kolaf

After some more googling and reading (apparently you have never read enough) I realised that it might be my docked 2D panels placed on a fourth monitor that are causing the problems. Sure enough, starting the simulator without the panels undocked (I had a saved flight with the undocked panels so I never saw the change myself) I immediately had much better performance. It appears that undocking even a single panel to a secondary monitor significantly reduces my simulator performance. Both CPU and GPU usage is drastically reduced, leading to the symptoms I described above. The weird thing was that removing the undocked panel did not restore the performance. I had to restart the simulator to achieve the performance I had before undocking the panel. I found a thread on the p3d for from 2013 that discussed this issue, but it was also claimed to be solved in version 2.1. Maybe it has been reintroduced? Anyway, I will head over to that forum to describe measures and see what the Masters say. A link to the thread discussing and apparently fixing the issue: http://www.prepar3d.com/forum/viewtopic.php?f=6312&t=9630&start=30

Hi guys, I thought I was pretty good at configuring p3d, but I'm stumped getting it to work smoothly on my new system. Maybe I'm expecting too much, in that case I am sure you will let me know ;-). My system is a new build with a [email protected] GHz, 32 GB RAM at 2400 GHz,a MSI GTX 9700, a few SSDs, and so on. I have a completely fresh installation of p3d, and the only tweak I am running is affinityMask = 85 since I am running with HT on. The scenario I am testing is pmdg 737 NGX at aerosoft ENGM with FTX Norway. I have connected three screens to my main GPU and I have dragged the window to fill the three screens using Nvidia surround (using the new view groups functionality in 3.3 seems to have an even greater hit on performance than simply stretching the screen, even though it looks much better). The total screen resolution is thus 5900*1080. I should note that reducing the size of the window does not significantly affect the frame rate. The CPU utilisation remains more or less the same, while the GPU utilisation decreases drastically. This probably indicates a CPU bottleneck, but I will get back to that further down. The settings are not too demanding, I would think. I think I have a lot to go on on GPU, but I need to be careful with CPU management. Level of detail range is high, all scenery complexity and density is "dense", special effects are medium both, fxaa = off,msaa = 8, anisotropic 16x, HDR off, and some cloud shadows. Full settings can be viewed in this album: https://goo.gl/photos/8S36VhdrnrHUUaZP8 Running this on a rainy day I get between 14 and 20 FPS on the ground. This would have been fine if the system was working hard to give me this. However, the GPU load is less than 50%, and the load on the main CPU core is around 80%, while the other three (logical cores) that are assigned to p3d hover around 20 to 30% (presumably because I am not moving so little texture loading is required). A snapshot of these numbers are also visible in the album above. So, what to do. Why is not the CPU working full tilt? If both the CPU and GPU are running at less than 100%, then what is the bottleneck? I am assuming it is probably still the CPU, so what are the most CPU intensive settings? Has anyone ever produced an overview of the CPU and GPU impact of the different settings? This would be quite invaluable when trying to tune the performance of a specific system. I'm sorry that my questions are all over the place, but as I said, I'm not entirely sure how to proceed. Obviously I have to reduce some settings, but I don't want to do that :-). In the worst-case, though, which settings should go first? Thanks for listening.

For reference, this is what I ordered a few weeks ago: https://www.komplett.no/wishlists/70547e42-17cd-4cdd-b252-a351d0c625b3 (Or at least close to it, 6700k instead of 6700). I have been running it for about a week and I am quite satisfied. Just overclocked it to 4.5 yesterday without issue, running between 60 and 70° with prime95. I have still to find the correct settings balance since still seem to be more CPU bound than I expected.

If I may follow up on my own post: I completed setting up my three monitors using Nvidia surround (view groups had too much of a performance hit for me). It turns out that I am still very CPU bound, and even though I can turn up stuff like AA and shadows (and still only utilise 50-60% of my 1070), I still have to be quite restrictive on scenery detail. I need to limit my scenery to dense on all three sliders in order to remain mostly locked at 30 FPS with the 737ngx. Time for some overclocking... Anyway, I'm guessing that any 1080 system will be similarly CPU limited for large displays without a significant overclocking.

Brilliant. I am in the process of building a similar machine, only with a 1070 instead of a 1080. I am also moving from one to 3 screens, which should put a significantly heavier load on the GPU. Your report makes me confident that my new build should give me similar performance to my old build (2600K@4GHz, gtx 780) despite the increased screen real estate.

FWIW: http://www.3dmark.com/3dm11/11341974

Thanks for the clarification. That is a pretty inconvenient bug. Based on this, and the fact that I have had to clock down my CPU lately because it has been getting warmer, it looks like I will pull the trigger and go for a 6700 K.

Sorry for hijacking this thread, but I'm confused. I'm considering upgrading from an [email protected] GHz to a 6700k or 6800k and clocking it as far as it can go on a decent cooler. Looking at the CPUboss page it looks like the single core performance of the 2600 and 6700 is identical, even though the clock speed of the latter is 0.6 GHz higher. That can't be right? Has the really been no improvement in single core performance for the past four years?

Thanks, this helped me figure out how to do it with Linda. I have now set up two user functions such as this one to toggle the oxygen on and off :-) function Oxygen_on () ipc.writeLvar("RASOxySwitch", 1) DspShow ("Oxygen", "on") end

Thanks for the information, it should save me some trouble when you release the update :-)

A good read. This explains why I cannot get the reboot editionclick to work on my brand-new P3D v3 installation. I keep mashing the auto feather arm command, but nothing happens. The crazy thing is that the installer seems to work partially since it managed to update the correct panel config files.

Nevermind, all my problems were caused by other issues. The stutters turned out to be caused by an old FSUIPC configuration file. Letting it generate a new one solved the stutters. The crash seems to have been caused by the CPU overheating, which apparently was not triggered by the simpler default aircraft. Reducing the amount of over clock down to 4 GHz (a relatively old CPU) seems to have solved the problem and I'm now 15 minutes into my first flight :-). Unfortunately the FPS is a bit worse, but I think it is worth it :-). Feel free to delete my thread if this is not useful for anyone.

Hi guys, I have a fresh Windows 10 installation, with a brand-new Prepar3d V3 installation, together with the PMDG 737 NGX. I've installed one of aircraft from A2A as well as the global base package from ORBX. With the default aircraft and the other add-on aircraft the Sim is smooth as butter. However, as soon as I load the NGX, I get significant stutters. The simulation is smooth for 10 seconds, and then it freezes for a couple of seconds where the CPU utilisation drops to 0. The smoothness then returns for another 10 seconds, followed by another freeze, and so on. During my last attempt the whole thing ended in a bluescreen. Windows 10 spent some time recording data before rebooting, but I do not know at the moment where this data is located for me to check it out. I'm wondering if anyone else is fighting the same issue?

Hi, I'm wondering if it is possible to have the FO react to my voice commands while on the ground. I'm thinking about commands to set up the panels, e.g. set speed, set altitude, heading, et cetera. It appears that these commands are not working until airborne. I know it is quite possible to tune these dials myself while on the ground, but am suffering from tendinitis so I am trying to avoid as much as possible fiddling with the mouse. Thanks.

@ohsirus: I just read your build thread with great interest. It fired off a bunch of questions that hopefully you or someone else could answer. I assume you use something like proSim for the MIP displays, or do you something else? In which case, is it sufficient just to get the display module, or do you have to get the entire pack? I'm very impressed by your diligence and ability to build your entire system without the ability to use it and play with it in the process. I'm definitely not like that, so I'm thinking of a plan looking something like this: 1. Get the MCP and EFIS combination and have a blast. 2. Use my existing saitek yoke, pedals, and two throttle quadrants to get the basic functionality running. I guess that using the six levers and 12 buttons should be sufficient to replicate most of the functionality of the 737 throttle quad. I have a home-made panel with several switches for lights and stuff based on an Arduino Mega which I have interfaced through FSUIPC. This can be reworked to cover many of the small elements required for the MIP as I see the need for them. 3. Use my existing 14 inch touchscreen for CDU and maybe lower centre screen (but perhaps it is not big enough for both?) 4. buy some cheap screens off of eBay together with a cheap computer to run all the instrument displays and build this into a simple wooden rack. Power these using something like prosim. 5. Perhaps buy a hardware CDU 6. Expand with a front overhead panel (, and maybe an aft panel) 7. At some point add the rest as time, interest, and money allows. Does this sound like a reasonable plan? Is there anything I have forgotten, or should the order of some of the points perhaps be altered? One of the big selling points of this approach is to gradually grow into the hobby in terms of time and money spent, allowing me to tune the direction and investments according to my interests. Makes sense? Is there perhaps a better place to post questions like these? Thanks.

From what I can see it should be possible to get it working worth the iFly without much hassle. The next question, is the fun worth 460 euros?

If you want to go the Arduino route, incremental inputs are simply potmeters they create a variable resistance for an analog input in the microcontroller. I've used this to build for instance a Lego flap lever connected to an Arduino and interfaced via FSUIPC to P3D. You could easily build a throttle using this, but with my skill set I would be more worried about getting the mechanics right than dealing with the inputs which is relatively easy ;-)

Thanks, and they are not too expensive, either. What you think about opencockpits? They have a decent offer for the MCP and EFIS combo: http://www.opencockpits.com/catalog/mcpv3efis-christmas-offer-p-333.html?cPath=68

Hi guys, I'm wondering if I should increase the immersion factor for my 737 simming experience and considering purchasing a 737 MCP and possibly also an EFIS panel. I'm thinking that these two other ones I would get the most joy out of having in real hardware. I already have a touchscreen which I placed the FMC on, so there is no rush to get a hardware FMC. My very long-term plan is perhaps to slowly build things out to a full cockpit, but there is no rush. It is therefore of some importance that the hardware I purchase can work well in a full set up. I've looked around a bit and see that there are a number of suppliers of these two panels, and am having trouble deciding which I should choose. I use P3D v2.4 and the iFLY 737 NG, but I also fly some GA and it would be nice if I could use the MCP panel to control the autopilot (even though this is entirely unrealistic, I know). It is therefore critical that the hard work and interface with p3d and ifly, and it is a bonus if it relies on software we can configure heavily ourselves, e.g. FSUIPC., Instead of proprietary software. Still, it should preferably work out of the box without too much configuration fiddling. Price is of course an issue, but I'm not expecting it to be cheap, either :-). I really hope you guys have some good suggestions. Thanks, and happy New Year.

I don't know, perhaps I'm just naive in expecting that my hardware would perform better than what I am seeing :-)

I dropped autogen to sparse and scenery complexity to dense, turned off all shadows except inside the cockpit and on the aircraft, and I still am not able to maintain 30 FPS. I have always thought but my platform handled relatively high settings quite well using simple aircraft, but I'm a bit surprised by the significant impact the iFLY 737 has on system performance.

Thanks guys. I am well aware of the need to compromise, I just didn't think it applied to me I guess my main question was what can I turn down which has the least impact on my experience and the highest impact on performance? I will try the suggested scenery objects shadows (I already have tessellation at high, not maximum, but I guess I could turn it down even further for IFR). With this and tweaking the autogen settings even lower I guess I should be able to create an IFR profile as mentioned above. Still, I want more eye candy in the VFR settings with better frame rate than I have now 0:-).

Hi guys, I'm running P3D 2.4 on a Intel 2700@ 4.5 GHz with a GTX 780 GPU. This allows me to have autogen density at thence, scenery complexity at very dense, cloud coverage at maximum at a distance of 90nm, shadows at medium with default values, and all the other sliders relatively high (except traffic which is around 15%). Running on a 1080p monitor this allows me to look the frame rate relatively steady at 33 using the A2A C182 in FTX Norway in relatively cloudy weather. When I approach Oslo performance drops to around 20-25. If I uncap the frames (using the internal limiter) it jumps somewhere between 30 and 60 most of the time, but the simulation becomes a bit more stuttery. If I'm running a more heavy add-on aircraft such as the iFLY 737 or MJC D8 I'm never able to maintain 33, it goes between 15 and 25 depending on how much ground and sky I see. I've attempted to improve performance for the complex aircraft both by reducing autogen by a notch and reducing shadow quality, but nothing seems to have any significant impact. Which settings would be most beneficial for me to tweak to improve performance for tube liners? Also, according to what I'm reading in the forums I feel that I should be able to have better settings with my hardware with better performance. Thanks.

Hi guys, I was very lucky and got a leap motion for Christmas. I have been dying to try it out on my flight simulator, but am having some difficulties with figuring out the best way to utilise it and I was hoping that someone he has some experience they would like to share. My first impulse was to use is as a simple mouse emulator. The concept is quite simple, you move your hand around over the sensor to control the mouse pointer on the screen. This approach has two drawbacks: 1. it is a bit imprecise and it quickly gets tiring to hold your hand in the air while figuring out the pointing. 2. The mouse emulation did not seem to work well with my P3D 2.4 installation, it refused to register any of my virtual mouse clicks. Since this doesn't seem to work very well I'm thinking of two alternative approaches. The first is also quite intuitive, having the leap motion turn my game screen into a touchscreen by detecting when I place my finger on/near the screen and simulating a click. You could for instance use one finger for left click and two fingers for right click. However, I have not found any ready-made application to support this concept which I find a bit weird. My third idea is a bit more flight simulator specific. Instead of using the screen as a touchscreen you could use any kind of flat area as a touch interface. I could for instance printout the overhead panel of a 737 and place the leap motion at the bottom of this. I could then map the depicted switch positions on the panel to float similar to controls, probably through FSUIPC. I could support new planes simply printing out new control interfaces and calibrating them. I have no idea how easy this would be, but assuming that their SDK is relatively easy to use, it should be no more difficult to do this than to do regular mouse emulation. The only additional step would be the FS UIPC which shouldn't be difficult. I'm writing this post in the hope that maybe some of you have some experience/ideas/no of software that could be useful for integrating the motion with a flight simulator, either FSX or P3D. Merry Christmas.

I'm using a relatively old 14 inch LG monitor, L1510SF, with a USB connection for the touch interface.