Board index FlightGear Development Aircraft

Su-15

Questions and discussion about creating aircraft. Flight dynamics, 3d models, cockpits, systems, animation, textures.

Re: Su-15

Postby abassign » Mon Nov 09, 2015 7:54 pm

The performance of the SU15, are definitely not different from the performance I get from F14 and F15. With my quad-core (but only two CPUs are used by FGFS) and the NVIDIA GeForce GTX 870M with 6 GB of video RAM, the performance remain fairly stable. I turned on the ALS, with 4 covering trees, and advanced weather; all the effects are configured to the maximum value.
Certainly the amount of polygons is high, but well below the 500,000 above which starts detecting a real degradation in performance (-20%), as I could check in earlier tests. It is likely that under-performing graphics cards there may be a slowdown much more noticeable, but still in line with F14 and F15.
This beautiful work of Vitos shows that it is possible to increase the quality of aircraft for Fgfs without losing too much in the quality of the simulation. Instead I have always noticed a big problem for the trees, which are, in my configuration, extremely costly, much more than I expected.

Image
Developer of the program https://wiki.flightgear.org/Julia_photoscenery_generator
FDM developer of the G91R1B aircraft https://wiki.flightgear.org/FIAT_G91R1B
JSBSim collaborator
abassign
 
Posts: 947
Joined: Mon Feb 27, 2012 6:09 pm
Location: Italy (living 5 Km from airport LIME)
Callsign: I-BASSY
Version: 2020.4
OS: Ubuntu 20.10

Re: Su-15

Postby MIG29pilot » Mon Nov 09, 2015 8:10 pm

[gasping for breath]Did you say 59 fps?[/gasping ofr breath]
User avatar
MIG29pilot
 
Posts: 1465
Joined: Tue May 19, 2015 5:03 pm
Location: 6 feet under Snow
Callsign: MIG29pilot
Version: 2020.1.3
OS: Windows 10

Re: Su-15

Postby dg-505 » Mon Nov 09, 2015 8:13 pm

Well, with a GTX 870 and 6 GB VRAM that's not surprising... :roll: But the price of such a machine is breathtaking :shock:
A mile of road will take you a mile, but a mile of runway will take you everywhere.

DHC-6 Twin Otter development
User avatar
dg-505
 
Posts: 677
Joined: Mon Jun 02, 2014 3:36 pm
Location: Bonn, Germany
Callsign: dg-505
Version: next
OS: Ubuntu 20.04.4 LTS

Re: Su-15

Postby MIG29pilot » Mon Nov 09, 2015 8:15 pm

Ten dollars per frame per second, I believe.
User avatar
MIG29pilot
 
Posts: 1465
Joined: Tue May 19, 2015 5:03 pm
Location: 6 feet under Snow
Callsign: MIG29pilot
Version: 2020.1.3
OS: Windows 10

Re: Su-15

Postby Thorsten » Mon Nov 09, 2015 8:16 pm

It is likely that under-performing graphics cards there may be a slowdown much more noticeable, but still in line with F14 and F15.


It beats me why you would say that, as

a) your GPU is sort of the maximum money can buy, so it seems a bit odd to call everything else 'under-performing' and
b) we've already established that for most people the difference between the Su-15 is not in line with the F-14 and F-15, which you could have known at the simple expense of reading other posts.

Oh, and your screenshots still don't show.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Su-15

Postby dg-505 » Mon Nov 09, 2015 8:17 pm

Make 20-30 Dollar, that matches more likely
A mile of road will take you a mile, but a mile of runway will take you everywhere.

DHC-6 Twin Otter development
User avatar
dg-505
 
Posts: 677
Joined: Mon Jun 02, 2014 3:36 pm
Location: Bonn, Germany
Callsign: dg-505
Version: next
OS: Ubuntu 20.04.4 LTS

Re: Su-15

Postby Hooray » Mon Nov 09, 2015 8:18 pm

It is likely that under-performing graphics cards there may be a slowdown much more noticeable

exactly, which is why some of our longest-standing issues went unnoticed for several months, and even years - including a number of segfaults, race conditions, memory leaks, leaking listeners, and redundant callbacks.

Without using some fairly sophisticated tools and techniques, you can no longer identify certain bottlenecks as long as your hardware is simply powerful enough - which applies usually to the Nasal GC, but also other subsystems containing problematic code
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

Re: Su-15

Postby MIG29pilot » Mon Nov 09, 2015 8:41 pm

dg-505 wrote in Mon Nov 09, 2015 8:17 pm:Make 20-30 Dollar, that matches more likely

AAAAHHHH!!! :shock:
User avatar
MIG29pilot
 
Posts: 1465
Joined: Tue May 19, 2015 5:03 pm
Location: 6 feet under Snow
Callsign: MIG29pilot
Version: 2020.1.3
OS: Windows 10

Re: Su-15

Postby Hooray » Tue Nov 10, 2015 8:02 am

the NVIDIA GeForce GTX 870M with 6 GB of video RAM [...] Instead I have always noticed a big problem for the trees, which are, in my configuration, extremely costly, much more than I expected.


with that sort of hardware, you may want to open a new thread and post your results with different startup/runtime configurations - i.e. regardless of the location/aircraft you are using, your graphics card is definitely very capable to deal with shaders.
So I would suggest that you try different benchmarks - ideally, without using any complex location/aircraft at all, and then post the results using different shader/detail levels - possibly in conjunction with the osg-on-screen stats.
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

Re: Su-15

Postby abassign » Wed Nov 11, 2015 12:49 am

@Hooray

you may want to open a new thread and post your results with different startup/runtime configurations - i.e. regardless of the location/aircraft you are using, your graphics card is definitely very capable to deal with shaders.


I never expected to generate so much hilarity with my test! The GTX 870 graphics card with 6 GB VRAM currently costs about $ 300 is inserted into a laptop PC quad-processor and operating system Linux (Kubuntu) undertaken by a small Italian company named Santech that is specialized in making high-performance laptops, the portable finished costs about $ 1,200. Currently not sold more than the GTX 870 card but in its place is sold the GTX 970 that is at least 20% faster with 8 GB of VRAM, the price is the same +/-. But in the end the graphics card is not used in the best way. I tried to install the same program on another PC desktop, always with Kubuntu, but this time with much lower performance: GeForce GT640 2GB VRAM (cost $ 80) is a dual-processor machine with 8 GB of RAM.
On paper, the speed of the desktop used for testing should be at least 4-5 times less ... Surprise! the speed instead it is about 50% of the speed I get with my laptop !!

Image

This means that there are big bottlenecks! As the saying Hooray, the workload of the GPU is not likely compensated by execution of the program by the processor core.
My considerations are only hypotheses, I did not do more precise work. I then noticed that when active parameter <multitheading-mode> system improves performance by up to 30% with a workload of the two CPUs of 120-150%.

Therefore there is not much to laugh about, the GTX870 with 6GB of RAM is good, but can not express all its possibilities. In other words: 50% of the value of the GPU are wasted money ;)

For many months, I think the problem, at least for my PC, is related to the low CPU usage (for the two machines I've tested, it never exceeds 30% of that available)!
Developer of the program https://wiki.flightgear.org/Julia_photoscenery_generator
FDM developer of the G91R1B aircraft https://wiki.flightgear.org/FIAT_G91R1B
JSBSim collaborator
abassign
 
Posts: 947
Joined: Mon Feb 27, 2012 6:09 pm
Location: Italy (living 5 Km from airport LIME)
Callsign: I-BASSY
Version: 2020.4
OS: Ubuntu 20.10

Re: Su-15

Postby Hooray » Wed Nov 11, 2015 8:52 am

it is like Thorsten said, you simply have an extremely powerful graphics card (and system overall), FlightGear is not particularly good at leveraging all this power currently - depending on the settings you are using, you may see that FlightGear is GPU limited or CPU limited - ironically, on a system like yours, many other (deeper) problems (such as the Nasal GC scheme causing stuttering, RAM and listeners leaking) may never show at all.

However, with your sort of hardware, you would be in an excellent position to help benchmark/profile various aspects of FG to determine bottlenecks in different systems.

Overall, I would reiterate my previous suggestion to get involved in similar benchmarking efforts:
Subject: Polygon Limits
You may want to get in touch with hamzaalloush, who's been working on similar tests:
Subject: How to tell if you are CPU or GPU limited (split)
Image
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

Re: Su-15

Postby Hooray » Wed Nov 11, 2015 11:31 am

ok, somebody was so kind to send me a link to vitos' most recent zip file - I was surprised seeing that it was ~350 mb in size.
I still downloaded the whole thing to look at the Nasal/Canvas code implementing the radar.
To be honest, much of the code at the beginning is looking pretty good to me, and pretty much in line with the docs on the wiki (if not to say copy/paste "in line") - where things are breaking down quite terribly is the whole update logic, like vitos mentioned previously: he didn't figure out a way to address pre-existing canvas elements/nodes and selectively create/update and remove those. This is exactly where MapStructure, or even just its underlying concepts, could have helped tremendously.
That said, the other issue is massively complex nested/conditional (if/else) blocks, including the corresponding settimer() logic to update this at 0.1 second intervals.
Overall, the update() function is over 300 LOC in size - i.e. 60% of the whole file.

Anybody interested in the nitty-gritty details can take a look at: http://codepad.org/C8IfU4ZL

While I have personally found my interactions with 5H1N0B1 (and his Canvas/radar effort) much more fruitful, I would like to re-iterate my offer to vitos, to follow up on my previous suggestions so that we can work out a way to optimize his radar related code, and maybe even integrate/unify it with the code that 5H1N0B1 has been working on for his mirage2000-5 - should vitos be interested in pursuing this, he would not even have to deal with me directly.

In summary, the code looks like a lot of work, because vitos apparently managed to come up with a working piece of code despite using tons of difficult workarounds because he was obviously missing some Nasal/Canvas related context - it kinda reminds me of the original 777/EFB code to be honest.
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

Re: Su-15

Postby Müller » Wed Nov 11, 2015 3:31 pm

I currently have on my computer a graphics card HD 6850 and runs FlighGear well.
A week ago I buy a graphics card GTX970 (not yet installed on my computer), I agree that to run the FG is an exaggeration, currently has no need of this graphics card, others with a much lower cost will run normally FG .
I do not see as wasted money, will depend on what kind of games and quality graphics that you want to run on your computer.. :D
I7-8700k - MSi Z370 Plus Sli - Galax RTX 2070 - G.Skill 16gb 3000mhz - SSD MX 500 -500gb, Hd 1tb Seagate, HD 2tb Wd - WC Corsair H100iv2 - Corsair RM 650 - Monitor 24'' Asus - Gab.Aerocool 800
Müller
 
Posts: 411
Joined: Mon Feb 20, 2012 1:37 am

Re: Su-15

Postby Thorsten » Wed Nov 11, 2015 3:44 pm

For the record - I don't think it's problematic to develop for high end computers

I think what is problematic is

a) if it's not optional and you're in essence asking everyone else to upgrade - which is why the core team is very conservative about non-optional features

b) if it's needlessly so and you could make it much faster by just optimizing a bit

I think the Su-15 falls under b) - it could run with the same visuals for more users if it were just optimized for performance.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Su-15

Postby Hooray » Wed Nov 11, 2015 3:53 pm

those people with very powerful hardware who are still getting only ~30 fps may want to hide the aircraft using /sim/rendering/draw-masks to see for themselves how much of an impact the 3D model/texturing vs. scenery/clouds etc has, note that this will not have any effect on running Nasal code
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

PreviousNext

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 23 guests