September 12th, 2019 Update - Alpha sign-up

[mattskiam 1]

On 9/13/2019 at 3:45 PM, pegruder said:

Anyone ever able to sign up - mine hangs on the dxdiag upload forever.  Tried 3 diff browers - and 3 PCs lmao.

Same problem with me.

[pegruder 65]

1 hour ago, mattskiam said:

Same problem with me.

I guess no testing for us 😞

[FDEdev 955]

If you are using win10 have you tried with Edge?

[HughesMDflyer4 470]

6 hours ago, FDEdev said:

Aerofly is a flying game which is even much less complex than Ace combat. With almost zero features it's no wonder that the FPS is so high.

Please stop spreading this myth that anything considered a "game" is automatically doing less behind the scenes, and can therefore magically perform better. Aerofly's scenery engine (at least when I tried it a couple years ago) can handle an impressively high poly count/object density, which is one of the main performance killers in other sims. Aircraft systems and flight dynamics aren't very intensive/demanding calculations.

[FDEdev 955]

19 minutes ago, HughesMDflyer4 said:

Please stop spreading this myth that anything considered a "game" is automatically doing less behind the scenes, and can therefore magically perform better.

Aircraft systems and flight dynamics aren't very intensive/demanding calculations.

I never said that. IMO Aerofly is a game because it's missing way too much features to be considered a simulator, that's all.

Btw, I doubt that e.g. the FSLabs A320 systems simulation isn't intensive/demanding.

Edited September 17, 2019 by FDEdev

[Paraffin 2,197]

20 minutes ago, HughesMDflyer4 said:

Aircraft systems and flight dynamics aren't very intensive/demanding calculations.

True up to a point, but it can depend on how many systems are being modeled and how deeply. I have one aircraft model that's a guaranteed 8-10 fps drop over other aircraft, just from loading it.

Weather modeling can also be demanding. I get my worst frame rates when multiple (separate) vertical layers of thick clouds are generated, and it's much faster in clear skies. This is one reason why Aerofly FS2 has such high frame rates, with no real weather modeling (last I heard).

[Krakin 5,853]

Check your emails. They said the Tech Alpha is just about ready for testing and they're reminding everyone to get signed up. First legit email from the team i think.

[HughesMDflyer4 470]

59 minutes ago, FDEdev said:

I doubt that e.g. the FSLabs A320 systems simulation isn't intensive/demanding.

 

51 minutes ago, Paraffin said:

True up to a point, but it can depend on how many systems are being modeled and how deeply. I have one aircraft model that's a guaranteed 8-10 fps drop over other aircraft, just from loading it.

Weather modeling can also be demanding. I get my worst frame rates when multiple (separate) vertical layers of thick clouds are generated, and it's much faster in clear skies. This is one reason why Aerofly FS2 has such high frame rates, with no real weather modeling (last I heard).

 

Having worked on a complex airliner add-on before and having talked to the programmers working on it, I can tell you that systems should not impact your FPS in any noticeable way. There are a few things that can have an impact though:

- Digital displays: This is a huge one. The gauge rendering system in FSX/P3D is ridiculously inefficient and can bring a well performing aircraft to its knees if used improperly. P3Dv4 (and maybe v3?) integrates a new system known as "Render to Texture" via the PDK element of the SDK. A few devs have used this system and the performance impact of displays is basically nothing. All P3D add-ons should move to this system IMO. Rendering of the shapes/text for displays can also have a bit of an impact if done improperly, as most devs do this on the CPU still. These calculations alone are still more taxing than anything directly systems related.

- Unoptimized models: While graphics engines can easily handle several million polys, there is a point in every engine where it will begin to slow down, and a high poly model that uses many duplicate materials will push the renderer closer to that point. Many flight sim developers are not experienced game developers and have little knowledge of the techniques used to create efficient art assets where strict performance budgets are followed. This begins to show when you load up complex aircraft with many scenery add-ons.

Weather however is indeed a cause of poor FPS. The FSX/P3D cloud system has an insane amount of overdraw (transparent sprites being drawn on top of one another miles and miles away even though they're not visible), which is ridiculously inefficient. This is why you get worse FPS in big storms vs light clouds.

Edited September 17, 2019 by HughesMDflyer4

[Paraffin 2,197]

8 minutes ago, HughesMDflyer4 said:

Weather however is indeed a cause of poor FPS. The FSX/P3D cloud system has an insane amount of overdraw (transparent sprites being drawn on top of one another miles and miles away even though they're not visible), which is ridiculously inefficient. This is why you get worse FPS in big storms vs light clouds.

My example above was in XP11, which I think suggests that weather depiction is universally problematic in terms of CPU/GPU load for all sims. It's just a question of how optimized it is, and you get maximum optimization by not modeling it at all, like Aerosoft FS2. 🙂

There is no free lunch when you're trying to depict weather realistically. For example, the move away from 2D billboards to 3D volumetric clouds in one weather plugin (XEnviro) caused complaints from users about lowered frame rates. No free lunch there. Certain other weather-related effects like raindrops on the windscreen, or visual icing buildup on the wings (which is in one plane model I fly) eats up more CPU/GPU cycles.

People have been blown away by what looks like vastly better weather modeling in the MSFS previews, but if what we see is actual dynamic weather and not just a painting on the skybox, then there will be no free lunch in generating all those visuals and corresponding effects on the aircraft.

[Baber20 1,491]

1 hour ago, Krakin said:

Check your emails. They said the Tech Alpha is just about ready for testing and they're reminding everyone to get signed up. First legit email from the team i think.

Recieved one too today.

[Deleted 250]

8 minutes ago, Baber20 said:

Recieved one too today.

Guys, this is an automatic mail that goes to everyone who has registered. So that everyone knows it. So everyone got an email. 

[Gulfstream 416]

2 hours ago, Paraffin said:

True up to a point, but it can depend on how many systems are being modeled and how deeply. I have one aircraft model that's a guaranteed 8-10 fps drop over other aircraft, just from loading it.

 

You all are thinking of how these engines were written literally DECADES ago.

We have simulators these days that simulate spaceships everywhere from orbit, through the atmosphere, and then down to the planet's surface where you can walk around (see: Star Citizen). 

Loading a complex 747 into a earth-based Flight Simulator in 2019 should not be an issue if it is written to today's standards.   It doesn't matter how complex your flap system is, or your FMS, or your HUD, or what the texture resolution is.

Everyone is thinking about old flight simulators, maybe that's because we've been stuck there for so long.

My laptop's graphics card is capable of 1.69 trillion floating point operations per second.  And it's not a top of the line card at this point.  My CPU can do 354 billion per second.

We'll be ok, as long as this is coded with a new engine targeting 2019 and beyond.

Edited September 17, 2019 by Gulfstream

[stripealipe 87]

Good! Don't make me fill in those forms again lol... just gimme the link and lemme in... fingers crossed.. i have the hardware to push it through its paces...

[HughesMDflyer4 470]

2 hours ago, Paraffin said:

My example above was in XP11, which I think suggests that weather depiction is universally problematic in terms of CPU/GPU load for all sims. It's just a question of how optimized it is, and you get maximum optimization by not modeling it at all, like Aerosoft FS2. 🙂

There is no free lunch when you're trying to depict weather realistically. For example, the move away from 2D billboards to 3D volumetric clouds in one weather plugin (XEnviro) caused complaints from users about lowered frame rates. No free lunch there. Certain other weather-related effects like raindrops on the windscreen, or visual icing buildup on the wings (which is in one plane model I fly) eats up more CPU/GPU cycles.

People have been blown away by what looks like vastly better weather modeling in the MSFS previews, but if what we see is actual dynamic weather and not just a painting on the skybox, then there will be no free lunch in generating all those visuals and corresponding effects on the aircraft.

I don't think a third party integration of weather is really a fair performance comparison. Anyone besides those with access to the P3D or XP11 code is going to struggle to implement an entirely new weather system in the best and most efficient way possible way. Lots of corners have to be cut in add-on development.

Flight Sim World has rather advanced cloud rendering and seem unaffected by weather. In testing, I'm seeing the same performance in Flight Sim World in both light cloud coverage and heavy storms.

The effects you mention (rain/ice) are just shaders (I believe rain could even be done through post processing). Many games do those and many more effects in real time. There's nothing particularly demanding about them.

[188AHC 2,141]

Sadly I probably won't be chosen as I was filling out the application and the Wi-Fi failed dropping me out. I was hoping but we will see. 

I was able to finish the application when I got home but couldn't go back and correct anything,