Re: [Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-07 Thread Timothy Arceri
On Fri, 2017-01-06 at 13:29 +0100, Iago Toral wrote: > Hi Tim, > > it's been a while, so off the top of my head I don't have any > particular suggestions or the capacity to tell whether your fix is > correct or not :-/. > > I'll try to spend some time re-acquainting myself with that code on > Mon

Re: [Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-06 Thread Iago Toral
Hi Tim, it's been a while, so off the top of my head I don't have any particular suggestions or the capacity to tell whether your fix is correct or not :-/. I'll try to spend some time re-acquainting myself with that code on Monday and see if I can see what is going on here. Iago On Fri, 2017-0

Re: [Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-06 Thread Timothy Arceri
On Fri, 2017-01-06 at 12:36 +0100, Nicolai Hähnle wrote: > On 06.01.2017 00:26, Timothy Arceri wrote: > > Marking operations as redundant if they are equal to the base > > range is fine when the tree structure is something like this: > > > > max > >   / \ > >  max b > >    

Re: [Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-06 Thread Nicolai Hähnle
On 06.01.2017 00:26, Timothy Arceri wrote: Marking operations as redundant if they are equal to the base range is fine when the tree structure is something like this: max / \ max b / \ 3max / \ 3 a But the opt falls apart with a tree li

Re: [Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-06 Thread Timothy Arceri
Hi Iago/Petri, Just Ccing you guys in case you have a better solution (I know its been a while since this was written). This is causing incorrect shaders in at least Serious Sam 3 possibly others. I've sent some piglit tests to reproduce it to the piglit list. Thanks, Tim On Fri, 2017-01-06 at

[Mesa-dev] [PATCH] glsl: fix opt_minmax redundancy checks against baserange

2017-01-05 Thread Timothy Arceri
Marking operations as redundant if they are equal to the base range is fine when the tree structure is something like this: max / \ max b / \ 3max / \ 3 a But the opt falls apart with a tree like this: max / \ max