ShuriZma
/
pcsx2
mirror of https://github.com/PCSX2/pcsx2.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

glsl: disable computing of extra alpha coeff in SW blending 

Hum, I'm curious of the impact to enable only this code when it is actually used.
This commit is contained in:
Gregory Hainaut 2015-07-17 17:53:00 +02:00
parent 784822a5c2
commit e3751f6cd9
2 changed files with 36 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
plugins/GSdx/res/glsl/tfx_fs.glsl
View File

@ -35,7 +35,9 @@ in SHADER
// Same buffer but 2 colors for dual source blending // Same buffer but 2 colors for dual source blending
layout(location = 0, index = 0) out vec4 SV_Target0; layout(location = 0, index = 0) out vec4 SV_Target0;
#if !SW_BLEND
layout(location = 0, index = 1) out vec4 SV_Target1; layout(location = 0, index = 1) out vec4 SV_Target1;
#endif
#ifdef ENABLE_BINDLESS_TEX #ifdef ENABLE_BINDLESS_TEX
layout(bindless_sampler, location = 0) uniform sampler2D TextureSampler; layout(bindless_sampler, location = 0) uniform sampler2D TextureSampler;
@ -218,6 +220,7 @@ mat4 sample_4p(uvec4 u)
vec4 sample_color(vec2 st, float q) vec4 sample_color(vec2 st, float q)
{ {
//FIXME: maybe we can set gl_Position.w = q in VS
#if (PS_FST == 0) #if (PS_FST == 0)
st /= q; st /= q;
#endif #endif
@ -287,6 +290,7 @@ vec4 sample_color(vec2 st, float q)
return t; return t;
} }
// FIXME Precompute the factor 255/128 in VS
#ifndef SUBROUTINE_GL40 #ifndef SUBROUTINE_GL40
vec4 tfx(vec4 t, vec4 c) vec4 tfx(vec4 t, vec4 c)
{ {
@ -552,17 +556,29 @@ void ps_main()
c.rb = c.rr; c.rb = c.rr;
#endif #endif
// FIXME precompute my_TA & 0x80
// Write GA part. Mask will take care of the correct destination // Write GA part. Mask will take care of the correct destination
// Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\n"
// However Nvidia emulate it with an if (at least on kepler arch) ...\n"
#if PS_READ_BA #if PS_READ_BA
if (bool(denorm_c.a & 0x80u)) if (bool(denorm_c.a & 0x80u))
c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f); c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);
else else
c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f); c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);
#else #else
if (bool(denorm_c.g & 0x80u)) if (bool(denorm_c.g & 0x80u))
c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f); c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);
else else
c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f); c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);
// Nice idea but step/mix requires 4 instructions
// set / trunc / I2F / Mad
//
// float sel = step(128.0f/255.0f, c.g);
// vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u)) / 255.0f;
// c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);
#endif #endif
#endif #endif
@ -601,7 +617,9 @@ void ps_main()
ps_fbmask(c); ps_fbmask(c);
SV_Target0 = c; SV_Target0 = c;
#if !SW_BLEND
SV_Target1 = vec4(alpha_blend); SV_Target1 = vec4(alpha_blend);
#endif
} }
#endif #endif

18
plugins/GSdx/res/glsl_source.h
View File

@ -928,7 +928,9 @@ static const char* tfx_fs_all_glsl =
"\n" "\n"
"// Same buffer but 2 colors for dual source blending\n" "// Same buffer but 2 colors for dual source blending\n"
"layout(location = 0, index = 0) out vec4 SV_Target0;\n" "layout(location = 0, index = 0) out vec4 SV_Target0;\n"
"#if !SW_BLEND\n"
"layout(location = 0, index = 1) out vec4 SV_Target1;\n" "layout(location = 0, index = 1) out vec4 SV_Target1;\n"
"#endif\n"
"\n" "\n"
"#ifdef ENABLE_BINDLESS_TEX\n" "#ifdef ENABLE_BINDLESS_TEX\n"
"layout(bindless_sampler, location = 0) uniform sampler2D TextureSampler;\n" "layout(bindless_sampler, location = 0) uniform sampler2D TextureSampler;\n"
@ -1111,6 +1113,7 @@ static const char* tfx_fs_all_glsl =
"\n" "\n"
"vec4 sample_color(vec2 st, float q)\n" "vec4 sample_color(vec2 st, float q)\n"
"{\n" "{\n"
" //FIXME: maybe we can set gl_Position.w = q in VS\n"
"#if (PS_FST == 0)\n" "#if (PS_FST == 0)\n"
" st /= q;\n" " st /= q;\n"
"#endif\n" "#endif\n"
@ -1180,6 +1183,7 @@ static const char* tfx_fs_all_glsl =
" return t;\n" " return t;\n"
"}\n" "}\n"
"\n" "\n"
"// FIXME Precompute the factor 255/128 in VS\n"
"#ifndef SUBROUTINE_GL40\n" "#ifndef SUBROUTINE_GL40\n"
"vec4 tfx(vec4 t, vec4 c)\n" "vec4 tfx(vec4 t, vec4 c)\n"
"{\n" "{\n"
@ -1445,17 +1449,29 @@ static const char* tfx_fs_all_glsl =
" c.rb = c.rr;\n" " c.rb = c.rr;\n"
"#endif\n" "#endif\n"
"\n" "\n"
" // FIXME precompute my_TA & 0x80\n"
"\n"
" // Write GA part. Mask will take care of the correct destination\n" " // Write GA part. Mask will take care of the correct destination\n"
" // Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\\n\"\n"
" // However Nvidia emulate it with an if (at least on kepler arch) ...\\n\"\n"
"#if PS_READ_BA\n" "#if PS_READ_BA\n"
" if (bool(denorm_c.a & 0x80u))\n" " if (bool(denorm_c.a & 0x80u))\n"
" c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);\n" " c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);\n"
" else\n" " else\n"
" c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);\n" " c.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);\n"
"\n"
"#else\n" "#else\n"
" if (bool(denorm_c.g & 0x80u))\n" " if (bool(denorm_c.g & 0x80u))\n"
" c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);\n" " c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)) / 255.0f);\n"
" else\n" " else\n"
" c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);\n" " c.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)) / 255.0f);\n"
"\n"
" // Nice idea but step/mix requires 4 instructions\n"
" // set / trunc / I2F / Mad\n"
" //\n"
" // float sel = step(128.0f/255.0f, c.g);\n"
" // vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u)) / 255.0f;\n"
" // c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);\n"
"#endif\n" "#endif\n"
"\n" "\n"
"#endif\n" "#endif\n"
@ -1494,7 +1510,9 @@ static const char* tfx_fs_all_glsl =
" ps_fbmask(c);\n" " ps_fbmask(c);\n"
"\n" "\n"
" SV_Target0 = c;\n" " SV_Target0 = c;\n"
"#if !SW_BLEND\n"
" SV_Target1 = vec4(alpha_blend);\n" " SV_Target1 = vec4(alpha_blend);\n"
"#endif\n"
"}\n" "}\n"
"\n" "\n"
"#endif\n" "#endif\n"