Page 2 of 2

Re: Hardware change, p3d

Posted: Mon Sep 21, 2020 3:56 pm
by Martyson
No, my laptop has an i7-7700.

Ok, I got it, I think.

You never tested the i9700 , right?

Re: Hardware change, p3d

Posted: Mon Sep 21, 2020 4:01 pm
by JorgenSA


Re: Hardware change, p3d

Posted: Tue Sep 22, 2020 8:38 am
by KevinKaessmann
Sorry if I was misunderstood, my observations and measurements lead me to the conclusion that one reason for stuttering are the alterning clock frequencies of the different cores, regardless whether desktop or laptop processor.

With default settings, a limited number of cores runs with a higher clock frequency, but as we cannot decide which core runs with what frequency at a specific point in time, it could be the core with the base thread but also one of the rendering thread cores.

In my case, with a four core 88W TDP 4790K, a big cooler and no overclocking, it was easy to fix it by enabling the BIOS "Multicore Enhancement" and increasing the minimum frequency in the windows power scheme or setting it to high performance : Win 7="High Performance", Win 10= "Ultimate Performance". (Details about frequency switching delays e.g. in this document)

The setting of BIOS Multicore Enhancement (ASUS names it "sync all cores") keeps the primary thread running with the same frequency all the time (no frequency reduction for primary thread core when rendering threads start on the other cores).

The setting of the Windows high performance scheme avoids the xx microsecond delay of the rendering thread start (as they run not continuously but "from time to time", say every some seconds, depending on the location and speed of the aircraft). Also, it avoids the frequency jump delays of the first, "free" processor too (affinity mask) if Windows and other tasks want to run (e.g. I/O isn't done by the application but by Windows tasks).

So in my opinion, instead of overclocking one core to the maximum, it may smooth the system by running all cores with a stable constant frequency that doesn't have to change under different load situations.

But there are other reasons for stuttering (GPU limit in cloudy/rainy vs. cavok situations, add-ons with interval activity), so one after the other has to be analysed and eliminated.

Re: Hardware change, p3d

Posted: Tue Sep 22, 2020 9:04 am
by JorgenSA
Thank you for this, this is a very good explanation of a complex topic!


Re: Hardware change, p3d

Posted: Thu Oct 15, 2020 10:46 am
by Blaunarwal
but still these spikes after 20 to 120 seconds
May they result from the Intel Turbo behaviour ?

In my FSX times, I found that one of the reasons for stuttering every some seconds was the "on-off" usage of the rendering threads.

Depending on the scene complexity and aircraft speed,the rendering isn't done continuously but in chunks of multiple seconds up to minutes. In my case, everytime the rendering threads started, the frequency of the P3D main process core dropped and stuttering appeared.

I cured it by enabling the "Multicore Enhancement" in BIOS (4790K: all cores run with maximum turbo frequency 4400 MHz - provided a good cooling) and a specific energy saving profile for FSX/P3D that keeps the processor clocks on a minumum frequency, in my case 2800 MHz.
  • The first leads to rendering threads not disturbing the frequency of the main thread's core (that I isolated from rendering threads by FSX/P3D affinity mask).
  • The second avoids big frequency switching delays of speed step when the rendering threads become active - we talk about milliseconds, not microseconds. This Speed Step/Speed Shift delay hits not only the rendering threads but also - after they're finished - the main thread's core while increasing back its clock frequency.

On the newer Intels, one has to be careful not to burn the processor by overclocking in combination with Multicore Enhancement - running all cores on a very high frequency with very high power limits (= not ending Turbo) requires good cooling solutions...

Maybe it's worth having a look on Windows 10 Processor power management options overview for further optimization ?
Saw your answer to my post pretty late. Thanks a lot for these indeept explanations.
In the mean time I came to the same conclusion about the cpu. But there are also problems with old scenery files like I found now when adding them to the better working V5. I had stutters right away after adding and could identify the problematic files.
In the meantime I also bought a vaio cooler and replaced my air cooler. It is now certainly no longer a problem of temperature. I read the opposite recommendation of yours regarding Multicore Enhancement, stating it should be off to force all cores to run at full speed. It is an interesting approach to have it enabled and the cpu not always on full OC frequency. I'll give it a try.

Re: Hardware change, p3d

Posted: Fri Oct 16, 2020 7:18 pm
by KevinKaessmann
As far as I know, the recommendation is to disable Multicore Enhancement for extreme overclocking as it runs all your cores with the highest frequency - instead of just one or two cores - and that generates more heat as without MCE, where the Intel defined frequency differences apply (see ... -explained).

BTW an interesting video is embeded in this discussion: ... cement_on/

Another interesting point is, that the Intel default allows the processor to draw maximum energy just for a limited number of seconds,
see ... -watt-cpu/ (PL1, PL2, Tau).
After this time, the processor reduces the boost.

Looking on my old H170 mainboard with the Intel Extreme Tuning Utility (XTU, I use it just for monitoring), I found a value "Maximum Power" and this was set to 4096 Watt (BIOS does this but this value isn't changeable or even shown). This explains, why my 4790K cores run on 4400MHz all the time - it doesn't relatet MCE as MCE simply allows all cores to run with the same maximum frequency at the same time.

In a newer board, the power management related parameters PL1, PL2 and Tau can be set to a non-Intel value.
If these parameters are set to Intel default, even a 10900K will run with maximum turbo frequency just for a minute, then throttle down. This could explain stuttering in intervals - although it's said that many boards increase these values by default.

So to analyse stutter, it needs more than just the simple CPU usage graphs. One has to monitor the individual core clock frequencies in parallel.

In my case, running with the ASUS build-in power setting of 4096 Watt (means: no limit) and MCE activated, my cores can run at 4400 MHz endless. Together with my Windows power management setting that doesn't allow a core to fall below 2800 MHz (avoiding the frequency step-up delay if a thread requests processing power), I get smooth behaviour without stutter (on my old rig).

One should keep in mind that a default Windows computer is designed by Intel and Microsoft to save energy in contradiction to P3D requesting high processing power on all cores all the time.

Re: Hardware change, p3d

Posted: Fri Oct 16, 2020 8:19 pm
by choppa
As a total ignoramus when talking about computer specs...I have a basic question to ask.
My 4 year old system needs upgrading...i5 6600 3.4ghz, Geforce 1080 8GB, with 16 GB RAM.
V5 generally runs pretty smooth....I am not a frame rate chaser, if it looks smooth to my eyes that does me. I have the sliders pretty well to the right, but Dynamic Veg Off...found that to be the main smoothness killer. Plenty of Orbx and Flytampa add ons, plus A2A, PMDG, Majestic and Carenado aircraft.
My question is related to the choice of motherboards and processors.
A number in our flight sim group have steered toward AMD processors and AMD Video Cards, with superb results.
That is my quandary....Intel and Nvidia....or AMD?
I apologise up front if my terminology is off, but hopefully you'll get my drift.

Re: Hardware change, p3d

Posted: Sat Oct 17, 2020 7:44 am
by JorgenSA
Phil, don't worry about your terminology, you're doing very well!

From what I have read in various places, Intel CPUs should be more efficient than the AMD ones, because they process more instructions per clock cycle.

Now, this is what I have read, and this is an ever-changing game between Intel and AMD, so this information may very well be outdated.

Personally, and I stress personally, I will go with Intel and nVidia. Did I say personally? ;-)

And finally allow me a quote: "if it looks smooth to my eyes that does me". Good one, that about sums it all up!


Re: Hardware change, p3d

Posted: Sat Oct 17, 2020 9:01 am
by KevinKaessmann
Just my thoughts:

Intel's next generation "Rocket Lake-S" seem to be available in march next year: ... -q1-20212/
AMD's new generation is expected the next weeks: ... rformance/

It's said thad the new Ryzen 5000 will beat the actual Intel processors,
but Rocket Lake-S may beat Ryzen 5000 as it will get fewer cores with higher frequency.

Regarding AMD, I would wait until Nov/Dec, then check the real benchmarks/experiences,
then decide to go for Ryzen 5000 or or wait until march 2021 for Intel.

I was an Intel fan in the past,
but it seems, the times they are a-changing...