Offloudef: yeah, SM4 modifiers have nothing to do with the issue. you have only two different modifiers in the log: 0x00155543 and 0x800000c2. these are return types and resource types for resources access instructions (sample etc)
Offloudef
jkucia: good to know, I tried to look them up without anything
glueless about hlsl in general
s/glueless/clueless/
NP-HardBot
Correction(Offloudef): clueless about hlsl in general
ehoover joined the channel
m4sk1n joined the channel
Andre_H has quit
jkucia
Offloudef: BTW, is there a wine bug for the problem you're trying to solve?
Offloudef
jkucia: nope planning to create one now with this information
though winehq seems down
jkucia
it's slow
Offloudef: fwiw, I expect the shader translation for this shader to be correct. I guess you need to examine shader inputs, the constant buffer and the texture to check what's wrong
m4sk1n has quit
Offloudef
jkucia: I could make my LD_PRELOAD hack to dump those I guess
jkucia: with that one line commented the output on screen at least looks correct. It makes me wonder if that texkill is implemented right in SM4 codepath
the comment in wined3d's source says "If any of the first 3 components are < 0, discard this pixel "
but the sm4 implementation seems to cast it to bool
changing that particular line to SM2's version: if (any(lessThan(R1.xyzw, vec4(0.0)))) discard; seems to also work