MTM2 Beta Patch 0.47 - Improving from .46

[OG_RuDeE]

Ok, shadows missing on Kryptik Mountains plants and stock Voodoo Island on my other machine with 0.47.

Reverting to 0.46, shadows return on Kryptik and Voodoo plants, but still no shadows on QuidPro. Flickering on fences goes away though with 0.46. AMD thing?

Edit: Didn't see post above, will test.

[Fila]

Ok Quid Pro Quo is a special case. It has the sun settings as topdown, so the game registers as the sun being at a 0 angle (or 180 angle whichever way you wanna read it). You will only get shadows under your truck, a small shadow under those lamposts or a shadow in that hangar door. The fence flickering is real though, good thing you spotted that, need to fix it.

The flickering might actually be tied to the sun being top down and the game tries to render shadows but fails to place them properly. lol

[OG_RuDeE]

Does this work: https://drive.google.com/file/d/1IkJWZd ... drive_link

Edit: I may also need to run the test on my other machine (6650 XT), but it's time to put the kids to bed! Will try in the morning or at the latest tomorrow evening. Sorry to leave you hanging!

[Fila]

Yep, got it. Will now look into it.

[Fila]

Does your AI behave the same on both? that's the important bit. I need the one from the machine that stutters if they don't behave the same.

[KF2416]

I'm curious if I should disable Nglide as it interpecting, which causes the stadium placement to be opposite.

[Fila]

Ok, shadows missing on Kryptik Mountains plants and stock Voodoo Island on my other machine with 0.47.

Reverting to 0.46, shadows return on Kryptik and Voodoo plants, but still no shadows on QuidPro. Flickering on fences goes away though with 0.46. AMD thing?

Edit: Didn't see post above, will test.

To answer in short, yes. I've done some changes to increase performance which worked on my end but it seems that AMD doesn't like it. I will work on a fallback case.

In the meantime I am also working on porting this to DX11 and OpenGL, AMD might react differently to those. DX9 is ancient after all.

[Fila]

I'm curious if I should disable Nglide as it interpecting, which causes the stadium placement to be opposite.

It doesn't hurt to try. With that being said, I am slowly working towards DX11 version too, so maybe that would behave differently. Also both dgvoodoo and nglide ignore DX11 I believe.

[OG_RuDeE]

Does your AI behave the same on both? that's the important bit. I need the one from the machine that stutters if they don't behave the same.

New link from the machine that stutters: https://drive.google.com/file/d/1GuWbmE ... sp=sharing

I can see stuttering even just sitting at the start point. Also, no shadows.

Now, I'm off. Good luck! Will be back tomorrow or the next day.

[Fila]

Oh that's bad. It means the stuttering fix I tried didn't work :/

[KF2416]

I'm curious if I should disable Nglide as it interpecting, which causes the stadium placement to be opposite.

It doesn't hurt to try. With that being said, I am slowly working towards DX11 version too, so maybe that would behave differently. Also both dgvoodoo and nglide ignore DX11 I believe.

Should I either disable or uninstall nglide?

[Fila]

disable should be good enough.

[KF2416]

How do I disable it tho?

[Ford-chan]

How do I disable it tho?

On my PC, yours is Win10, so this may tricky for explanation, but simply go into Start, search Add or remove programs, from there nGlide should show up in the Installed Apps section where you can simply uninstall from that point onwards. The disabling part is something that is not an option within the nGlide app itself.

Not to mention, I still use it for old Papyrus games like NASCAR Racing 2 and 3.

[Fila]

Those old Papyrus games were soo good.

[Ford-chan]

No that's fine thanks. The texture bugs on the walls are from the model itself. It means that something is wrong on my end specifically, probably with the .pod file.

Be interesting to see if a direct copy and paste from MTM1's source code will have some different results, which is not possible since nobody has yet asked Mark for it.

Those old Papyrus games were soo good.

Better than some modern sims to a degree. Cale and this Youtuber GPLaps will tell you all about it.

[KF2416]

I even uninstalled nglide and the issue is still persisting.

[Fila]

Alright so the tests that I've had Rudee do confirm Ford-Chan's long standing position that MTM2 behaves differently on every machine.

Simple answer: Yes it does and now there's the data to back it up. It's not an Intel vs AMD thing, it's literally every CPU mixed with Windows/Linux (wherever it is run). The reason for this is that TRI used some non-deterministic code to compute AI trucks. The framerate limiter helps a lot for a different reason but the underlying issue is that the truck themselves read the maps differently on each different CPU.

Rudee... it is also connected to your different AMDs and why one stutters and one doesn't... or rather, what triggers the stuttering issue. I've compared both your CPU's to mine, your stuttering AMD has wildly different numbers whereas your non-stuttering one is closer to my Intel.

So the solution is to come up with a deterministic math equation. This means less random from the trucks BUT that randomness can eventually be added back on purpose rather than having it be completely wild.

Side note: apparently non-deterministic code is banned in multiplayer oriented games for... well, you guess why

[Fila]

I even uninstalled nglide and the issue is still persisting.

But the other mtm1 drags still work correct yeah?

It seems that for some reason it doesn't wanna render this particular arena and I am not sure why, there is no special code that sets them apart.

[KF2416]

I bet it was stadium model's size.

I wonder if there's a fix patch for TWA drag stock track.