use of the simulator. Maybe construct a scenario that will present the most excessive load and use that in your tests.

13 minutes ago, Ray Proudfoot said:

But I have fps limited to 30 via my monitor setting and VSync. Will FFTF make any difference?

If you remember from my first mention of FFTF it was stated as a means to test your system to find where the performance is reduced, rather than use a change of FFTF to run the simulator differently. I mentioned just now, some use it to keep fps high and ignore scenery loading speed or use the FFTF Dynamic program to alter it on the fly.

Reducing scenery load speed will have an effect also to reduce stutter depending on the settings. However, in some circumstances a lower FFTF or slower scenery loading might be seen with some scenery as the distant scene builds up slowly to the horizon.

But the main requirement is to keep fps high enough to stay beyond the VSYNC frequency or the external fps limit value in all cases. Whether that impacts scenery loading too much or not is up to the user.