Board index FlightGear Development Effects and shaders

Landmass problem

An exciting "new" option in FlightGear, that includes reflections, lightmaps, the particle system etc.. A lot is yet to be discovered/implemented!

Re: Landmass problem

Postby bugman » Wed Aug 16, 2017 9:55 am

I think abassign is trying to say that we should have specific hacks in FlightGear to work around short-term graphics card driver bugs. On the other side however, this is a nasty, non-maintainable, and slow workaround for a problem that absolutely should be fixed in the graphics card driver. This is even more the case as the Intel Linux drivers were known to be one of the buggiest Mesa/OpenGL drivers, and also one of the most rapidly changing. This article is similar in concept to this problem and is worth a read:

Regards,
Edward
bugman
Moderator
 
Posts: 1808
Joined: Thu Mar 19, 2015 10:01 am
Version: next

Re: Landmass problem

Postby abassign » Wed Aug 16, 2017 5:31 pm

bugman wrote in Wed Aug 16, 2017 9:55 am:I think abassign is trying to say that we should have specific hacks in FlightGear to work around short-term graphics card driver bugs. On the other side however, this is a nasty, non-maintainable, and slow workaround for a problem that absolutely should be fixed in the graphics card driver. This is even more the case as the Intel Linux drivers were known to be one of the buggiest Mesa/OpenGL drivers, and also one of the most rapidly changing. This article is similar in concept to this problem ...


I remember (I use KDE forever), however these are the easy mistakes, often due to the introduction of new features etc ...
Our case is much more complicated, meanwhile, because of a filter that introduces a fusion (transition.frag?) And occurs when there is a strong local color variation, as if the drive or HD was adopting some strategy Which depends on the previous frame, something similar to what MPEG 4. It would make sense to lower the output frequency of the card. I would like to check the same thing on Windows (which I did not have ...) or MAC without NVIDIA or AMD GPUs. For this I ask from the beginning of this post other examples to better understand the problem.
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: Landmass problem

Postby Thorsten » Wed Aug 16, 2017 5:55 pm

Our case is much more complicated, meanwhile, because of a filter that introduces a fusion (transition.frag?) And occurs when there is a strong local color variation, as if the drive or HD was adopting some strategy Which depends on the previous frame, something similar to what MPEG 4. It would make sense to lower the output frequency of the card.


Seriously :?: :?: :?:

(One might think that as the author of the code I would know of any 'filter that introduces a fusion' or the role of 'transition.frag' in it, but - alas - I do not. So I will humbly confess that I have no idea what you're talking about when analyzing my code).
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby abassign » Wed Aug 16, 2017 6:51 pm

Thorsten wrote in Wed Aug 16, 2017 5:55 pm:Seriously :?: :?: :?:


Very serious, absolutely serious, we do not know the true architecture of a GPU as it is covered by industrial secret.
However, if you have read my posts and their tests, the problem is when Landmass is at maximum value and only when Transation is at its maximum value (5 seems to me)! If you do not understand, I explain it to you in another way:

If (Landmass> = 5 AND Transation> = 5) : then there are false and flickering colors
else : the colors are correct!
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: Landmass problem

Postby Thorsten » Thu Aug 17, 2017 6:00 am

Very serious, absolutely serious, we do not know the true architecture of a GPU as it is covered by industrial secret.


Makes you wonder how on Earth they write those OpenSource drivers, knowing nothing about how the GPU internally works...

However, if you have read my posts and their tests, the problem is when Landmass is at maximum value and only when Transation is at its maximum value (5 seems to me)!


Some of us have moved beyond that observations a whole ago, noted that it actually is Agriculture that needs to be set to full, used that to identify the precise bit of code that's running for these settings and even tentatively identified a line.

Maybe you give the thread another read?
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby abassign » Thu Aug 17, 2017 9:19 am

Thorsten wrote in Thu Aug 17, 2017 6:00 am:Very serious, absolutely serious, we do not know the true architecture of a GPU as it is covered by industrial secret.
Makes you wonder how on Earth they write those OpenSource drivers, knowing nothing about how the GPU internally works...

Unfortunately, the complexity of hardware products such as the CPU-GPU is so high that real architecture is not often known. To develop a drive, and not sign an NDA, an official architecture is made known, which often covers the one actually present in the device. This is also logical if I develop some architecture to make a system more performing, why do I have to tell my business competitors?

For example, if the GPU has a mixed architecture with a CPU (the Intel case), do you think there are no mechanisms similar to the predictive pipes that handle the code? The use of a compression method in transferring data to RAM through algorithms that are conceptually similar to MPEG can be a good strategy to increase the RAM access speed without too much reduction of the usable bandwidth from the CPU.
For example, watch this comparison: http://www.game-debate.com/gpu/index.php?gid=3221&gid2=885&compare=hd-i7-6700k-vs-geforce-gt-740
It does not seem surprising that 20-24 core can compete with 384 core and with a non-dedicated system RAM!
Certainly, the Intel GPU on the card goes slower: Texture Rate 5.6 GTexel / s (Intel) versus 31.8 GTexel / s (GT740) ... but the games at low resolution (800x600) goes at the same speed.
This seal means that the system is efficient!
With the my current Intel i7 compared to the previous GTX870M with 6 GB of RAM with FGFS I go slower, at the same resolution of about 5 times but GTX870M has 1344 core instead of 24 of Intel (ratio 1/56), theoretically it should go slower than 50 times not 5!
This was my surprise when NVIDIA broke and I had to use the i7 GPU. I thought I could no longer use FGFS, but instead, even at low resolution (1/4), I still get 15-20 fps ... as before!
Of course, to get similar performance, you need to work well on internal architecture, with methods and tricks that the manufacturer hides as they are the technological advantage over the competition.

Thorsten wrote in Thu Aug 17, 2017 6:00 am:Some of us have moved beyond that observations a whole ago, noted that it actually is Agriculture that needs to be set to full, used that to identify the precise bit of code that's running for these settings and even tentatively identified a line.

Too bad I have seen, tried and verified that this method does not use "agriculture-ALS.frag" and therefore only serves to skip the problem...

I'm interested in solving problems, understanding and not simply skip it!
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: Landmass problem

Postby Thorsten » Thu Aug 17, 2017 9:39 am

Too bad I have seen, tried and verified that this method does not use "agriculture-ALS.frag" and therefore only serves to skip the problem...


*rofl*

You have a funny idea of humility. I wrote the whole effect - and you tell me that I have no idea what pieces of code it runs when? Seriously?

For the record, you have very wrong notions about how the quality sliders control the effects as well as very wrong notions about what happens internally in the rendering code.

You also use mighty words like 'Incompleteness theorem' without showing any sign of knowledge of what they actually mean. You frequently ignored actual information I and others have posted in favor of keeping your pet theories.

I'm interested in solving problems, understanding and not simply skip it!


Really? I didn't get the impression anywhere - at least you constantly reject any bit of information handed your way if it doesn't confirm what you already believe. You're unlikely to ever get any problem solved that way.

Anyway - please try solving it your way, I'll just point out to the innocent reader whatever you get wrong.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby bugman » Thu Aug 17, 2017 10:19 am

abassign wrote in Thu Aug 17, 2017 9:19 am:Unfortunately, the complexity of hardware products such as the CPU-GPU is so high that real architecture is not often known. To develop a drive, and not sign an NDA, an official architecture is made known, which often covers the one actually present in the device. This is also logical if I develop some architecture to make a system more performing, why do I have to tell my business competitors?


You know that the Intel Mesa/OpenGL open source drivers for Linux are written by Intel themselves, just like the Windows drivers? They build the CPU and GPU and write all the device drivers themselves. And that if you tell them about a bug in their driver, they will often fix it for the next release?

Regards,
Edward
bugman
Moderator
 
Posts: 1808
Joined: Thu Mar 19, 2015 10:01 am
Version: next

Re: Landmass problem

Postby abassign » Thu Aug 17, 2017 12:25 pm

@Thorsten

However, if you want to know if "agriculture-ALS.frag" is used or not ... just enter an error in its code and, like magic, you will see the black area. If you move the "Agriculture" slider to the left, you'll see that "magically" reappears the area and so it means using a method that does not call ... "agriculture-ALS.frag".
They are definitely primitive debugging techniques, but they permit, quickly, to understand many things.
Last edited by abassign on Thu Aug 17, 2017 12:49 pm, edited 2 times in total.
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: Landmass problem

Postby abassign » Thu Aug 17, 2017 12:46 pm

bugman wrote in Thu Aug 17, 2017 10:19 am:You know that the Intel Mesa/OpenGL open source drivers for Linux are written by Intel themselves, just like the Windows drivers? They build the CPU and GPU and write all the device drivers themselves. And that if you tell them about a bug in their driver, they will often fix it for the next release?


I've updated the drives using a nightly version, but nothing happened :(!
I continue to wonder if anyone who has Windows with an Intel i7 system can do the same test without using the external GPU. I would be curious to understand.
However, Linux drives are made by Intel and it is not said, being Linux opensource, which contain confidential calls that require you to sign up for an NDA. Microsoft, when it develops core drives, has two sets of NDA that securely protect each other.
At this point I think the method adopted in "agriculture-ALS.frag" + "transition.frag" could always cause problems for Linux when low performance HW is used.
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: Landmass problem

Postby Thorsten » Thu Aug 17, 2017 2:11 pm

They are definitely primitive debugging techniques, but they permit, quickly, to understand many things.


They might permit you to understand things if you would use them to do so, rather than to try to confirm what you believe to know. Yet sadly, you do not actually use them, otherwise you'd know by now where you're wrong.

At this point I think the method adopted in "agriculture-ALS.frag" + "transition.frag" could always cause problems for Linux when low performance HW is used.


Yes - but as pointed out above, since you have no real idea how the effects work or what the code does (or even what code runs), that's not very relevant.

And, as pointed out before - that's what the quality slider is for - if it doesn't run on low performance HW, take it down a notch. As simple as that. :D
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby Thorsten » Thu Aug 17, 2017 3:34 pm

@enrogue:

One thought occurred to me - it's unlikely to be the smoothstep function which yields a problematic result - it's more likely the color being mixed in that's off.

So while fiddling with the line 'cures' the issue by not mixing anything, the step that the driver doesn't process likely occurs before.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby abassign » Thu Aug 17, 2017 5:01 pm

Dear @Thorsten

A place that I recommend to go on vacation to have a bath of humility:

Image

Then when you have bathed in humility, you are beautiful and cleansed of all your pride, you can try to solve the problem.
Meanwhile, if you have a machine with Windows installed and an Intel i7 processor ... you know the proof that you should do ...
I wish you a very good bath!
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: Landmass problem

Postby Thorsten » Thu Aug 17, 2017 6:01 pm

You are confusing pride with actual knowledge.

I do understand the code in question quite well for the simple reason that I wrote it. I know precisely what part runs when for the same reason - I don't need to resort to crude tests or guesswork to find that out.

It's in fact pretty preposterous of you to try to tell me how my own code works when it's obvious that you don't even understand basic things about the effect framework. If you believe in humility, try for once to listen to what people are telling you rather than assuming you know best because you can delete a line of code someone else pointed you to.

you know the proof that you should do ...


I have all evidence I need to conclude that your Intel driver does something funky - since you seem to not have understood it the first time - as far as I'm concerned the case is closed, the problem is identified.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Landmass problem

Postby abassign » Thu Aug 17, 2017 8:15 pm

@Thorsten

Always to follow the way of humility I suggest you read this manual:
https://en.wikipedia.org/wiki/Little_Flowers_of_St._Francis
THis is the book on line: http://www.ccel.org/ccel/ugolino/flowers.i.html
Before you criticize those who note that your code does not work sometimes ...

With so much affection :)

P.S:
Strange however that with X-Plane 10 this problem I have not noticed ... maybe I see little and I'm aging ... :(
Last edited by abassign on Thu Aug 17, 2017 8:31 pm, edited 2 times in total.
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

PreviousNext

Return to Effects and shaders

Who is online

Users browsing this forum: No registered users and 3 guests