Loading Overlays

[sven]

I've just taken a shot at re-structuring some of the multithreading code so that we can show overlays and/or progress indicators while the game is processing.

The current progress indicators are pretty crude -- but, my main concern right now is seeing whether or not the new system is in fact stable. To enable the load screens, I'm now doing some things behind the scenes that *might* reintroduce some of the AMD-specific crashes we saw at the beginning of the beta. So, if you notice any crashes after updating to the latest stable build (r15147), please let me know

edited by sven: Looks like I may have gotten a bit ahead of myself. Testing on my own AMD hardware has indeed revealed some driver issues. As of r15149 (now marked stable), the loading overlays are turned off. Hopefully, I'll find a workaround for the bugs soon...

[sven]

Ok, as of the current 'dev' build, I have a version of the loading overlays that's not immediately crashing on my own AMD hardware.

I haven't had a chance to do careful stability testing on Intel/AMD drivers yet. But, if anyone wants to take it for a spin, the latest build is up on 'dev', r15158.

[NullVoid]

Starting a new game on dev crashes the game on universe generation with the 64-bit binary. 32-bit seems to work as intended. The crash.dmp file is 905,484,429 bytes in size, so I can't attach it to e-mail or to the forum. Is the updater capable of sending that?

[sven]

Starting a new game on dev crashes the game on universe generation with the 64-bit binary. 32-bit seems to work as intended. The crash.dmp file is 905,484,429 bytes in size, so I can't attach it to e-mail or to the forum. Is the updater capable of sending that?

I can replicate the crash. (I tested the 32 bit version of the 'dev' build on intel drivers last night, but, not the 64.) So, no need to send the diagnostic file. Thanks for helping to track down these bleeding-edge issues

[sven]

I can replicate the crash. (I tested the 32 bit version of the 'dev' build on intel drivers last night, but, not the 64.) So, no need to send the diagnostic file. Thanks for helping to track down these bleeding-edge issues

Ok, as of r15162, I've got a version that appears to be reasonably stable on all my AMD/Intel/NVidia test setups. I've just marked this build as 'stable', so, now we'll see how well it performs in the real world

[sven]

Ok, as of r15162, I've got a version that appears to be reasonably stable on all my AMD/Intel/NVidia test setups. I've just marked this build as 'stable', so, now we'll see how well it performs in the real world

Looks like the changes introduced a minor bug that could come up when you hit "Continue" from the start menu. Should be fixed in the new stable build, r15165. (Ty to bjg for the error upload.)

[SirDamnALot]

With the new overlays, the FX effect when fighters fire is not rendered properly (on AMD Radeon 200 series GPU):

Untitled.png (643.6 KiB) Viewed 16005 times

With parameter -no_overlays, the effect is working:

Untitled2.png (409.13 KiB) Viewed 16005 times

[sven]

With the new overlays, the FX effect when fighters fire is not rendered properly (on AMD Radeon 200 series GPU):

Wow, neat bug! I'll look into it. Many thanks.

[sven]

With the new overlays, the FX effect when fighters fire is not rendered properly (on AMD Radeon 200 series GPU):

Ok, I'm not able to replicate this glitch on my own AMD hardware. However, I can think of one thing that might fix it. Would you try updating to the latest 'dev' build (15167), and checking to if the fighter effects are still corrupted?

(The loading screens in this mode are still incomplete -- atm, they're just a blank white screen with some grey boxes -- but, if the effects bugs go away, I could update them to something a bit nicer looking.)

[SirDamnALot]

With the new overlays, the FX effect when fighters fire is not rendered properly (on AMD Radeon 200 series GPU):

Ok, I'm not able to replicate this glitch on my own AMD hardware. However, I can think of one thing that might fix it. Would you try updating to the latest 'dev' build (15167), and checking to if the fighter effects are still corrupted?

(The loading screens in this mode are still incomplete -- atm, they're just a blank white screen with some grey boxes -- but, if the effects bugs go away, I could update them to something a bit nicer looking.)

Updated to R15167 and some strange results.
Initially the effect is working, but then switches to the black boxes.
If the effect is working, I can even rewind it with the frame bar and it still works.
After it goes to black boxes, it stays with black boxes. Even if I load a savegame and fight a different fight.

Unbenannt.png (187.71 KiB) Viewed 15990 times

- game_988, Rasalgethi, Bombers VS a pirate cruiser/destroyers,
Sometimes the effect works for several bomber strikes, then goes dark. One time I could win the fight without the glitch occouring.
- game_987, Menkent I, strike fighter VS planetary defenses
several tries, it always goes black on the second attack.

[sven]

Updated to R15167 and some strange results.
Initially the effect is working, but then switches to the black boxes.
If the effect is working, I can even rewind it with the frame bar and it still works.

Ok, this is all very helpful. I have a few more theories, and have uploaded another patch (r15169). I know it's probably about time for you to head to bed, but, if you get a chance, I'd be curious to know what effect, if any, the latest 'dev' changes have on these glitches.

[SirDamnALot]

Updated to R15167 and some strange results.
Initially the effect is working, but then switches to the black boxes.
If the effect is working, I can even rewind it with the frame bar and it still works.

Ok, this is all very helpful. I have a few more theories, and have uploaded another patch (r15169). I know it's probably about time for you to head to bed, but, if you get a chance, I'd be curious to know what effect, if any, the latest 'dev' changes have on these glitches.

R15173 (blue background with "loading") with AMD Radeon 290x, still the same symptoms.
The effect works 1 or 2 times, then stays corrupted as boxes until Quit & restart of the game. (Both 32 and 64 exe)

[SirDamnALot]

R15173 (blue background with "loading"), now with Nvidia GTX 570.
No symptoms of the error to be found.
Also: the GPU load behaves itself in window mode.
(Caps at refresh rate, and 50% gpu load max. The AMD 290x caps too, but still works like mad)

[SirDamnALot]

R15173 (blue background with "loading"), now with AMD Radeon 5870.
Not really surprising, same driver, same error. Black boxes occour.
Thats all from my hardware archive

Maybe some other AMD users could do a sanity check ? Here are the 2 savegames that have the 2 fighter battles:
http://www.fastshare.org/download/SavedGames.zip

[sven]

R15173 (blue background with "loading"), now with AMD Radeon 5870.
Not really surprising, same driver, same error. Black boxes occour.
Thats all from my hardware archive

Hrm. I'm unable to replicate the error using my Radeon HD 7000, using driver version 14.501.1003.0 (which was released in 2014). When you installed SiS, it looks like your machine was also setup with driver version 14.501.1003.0. Have you updated your graphics driver since installing SiS?