Lastest Build WIP r759 Discuss (do not post builds) DAT+BAT*

Discuss Supermodel and your favorite Model 3 games. Show off your latest threads, this is the place to see and be seen.
Forum rules
Keep it classy!

  • No ROM requests or links.
  • Do not ask to be a play tester.
  • Do not ask about release dates.
  • No drama!

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Ian » Sat Aug 11, 2018 10:17 am

Besides slowness .. does it look more or less correct?
Ian
 
Posts: 1547
Joined: Tue Feb 23, 2016 9:23 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Ian » Sat Aug 11, 2018 11:58 am

For anyone that is reading this
YoYo, I found the reason for the crazy lines. A division by zero was happening in the shader. Was only happening I think with polys that were flattened to a single line I think. Anyway I put a simple check for div zero and it seems to fix it perfectly. Will release a build later if u wanna try :)

Before
Image

After div by zero check
Image
Ian
 
Posts: 1547
Joined: Tue Feb 23, 2016 9:23 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Ian » Sat Aug 11, 2018 12:38 pm

2nd test version if people want to try it out

https://ufile.io/2o0yt
Ian
 
Posts: 1547
Joined: Tue Feb 23, 2016 9:23 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Toshiko » Sat Aug 11, 2018 10:04 pm

I noticed a similar geometry bug in v1 that is mostly gone in v2, but still doesn't quite look correct - in the attract mode of Super GT (scuda), on the second-to-last corner of Beginner Night, there's a tile that was broken at the seams in v1, but is now warping into view in v2.
Toshiko
 
Posts: 119
Joined: Fri Mar 16, 2012 10:30 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Ian » Sun Aug 12, 2018 3:25 am

Post a screenshot ..
no way I can guess what the mysterious 1 missing tile is lol
Ian
 
Posts: 1547
Joined: Tue Feb 23, 2016 9:23 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Toshiko » Sun Aug 12, 2018 6:50 pm

It's fairly large, is in the middle of the screen, and it happens during a huge hit of slowdown, which makes it more obvious. Doesn't happen at all in the triangle renderers.

I'll capture it in OBS in a bit and make a GIF, since that'll both be easier and more apparent.
Toshiko
 
Posts: 119
Joined: Fri Mar 16, 2012 10:30 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Toshiko » Sun Aug 12, 2018 7:28 pm

Here it is~

Image
Toshiko
 
Posts: 119
Joined: Fri Mar 16, 2012 10:30 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby YoYo09 » Sun Aug 12, 2018 9:57 pm

Toshiko wrote:Here it is~

Image


Toshiko
Are you using an ati card?

I have not found this symptom.
I use gtx970.

Ian
like scene save file here
https://drive.google.com/file/d/1MDeX40 ... sp=sharing
YoYo09
 
Posts: 67
Joined: Mon Nov 28, 2016 7:43 pm

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Toshiko » Sun Aug 12, 2018 11:28 pm

ATi hasn't existed for years, but yes, it's an AMD card. I said as much two posts ago.

One of these days I'll get tired of avoiding the nvidia tax, but I'll need to be sufficiently well off to justify paying 30% more for the same performance with fewer bugs that almost exclusively exist in free software. Not that I don't covet my neighbor's 1080 Ti, but I'm not made of money.
Toshiko
 
Posts: 119
Joined: Fri Mar 16, 2012 10:30 am

Re: Lastest Build WIP r738 Discuss (do not post builds)

Postby Ian » Mon Aug 13, 2018 3:11 am

As a developer driver bugs are wearing. I wrote a small commercial opengl app some years ago, non gaming related. For non gamers intel integrated gfx had something like 50% of the market share. Only problem was their drivers were so fucking buggy it was ridiculous. In the end we shipped a MESA software opengl renderer with it, so on these cards it actually rendered in software. It rendered at like 5fps, but 5 was better than zero. One bug I encountered was glClear wouldn't work on a FBO (frame buffer object - basically offscreen render target). Well it did work, but it was using the wrong glViewport size. As a work around I wrote a custom glClear method to emulate what glClear does but with manual draw calls instead. But that still wasn't enough to fix all the bugs. In the end we just gave up and went the software route lol. AMD also had some show stopper driver bugs. But in older drivers they just go fuck you, your hardware is now legacy so these bugs won't get fixed for your hardware. NVidia has always had better opengl support and that's a fact.
Ian
 
Posts: 1547
Joined: Tue Feb 23, 2016 9:23 am

PreviousNext

Return to The Catwalk

Who is online

Users browsing this forum: No registered users and 9 guests