Skip to content

A
angle
Commit 8df71735 by Geoff Lang Committed by Commit Bot
Options

D3D9: Perform the MAD operation AFTER pre/unmultiply logic.

The multiply/add uniforms are to mask the output of the shader based on the destination format. The pre/unmuliply logic should be performed on the source data without masking applied. TEST=conformance/textures/canvas/tex-2d-rgb-rgb-unsigned_byte.html BUG=angleproject:2192 Change-Id: I6e5f29a7d56e18bd1a81ff56500edb4c6b55eb55 Reviewed-on: https://chromium-review.googlesource.com/730494 Commit-Queue: Geoff Lang <geofflang@chromium.org> Reviewed-by: 's avatarJamie Madill <jmadill@chromium.org>
parent 56ff3ce2
Showing with 42 additions and 49 deletions
src/libANGLE/renderer/d3d/d3d9/shaders/Blit.ps
...@@ -26,19 +26,19 @@ float4 luminanceps(float4 texcoord : TEXCOORD0) : COLOR ...@@ -26,19 +26,19 @@ float4 luminanceps(float4 texcoord : TEXCOORD0) : COLOR
float4 luminancepremultps(float4 texcoord : TEXCOORD0) : COLOR float4 luminancepremultps(float4 texcoord : TEXCOORD0) : COLOR
{ {
float4 luma = (tex2D(tex, texcoord.xy).xw * mult.xw + add.xw).xxxy; float4 luma = tex2D(tex, texcoord.xy).xxxw;
luma.rgb *= luma.a; luma.rgb *= luma.a;
return luma; return luma * mult + add;
}; };
float4 luminanceunmultps(float4 texcoord : TEXCOORD0) : COLOR float4 luminanceunmultps(float4 texcoord : TEXCOORD0) : COLOR
{ {
float4 luma = (tex2D(tex, texcoord.xy).xw * mult.xw + add.xw).xxxy; float4 luma = tex2D(tex, texcoord.xy).xxxw;
if (luma.a > 0.0f) if (luma.a > 0.0f)
{ {
luma.rgb /= luma.a; luma.rgb /= luma.a;
} }
return luma; return luma * mult + add;
}; };
// RGB/A Component Mask Pixel Shader // RGB/A Component Mask Pixel Shader
...@@ -51,17 +51,17 @@ float4 componentmaskps(float4 texcoord : TEXCOORD0) : COLOR ...@@ -51,17 +51,17 @@ float4 componentmaskps(float4 texcoord : TEXCOORD0) : COLOR
float4 componentmaskpremultps(float4 texcoord : TEXCOORD0) : COLOR float4 componentmaskpremultps(float4 texcoord : TEXCOORD0) : COLOR
{ {
float4 color = tex2D(tex, texcoord.xy) * mult + add; float4 color = tex2D(tex, texcoord.xy);
color.rgb *= color.a; color.rgb *= color.a;
return color; return color * mult + add;
}; };
float4 componentmaskunmultps(float4 texcoord : TEXCOORD0) : COLOR float4 componentmaskunmultps(float4 texcoord : TEXCOORD0) : COLOR
{ {
float4 color = tex2D(tex, texcoord.xy) * mult + add; float4 color = tex2D(tex, texcoord.xy);
if (color.a > 0.0f) if (color.a > 0.0f)
{ {
color.rgb /= color.a; color.rgb /= color.a;
} }
return color; return color * mult + add;
}; };
src/libANGLE/renderer/d3d/d3d9/shaders/compiled/componentmaskpremultps.h
...@@ -22,9 +22,9 @@ ...@@ -22,9 +22,9 @@
dcl t0.xy dcl t0.xy
dcl_2d s0 dcl_2d s0
texld r0, t0, s0 texld r0, t0, s0
mul r0.xyz, r0.w, r0
mov r1, c0 mov r1, c0
mad r0, r0, r1, c1 mad r0, r0, r1, c1
mul r0.xyz, r0.w, r0
mov oC0, r0 mov oC0, r0
// approximately 5 instruction slots used (1 texture, 4 arithmetic) // approximately 5 instruction slots used (1 texture, 4 arithmetic)
...@@ -44,7 +44,7 @@ const BYTE g_ps20_componentmaskpremultps[] = { ...@@ -44,7 +44,7 @@ const BYTE g_ps20_componentmaskpremultps[] = {
54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31, 54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31,
0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144, 0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144,
0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228, 0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228,
160, 1, 0, 0, 2, 1, 0, 15, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 160, 5, 0, 0, 3, 0, 0, 7, 128, 0, 0, 255, 128, 0, 0, 228, 128, 1, 0,
15, 128, 0, 0, 228, 128, 1, 0, 228, 128, 1, 0, 228, 160, 5, 0, 0, 3, 0, 0, 2, 1, 0, 15, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 15, 128, 0,
0, 7, 128, 0, 0, 255, 128, 0, 0, 228, 128, 1, 0, 0, 2, 0, 8, 15, 128, 0, 228, 128, 1, 0, 228, 128, 1, 0, 228, 160, 1, 0, 0, 2, 0, 8, 15, 128,
0, 0, 228, 128, 255, 255, 0, 0}; 0, 0, 228, 128, 255, 255, 0, 0};
src/libANGLE/renderer/d3d/d3d9/shaders/compiled/componentmaskunmultps.h
...@@ -22,11 +22,11 @@ ...@@ -22,11 +22,11 @@
dcl t0.xy dcl t0.xy
dcl_2d s0 dcl_2d s0
texld r0, t0, s0 texld r0, t0, s0
rcp r1.w, r0.w
mul r1.xyz, r0, r1.w
cmp r0.xyz, -r0.w, r0, r1
mov r1, c0 mov r1, c0
mad r0, r0, r1, c1 mad r0, r0, r1, c1
rcp r1.x, r0.w
mul r1.xyz, r0, r1.x
cmp r0.xyz, -r0.w, r0, r1
mov oC0, r0 mov oC0, r0
// approximately 7 instruction slots used (1 texture, 6 arithmetic) // approximately 7 instruction slots used (1 texture, 6 arithmetic)
...@@ -46,9 +46,9 @@ const BYTE g_ps20_componentmaskunmultps[] = { ...@@ -46,9 +46,9 @@ const BYTE g_ps20_componentmaskunmultps[] = {
54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31, 54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31,
0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144, 0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144,
0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228, 0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228,
160, 1, 0, 0, 2, 1, 0, 15, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 160, 6, 0, 0, 2, 1, 0, 8, 128, 0, 0, 255, 128, 5, 0, 0, 3, 1, 0,
15, 128, 0, 0, 228, 128, 1, 0, 228, 128, 1, 0, 228, 160, 6, 0, 0, 2, 1, 7, 128, 0, 0, 228, 128, 1, 0, 255, 128, 88, 0, 0, 4, 0, 0, 7, 128, 0,
0, 1, 128, 0, 0, 255, 128, 5, 0, 0, 3, 1, 0, 7, 128, 0, 0, 228, 128, 0, 255, 129, 0, 0, 228, 128, 1, 0, 228, 128, 1, 0, 0, 2, 1, 0, 15, 128,
1, 0, 0, 128, 88, 0, 0, 4, 0, 0, 7, 128, 0, 0, 255, 129, 0, 0, 228, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 15, 128, 0, 0, 228, 128, 1, 0, 228,
128, 1, 0, 228, 128, 1, 0, 0, 2, 0, 8, 15, 128, 0, 0, 228, 128, 255, 255, 128, 1, 0, 228, 160, 1, 0, 0, 2, 0, 8, 15, 128, 0, 0, 228, 128, 255, 255,
0, 0}; 0, 0};
src/libANGLE/renderer/d3d/d3d9/shaders/compiled/luminancepremultps.h
...@@ -22,14 +22,12 @@ ...@@ -22,14 +22,12 @@
dcl t0.xy dcl t0.xy
dcl_2d s0 dcl_2d s0
texld r0, t0, s0 texld r0, t0, s0
mov r1.xw, c0 mul r0.xyz, r0.w, r0.x
mad r0.x, r0.x, r1.x, c1.x mov r1, c0
mad r0.y, r0.w, r1.w, c1.w mad r0, r0, r1, c1
mul r1.xyz, r0.y, r0.x mov oC0, r0
mov r1.w, r0.y
mov oC0, r1
// approximately 7 instruction slots used (1 texture, 6 arithmetic) // approximately 5 instruction slots used (1 texture, 4 arithmetic)
#endif #endif
const BYTE g_ps20_luminancepremultps[] = { const BYTE g_ps20_luminancepremultps[] = {
...@@ -46,9 +44,7 @@ const BYTE g_ps20_luminancepremultps[] = { ...@@ -46,9 +44,7 @@ const BYTE g_ps20_luminancepremultps[] = {
54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31, 54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31,
0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144, 0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144,
0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228, 0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228,
160, 1, 0, 0, 2, 1, 0, 9, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 160, 5, 0, 0, 3, 0, 0, 7, 128, 0, 0, 255, 128, 0, 0, 0, 128, 1, 0,
1, 128, 0, 0, 0, 128, 1, 0, 0, 128, 1, 0, 0, 160, 4, 0, 0, 4, 0, 0, 2, 1, 0, 15, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 15, 128, 0,
0, 2, 128, 0, 0, 255, 128, 1, 0, 255, 128, 1, 0, 255, 160, 5, 0, 0, 3, 0, 228, 128, 1, 0, 228, 128, 1, 0, 228, 160, 1, 0, 0, 2, 0, 8, 15, 128,
1, 0, 7, 128, 0, 0, 85, 128, 0, 0, 0, 128, 1, 0, 0, 2, 1, 0, 8, 0, 0, 228, 128, 255, 255, 0, 0};
128, 0, 0, 85, 128, 1, 0, 0, 2, 0, 8, 15, 128, 1, 0, 228, 128, 255, 255,
0, 0};
src/libANGLE/renderer/d3d/d3d9/shaders/compiled/luminanceunmultps.h
...@@ -22,16 +22,14 @@ ...@@ -22,16 +22,14 @@
dcl t0.xy dcl t0.xy
dcl_2d s0 dcl_2d s0
texld r0, t0, s0 texld r0, t0, s0
mov r1.xw, c0 rcp r1.w, r0.w
mad r0.x, r0.x, r1.x, c1.x mul r1.x, r0.x, r1.w
mad r0.y, r0.w, r1.w, c1.w cmp r0.xyz, -r0.w, r0.x, r1.x
rcp r0.z, r0.y mov r1, c0
mul r0.z, r0.z, r0.x mad r0, r0, r1, c1
cmp r1.xyz, -r0.y, r0.x, r0.z mov oC0, r0
mov r1.w, r0.y
mov oC0, r1
// approximately 9 instruction slots used (1 texture, 8 arithmetic) // approximately 7 instruction slots used (1 texture, 6 arithmetic)
#endif #endif
const BYTE g_ps20_luminanceunmultps[] = { const BYTE g_ps20_luminanceunmultps[] = {
...@@ -48,10 +46,9 @@ const BYTE g_ps20_luminanceunmultps[] = { ...@@ -48,10 +46,9 @@ const BYTE g_ps20_luminanceunmultps[] = {
54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31, 54, 46, 51, 46, 57, 54, 48, 48, 46, 49, 54, 51, 56, 52, 0, 171, 171, 171, 31,
0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144, 0, 0, 2, 0, 0, 0, 128, 0, 0, 3, 176, 31, 0, 0, 2, 0, 0, 0, 144,
0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228, 0, 8, 15, 160, 66, 0, 0, 3, 0, 0, 15, 128, 0, 0, 228, 176, 0, 8, 228,
160, 1, 0, 0, 2, 1, 0, 9, 128, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 160, 6, 0, 0, 2, 1, 0, 8, 128, 0, 0, 255, 128, 5, 0, 0, 3, 1, 0,
1, 128, 0, 0, 0, 128, 1, 0, 0, 128, 1, 0, 0, 160, 4, 0, 0, 4, 0, 1, 128, 0, 0, 0, 128, 1, 0, 255, 128, 88, 0, 0, 4, 0, 0, 7, 128, 0,
0, 2, 128, 0, 0, 255, 128, 1, 0, 255, 128, 1, 0, 255, 160, 6, 0, 0, 2, 0, 255, 129, 0, 0, 0, 128, 1, 0, 0, 128, 1, 0, 0, 2, 1, 0, 15, 128,
0, 0, 4, 128, 0, 0, 85, 128, 5, 0, 0, 3, 0, 0, 4, 128, 0, 0, 170, 0, 0, 228, 160, 4, 0, 0, 4, 0, 0, 15, 128, 0, 0, 228, 128, 1, 0, 228,
128, 0, 0, 0, 128, 88, 0, 0, 4, 1, 0, 7, 128, 0, 0, 85, 129, 0, 0, 128, 1, 0, 228, 160, 1, 0, 0, 2, 0, 8, 15, 128, 0, 0, 228, 128, 255, 255,
0, 128, 0, 0, 170, 128, 1, 0, 0, 2, 1, 0, 8, 128, 0, 0, 85, 128, 1, 0, 0};
0, 0, 2, 0, 8, 15, 128, 1, 0, 228, 128, 255, 255, 0, 0};
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment