Mostly there

This commit is contained in:
John Hunter
2023-03-21 15:50:53 +00:00
parent addb4a9dba
commit 3df909a6e8
9 changed files with 4090 additions and 470 deletions
+8 -5
View File
@@ -2,6 +2,7 @@
#version 460
#extension GL_EXT_mesh_shader:require
#define DescriptionIndex gl_PrimitiveID>>11
#include "include.glsl"
#ifdef implicit
@@ -107,8 +108,10 @@ void main(){
vec3 raydir=normalize(raypos-(inverse(pc.world)*vec4(camera_uniforms.campos,1)).xyz);
//raypos-=vec3(5);
//f_color=vec4(raydir,1.);
//return;
/*f_color=vec4(raydir,1.);
return;*/
/*f_color=vertexInput.position;
return;*/
#ifdef debug
f_color=vec4(sceneoverride(raypos,false),1);
@@ -116,14 +119,14 @@ void main(){
#endif
vec2 td=spheretracing(raypos,raydir,p);
#ifdef debug
/*#ifdef debug
f_color=vec4(td,0,1);
return;
#endif
#endif*/
vec3 n=getNormal(p,td.y);
if(td.y<EPSILON)
{
f_color=vec4(1.);
//f_color=vec4(1.);
f_color=vec4(shading(n),1.);
vec4 tpoint=camera_uniforms.proj*camera_uniforms.view*pc.world*vec4(p,1);
+70 -48
View File
@@ -3,6 +3,8 @@
#version 460
#extension GL_EXT_mesh_shader:require
uint DescriptionIndex;
#include "include.glsl"
#include "intervals.glsl"
@@ -14,35 +16,40 @@ layout(location=0)out VertexOutput
vec4 position;
}vertexOutput[];
struct MeshMasks
{
uint8_t masks[32][masklen]; //928
uint8_t enabled[32]; //32
vec3 bottomleft; //12
vec3 topright; //12
uint globalindex; //4
//uint objectindex; //4
}; //total = 992 bytes
taskPayloadSharedEXT MeshMasks meshmasks;
layout(set=0,binding=20)restrict writeonly buffer fragmentMasks{
uint8_t masks[][masklen];
}fragmentpassmasks;
void main()
{
uint localindex = uint(meshmasks.enabled[gl_WorkGroupID.x]);
DescriptionIndex = meshmasks.globalindex/2;
//clear_stacks();
default_mask();
#define CLIPCHECK 65536
float[6]bounds={
CLIPCHECK-sceneoverride(vec3(CLIPCHECK,0,0),false),
CLIPCHECK-sceneoverride(vec3(0,CLIPCHECK,0),false),
CLIPCHECK-sceneoverride(vec3(0,0,CLIPCHECK),false),
-CLIPCHECK+sceneoverride(vec3(-CLIPCHECK,0,0),false),
-CLIPCHECK+sceneoverride(vec3(0,-CLIPCHECK,0),false),
-CLIPCHECK+sceneoverride(vec3(0,0,-CLIPCHECK),false),
};
//default_mask();
mask = meshmasks.masks[gl_WorkGroupID.x];
vec3 bottomleft = meshmasks.bottomleft;
vec3 topright = meshmasks.topright;
vec4[8]positions={
vec4(bounds[3],bounds[4],bounds[5],1.),
vec4(bounds[3],bounds[4],bounds[2],1.),
vec4(bounds[3],bounds[1],bounds[5],1.),
vec4(bounds[3],bounds[1],bounds[2],1.),
vec4(bounds[0],bounds[4],bounds[5],1.),
vec4(bounds[0],bounds[4],bounds[2],1.),
vec4(bounds[0],bounds[1],bounds[5],1.),
vec4(bounds[0],bounds[1],bounds[2],1.),
vec4(bottomleft,1.),
vec4(bottomleft.x,bottomleft.y,topright.z,1.),
vec4(bottomleft.x,topright.y,bottomleft.z,1.),
vec4(bottomleft.x,topright.y,topright.z,1.),
vec4(topright.x,bottomleft.y,bottomleft.z,1.),
vec4(topright.x,bottomleft.y,topright.z,1.),
vec4(topright.x,topright.y,bottomleft.z,1.),
vec4(topright,1.),
};
//This can be optimised
@@ -51,21 +58,41 @@ void main()
uint vindex = gl_LocalInvocationID.x*8;
uint pindex = gl_LocalInvocationID.x*6;
int GlobalInvocationIndex = int(gl_GlobalInvocationID.z*gl_NumWorkGroups.x*gl_NumWorkGroups.y+gl_GlobalInvocationID.y*gl_NumWorkGroups.x+gl_GlobalInvocationID.x);
int GlobalInvocationIndex = int((meshmasks.globalindex*32+localindex)*32+gl_LocalInvocationID.x);
//adjust scale and position
for (int i = 0; i<8; i++)
{
positions[i] *= vec4(0.25,0.25,0.25,1.);
positions[i].x += (bounds[0]-bounds[3])*0.25 * (mod(gl_LocalInvocationID.x,4.)-1.5);
positions[i].y += (bounds[1]-bounds[4])*0.25 * (mod(floor(gl_LocalInvocationID.x/4.),4.)-1.5);
positions[i].z += (bounds[2]-bounds[5])*0.25 * (floor(gl_LocalInvocationID.x/16.)-1.5+gl_WorkGroupID.z*2.);
positions[i] *= vec4(0.25,0.25,0.5,1.);
positions[i].x += (topright.x-bottomleft.x)*0.25 * (mod(localindex,4.)-1.5);
positions[i].y += (topright.y-bottomleft.y)*0.25 * (mod(floor(localindex/4.),4.)-1.5);
positions[i].z += ((topright.z-bottomleft.z)*0.5 * (floor(localindex/16.)-0.5) + (topright.z+bottomleft.z)*0.25);
}
vec4 localtopright=positions[0];
vec4 localbottomleft=positions[7];
for (int i = 0; i<8; i++)
{
positions[i] *= vec4(0.25,0.25,0.5,1.);
positions[i].x += (localtopright.x-localbottomleft.x)*(0.25) * (mod(gl_LocalInvocationID.x,4.)-1.5) + (localtopright.x+localbottomleft.x)*0.375;
positions[i].y += (localtopright.y-localbottomleft.y)*(0.25) * (mod(floor(gl_LocalInvocationID.x/4.),4.)-1.5) + (localtopright.y+localbottomleft.y)*0.375;
positions[i].z += (localtopright.z-localbottomleft.z)*(0.5) * (floor(gl_LocalInvocationID.x/16.)-0.5) + (localtopright.z+localbottomleft.z)*0.25;
}
bvec3 signingvec=greaterThan((inverse(pc.world)*vec4(camera_uniforms.campos,1)).xyz,(positions[0].xyz+positions[7].xyz)/2);
float[2] check = scene(vec3[2](vec3(positions[0].xyz),vec3(positions[7].xyz)), true);
if ((check[0] < 0) && (check[1] > 0))
gl_MeshPrimitivesEXT[pindex+0].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+1].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+2].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+3].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+4].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+5].gl_PrimitiveID=GlobalInvocationIndex;
if ((check[0] < 0) && (check[1] > 0) && (gl_WorkGroupID.x < 32))
//if (true)
{
fragmentpassmasks.masks[GlobalInvocationIndex]=mask;
@@ -85,6 +112,14 @@ void main()
vertexOutput[vindex+5].position=(positions[5]);
vertexOutput[vindex+6].position=(positions[6]);
vertexOutput[vindex+7].position=(positions[7]);
/*vertexOutput[vindex+0].position=vec4(bottomleft,1.);
vertexOutput[vindex+1].position=vec4(bottomleft.x,bottomleft.y,topright.z,1.);
vertexOutput[vindex+2].position=vec4(bottomleft.x,topright.y,bottomleft.z,1.);
vertexOutput[vindex+3].position=vec4(bottomleft.x,topright.y,topright.z,1.);
vertexOutput[vindex+4].position=vec4(topright.x,bottomleft.y,bottomleft.z,1.);
vertexOutput[vindex+5].position=vec4(topright.x,bottomleft.y,topright.z,1.);
vertexOutput[vindex+6].position=vec4(topright.x,topright.y,bottomleft.z,1.);
vertexOutput[vindex+7].position=vec4(topright,1.);*/
if(signingvec.x){
gl_PrimitiveTriangleIndicesEXT[pindex+0]=uvec3(4,5,6)+uvec3(vindex);
@@ -108,22 +143,16 @@ void main()
gl_PrimitiveTriangleIndicesEXT[pindex+5]=uvec3(2,4,6)+uvec3(vindex);
}
gl_MeshPrimitivesEXT[pindex+0].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+1].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+2].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+3].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+4].gl_PrimitiveID=GlobalInvocationIndex;
gl_MeshPrimitivesEXT[pindex+5].gl_PrimitiveID=GlobalInvocationIndex;
} else
{
gl_MeshVerticesEXT[vindex+0].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+1].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+2].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+3].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+4].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+5].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+6].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+7].gl_Position=vec4(0,0,0,1);
gl_MeshVerticesEXT[vindex+0].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+1].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+2].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+3].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+4].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+5].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+6].gl_Position=mvp*(vec4(0,0,0,1));
gl_MeshVerticesEXT[vindex+7].gl_Position=mvp*(vec4(0,0,0,1));
gl_PrimitiveTriangleIndicesEXT[pindex+0]=uvec3(vindex);
gl_PrimitiveTriangleIndicesEXT[pindex+1]=uvec3(vindex);
@@ -131,12 +160,5 @@ void main()
gl_PrimitiveTriangleIndicesEXT[pindex+3]=uvec3(vindex);
gl_PrimitiveTriangleIndicesEXT[pindex+4]=uvec3(vindex);
gl_PrimitiveTriangleIndicesEXT[pindex+5]=uvec3(vindex);
gl_MeshPrimitivesEXT[pindex+0].gl_PrimitiveID=0;
gl_MeshPrimitivesEXT[pindex+1].gl_PrimitiveID=0;
gl_MeshPrimitivesEXT[pindex+2].gl_PrimitiveID=0;
gl_MeshPrimitivesEXT[pindex+3].gl_PrimitiveID=0;
gl_MeshPrimitivesEXT[pindex+4].gl_PrimitiveID=0;
gl_MeshPrimitivesEXT[pindex+5].gl_PrimitiveID=0;
}
}
+92
View File
@@ -0,0 +1,92 @@
// Implicit Mesh shader
#version 460
#extension GL_EXT_mesh_shader:require
#define DescriptionIndex gl_WorkGroupID.x
#include "include.glsl"
#include "intervals.glsl"
layout(local_size_x=32,local_size_y=1,local_size_z=1)in;
struct MeshMasks
{
uint8_t masks[32][masklen]; //928
uint8_t enabled[32]; //32
vec3 bottomleft; //12
vec3 topright; //12
uint globalindex; //4
//uint objectindex; //4
}; //total = 992 bytes
taskPayloadSharedEXT MeshMasks meshmasks;
shared uint index;
void main()
{
//clear_stacks();
default_mask();
meshmasks.masks[gl_LocalInvocationID.x] = mask;
if (gl_LocalInvocationID.x==0)
{
index=0;
}
#define CLIPCHECK 65536
float[6]bounds={
CLIPCHECK-sceneoverride(vec3(CLIPCHECK,0,0),false),
CLIPCHECK-sceneoverride(vec3(0,CLIPCHECK,0),false),
CLIPCHECK-sceneoverride(vec3(0,0,CLIPCHECK),false),
-CLIPCHECK+sceneoverride(vec3(-CLIPCHECK,0,0),false),
-CLIPCHECK+sceneoverride(vec3(0,-CLIPCHECK,0),false),
-CLIPCHECK+sceneoverride(vec3(0,0,-CLIPCHECK),false),
};
/*const float[6]bounds={
1,1,1,-1,-1,-1,
};*/
vec3 bottomleft = vec3(bounds[3],bounds[4],bounds[5]);
vec3 topright = vec3(bounds[0],bounds[1],bounds[2]);
vec3 center = (topright + bottomleft) / 2;
#define adjust(var) var -= center;\
var *= vec3(0.25,0.25,0.25);\
var.x += (bounds[0]-bounds[3]) * 0.25 * (mod(gl_LocalInvocationID.x,4.)-1.5) ;\
var.y += (bounds[1]-bounds[4]) * 0.25 * (mod(floor(gl_LocalInvocationID.x/4.),4.)-1.5) ;\
var.z += (bounds[2]-bounds[5]) * 0.25 * (floor(gl_LocalInvocationID.x/16.)-1.5+gl_WorkGroupID.z*2.) ;\
var += center;
adjust(bottomleft);
adjust(topright);
barrier();
float[2] check = scene(vec3[2](bottomleft,topright), false);
//float[2] check = scene(vec3[2](vec3(bounds[3],bounds[4],bounds[5]),vec3(bounds[0],bounds[1],bounds[2])), false);
if ((check[0] < 0) && (check[1] > 0))
//if ((bottomleft.x >= -1) && (bottomleft.y >= -1) && (bottomleft.z >= -1) && (topright.x <= 1) && (topright.y <= 1) && (topright.z <= 1))
//if ((gl_LocalInvocationID.x == 0) && (bottomleft.x >= 0))
{
uint localindex = atomicAdd(index, 1);
//if (localindex < 32) {
meshmasks.masks[localindex]=mask;
meshmasks.enabled[localindex]=uint8_t(gl_LocalInvocationID.x);
//}
}
if (gl_LocalInvocationID.x==0)
{
meshmasks.bottomleft = vec3(bounds[3],bounds[4],(bounds[5]*0.5) + ((bounds[2]-bounds[5])*0.5 * (-0.5+gl_WorkGroupID.z)));
meshmasks.topright = vec3(bounds[0],bounds[1],(bounds[2]*0.5) + ((bounds[2]-bounds[5])*0.5 * (-0.5+gl_WorkGroupID.z)));
meshmasks.globalindex = gl_WorkGroupID.x*2+gl_WorkGroupID.z;
//meshmasks.objectindex = DescriptionIndex;
}
barrier();
EmitMeshTasksEXT(index,1,1);
}
+219 -219
View File
@@ -1,219 +1,219 @@
const uint OPNop=__LINE__-1;
const uint OPStop=__LINE__-1;
const uint OPAddFloatFloat=__LINE__-1;
const uint OPAddVec2Vec2=__LINE__-1;
const uint OPAddVec2Float=__LINE__-1;
const uint OPAddVec3Vec3=__LINE__-1;
const uint OPAddVec3Float=__LINE__-1;
const uint OPAddVec4Vec4=__LINE__-1;
const uint OPAddVec4Float=__LINE__-1;
const uint OPSubFloatFloat=__LINE__-1;
const uint OPSubVec2Vec2=__LINE__-1;
const uint OPSubVec2Float=__LINE__-1;
const uint OPSubVec3Vec3=__LINE__-1;
const uint OPSubVec3Float=__LINE__-1;
const uint OPSubVec4Vec4=__LINE__-1;
const uint OPSubVec4Float=__LINE__-1;
const uint OPMulFloatFloat=__LINE__-1;
const uint OPMulVec2Vec2=__LINE__-1;
const uint OPMulVec2Float=__LINE__-1;
const uint OPMulVec3Vec3=__LINE__-1;
const uint OPMulVec3Float=__LINE__-1;
const uint OPMulVec4Vec4=__LINE__-1;
const uint OPMulVec4Float=__LINE__-1;
const uint OPDivFloatFloat=__LINE__-1;
const uint OPDivVec2Vec2=__LINE__-1;
const uint OPDivVec2Float=__LINE__-1;
const uint OPDivVec3Vec3=__LINE__-1;
const uint OPDivVec3Float=__LINE__-1;
const uint OPDivVec4Vec4=__LINE__-1;
const uint OPDivVec4Float=__LINE__-1;
const uint OPModFloatFloat=__LINE__-1;
const uint OPModVec2Vec2=__LINE__-1;
const uint OPModVec2Float=__LINE__-1;
const uint OPModVec3Vec3=__LINE__-1;
const uint OPModVec3Float=__LINE__-1;
const uint OPModVec4Vec4=__LINE__-1;
const uint OPModVec4Float=__LINE__-1;
const uint OPPowFloatFloat=__LINE__-1;
const uint OPPowVec2Vec2=__LINE__-1;
const uint OPPowVec3Vec3=__LINE__-1;
const uint OPPowVec4Vec4=__LINE__-1;
const uint OPCrossVec3=__LINE__-1;
const uint OPDotVec2=__LINE__-1;
const uint OPDotVec3=__LINE__-1;
const uint OPDotVec4=__LINE__-1;
const uint OPLengthVec2=__LINE__-1;
const uint OPLengthVec3=__LINE__-1;
const uint OPLengthVec4=__LINE__-1;
const uint OPDistanceVec2=__LINE__-1;
const uint OPDistanceVec3=__LINE__-1;
const uint OPDistanceVec4=__LINE__-1;
const uint OPNormalizeVec2=__LINE__-1;
const uint OPNormalizeVec3=__LINE__-1;
const uint OPNormalizeVec4=__LINE__-1;
const uint OPAbsFloat=__LINE__-1;
const uint OPSignFloat=__LINE__-1;
const uint OPFloorFloat=__LINE__-1;
const uint OPCeilFloat=__LINE__-1;
const uint OPFractFloat=__LINE__-1;
const uint OPSqrtFloat=__LINE__-1;
const uint OPInverseSqrtFloat=__LINE__-1;
const uint OPExpFloat=__LINE__-1;
const uint OPExp2Float=__LINE__-1;
const uint OPLogFloat=__LINE__-1;
const uint OPLog2Float=__LINE__-1;
const uint OPSinFloat=__LINE__-1;
const uint OPCosFloat=__LINE__-1;
const uint OPTanFloat=__LINE__-1;
const uint OPAsinFloat=__LINE__-1;
const uint OPAcosFloat=__LINE__-1;
const uint OPAtanFloat=__LINE__-1;
const uint OPMinFloat=__LINE__-1;
const uint OPMaxFloat=__LINE__-1;
const uint OPSmoothMinFloat=__LINE__-1;
const uint OPSmoothMaxFloat=__LINE__-1;
const uint OPMinMaterialFloat=__LINE__-1;
const uint OPMaxMaterialFloat=__LINE__-1;
const uint OPSmoothMinMaterialFloat=__LINE__-1;
const uint OPSmoothMaxMaterialFloat=__LINE__-1;
const uint OPDupFloat=__LINE__-1;
const uint OPDup2Float=__LINE__-1;
const uint OPDup3Float=__LINE__-1;
const uint OPDup4Float=__LINE__-1;
const uint OPAbsVec2=__LINE__-1;
const uint OPSignVec2=__LINE__-1;
const uint OPFloorVec2=__LINE__-1;
const uint OPCeilVec2=__LINE__-1;
const uint OPFractVec2=__LINE__-1;
const uint OPSqrtVec2=__LINE__-1;
const uint OPInverseSqrtVec2=__LINE__-1;
const uint OPExpVec2=__LINE__-1;
const uint OPExp2Vec2=__LINE__-1;
const uint OPLogVec2=__LINE__-1;
const uint OPLog2Vec2=__LINE__-1;
const uint OPSinVec2=__LINE__-1;
const uint OPCosVec2=__LINE__-1;
const uint OPTanVec2=__LINE__-1;
const uint OPAsinVec2=__LINE__-1;
const uint OPAcosVec2=__LINE__-1;
const uint OPAtanVec2=__LINE__-1;
const uint OPMinVec2=__LINE__-1;
const uint OPMaxVec2=__LINE__-1;
const uint OPDupVec2=__LINE__-1;
const uint OPDup2Vec2=__LINE__-1;
const uint OPDup3Vec2=__LINE__-1;
const uint OPDup4Vec2=__LINE__-1;
const uint OPAbsVec3=__LINE__-1;
const uint OPSignVec3=__LINE__-1;
const uint OPFloorVec3=__LINE__-1;
const uint OPCeilVec3=__LINE__-1;
const uint OPFractVec3=__LINE__-1;
const uint OPSqrtVec3=__LINE__-1;
const uint OPInverseSqrtVec3=__LINE__-1;
const uint OPExpVec3=__LINE__-1;
const uint OPExp2Vec3=__LINE__-1;
const uint OPLogVec3=__LINE__-1;
const uint OPLog2Vec3=__LINE__-1;
const uint OPSinVec3=__LINE__-1;
const uint OPCosVec3=__LINE__-1;
const uint OPTanVec3=__LINE__-1;
const uint OPAsinVec3=__LINE__-1;
const uint OPAcosVec3=__LINE__-1;
const uint OPAtanVec3=__LINE__-1;
const uint OPMinVec3=__LINE__-1;
const uint OPMaxVec3=__LINE__-1;
const uint OPDupVec3=__LINE__-1;
const uint OPDup2Vec3=__LINE__-1;
const uint OPDup3Vec3=__LINE__-1;
const uint OPDup4Vec3=__LINE__-1;
const uint OPAbsVec4=__LINE__-1;
const uint OPSignVec4=__LINE__-1;
const uint OPFloorVec4=__LINE__-1;
const uint OPCeilVec4=__LINE__-1;
const uint OPFractVec4=__LINE__-1;
const uint OPSqrtVec4=__LINE__-1;
const uint OPInverseSqrtVec4=__LINE__-1;
const uint OPExpVec4=__LINE__-1;
const uint OPExp2Vec4=__LINE__-1;
const uint OPLogVec4=__LINE__-1;
const uint OPLog2Vec4=__LINE__-1;
const uint OPSinVec4=__LINE__-1;
const uint OPCosVec4=__LINE__-1;
const uint OPTanVec4=__LINE__-1;
const uint OPAsinVec4=__LINE__-1;
const uint OPAcosVec4=__LINE__-1;
const uint OPAtanVec4=__LINE__-1;
const uint OPMinVec4=__LINE__-1;
const uint OPMaxVec4=__LINE__-1;
const uint OPDupVec4=__LINE__-1;
const uint OPDup2Vec4=__LINE__-1;
const uint OPDup3Vec4=__LINE__-1;
const uint OPDup4Vec4=__LINE__-1;
const uint OPPromoteFloatFloatVec2=__LINE__-1;
const uint OPPromoteFloatFloatFloatVec3=__LINE__-1;
const uint OPPromoteFloatFloatFloatFloatVec4=__LINE__-1;
const uint OPPromoteVec2FloatVec3=__LINE__-1;
const uint OPPromoteVec2FloatFloatVec4=__LINE__-1;
const uint OPPromoteVec2Vec2Vec4=__LINE__-1;
const uint OPPromoteVec3FloatVec4=__LINE__-1;
const uint OPAcoshFloat=__LINE__-1;
const uint OPAcoshVec2=__LINE__-1;
const uint OPAcoshVec3=__LINE__-1;
const uint OPAcoshVec4=__LINE__-1;
const uint OPAsinhFloat=__LINE__-1;
const uint OPAsinhVec2=__LINE__-1;
const uint OPAsinhVec3=__LINE__-1;
const uint OPAsinhVec4=__LINE__-1;
const uint OPAtanhFloat=__LINE__-1;
const uint OPAtanhVec2=__LINE__-1;
const uint OPAtanhVec3=__LINE__-1;
const uint OPAtanhVec4=__LINE__-1;
const uint OPCoshFloat=__LINE__-1;
const uint OPCoshVec2=__LINE__-1;
const uint OPCoshVec3=__LINE__-1;
const uint OPCoshVec4=__LINE__-1;
const uint OPSinhFloat=__LINE__-1;
const uint OPSinhVec2=__LINE__-1;
const uint OPSinhVec3=__LINE__-1;
const uint OPSinhVec4=__LINE__-1;
const uint OPTanhFloat=__LINE__-1;
const uint OPTanhVec2=__LINE__-1;
const uint OPTanhVec3=__LINE__-1;
const uint OPTanhVec4=__LINE__-1;
const uint OPRoundFloat=__LINE__-1;
const uint OPRoundVec2=__LINE__-1;
const uint OPRoundVec3=__LINE__-1;
const uint OPRoundVec4=__LINE__-1;
const uint OPTruncFloat=__LINE__-1;
const uint OPTruncVec2=__LINE__-1;
const uint OPTruncVec3=__LINE__-1;
const uint OPTruncVec4=__LINE__-1;
const uint OPFMAFloat=__LINE__-1;
const uint OPFMAVec2=__LINE__-1;
const uint OPFMAVec3=__LINE__-1;
const uint OPFMAVec4=__LINE__-1;
const uint OPClampFloatFloat=__LINE__-1;
const uint OPClampVec2Vec2=__LINE__-1;
const uint OPClampVec2Float=__LINE__-1;
const uint OPClampVec3Vec3=__LINE__-1;
const uint OPClampVec3Float=__LINE__-1;
const uint OPClampVec4Vec4=__LINE__-1;
const uint OPClampVec4Float=__LINE__-1;
const uint OPMixFloatFloat=__LINE__-1;
const uint OPMixVec2Vec2=__LINE__-1;
const uint OPMixVec2Float=__LINE__-1;
const uint OPMixVec3Vec3=__LINE__-1;
const uint OPMixVec3Float=__LINE__-1;
const uint OPMixVec4Vec4=__LINE__-1;
const uint OPMixVec4Float=__LINE__-1;
const uint OPSquareFloat=__LINE__-1;
const uint OPCubeFloat=__LINE__-1;
const uint OPSquareVec2=__LINE__-1;
const uint OPCubeVec2=__LINE__-1;
const uint OPSquareVec3=__LINE__-1;
const uint OPCubeVec3=__LINE__-1;
const uint OPSquareVec4=__LINE__-1;
const uint OPCubeVec4=__LINE__-1;
const uint OPSDFSphere=__LINE__-1;
const uint OPInvalid=__LINE__-1;
const uint OPNop=__LINE__-1; // //
const uint OPStop=__LINE__-1; //F //
const uint OPAddFloatFloat=__LINE__-1; //F F //F
const uint OPAddVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPAddVec2Float=__LINE__-1; //V2 F //V2
const uint OPAddVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPAddVec3Float=__LINE__-1; //V3 F //V3
const uint OPAddVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPAddVec4Float=__LINE__-1; //V4 F //V4
const uint OPSubFloatFloat=__LINE__-1; //F F //F
const uint OPSubVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPSubVec2Float=__LINE__-1; //V2 F //V2
const uint OPSubVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPSubVec3Float=__LINE__-1; //V3 F //V3
const uint OPSubVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPSubVec4Float=__LINE__-1; //V4 F //V4
const uint OPMulFloatFloat=__LINE__-1; //F F //F
const uint OPMulVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPMulVec2Float=__LINE__-1; //V2 F //V2
const uint OPMulVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPMulVec3Float=__LINE__-1; //V3 F //V3
const uint OPMulVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPMulVec4Float=__LINE__-1; //V4 F //V4
const uint OPDivFloatFloat=__LINE__-1; //F F //F
const uint OPDivVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPDivVec2Float=__LINE__-1; //V2 F //V2
const uint OPDivVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPDivVec3Float=__LINE__-1; //V3 F //V3
const uint OPDivVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPDivVec4Float=__LINE__-1; //V4 F //V4
const uint OPModFloatFloat=__LINE__-1; //F F //F
const uint OPModVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPModVec2Float=__LINE__-1; //V2 F //V2
const uint OPModVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPModVec3Float=__LINE__-1; //V3 F //V3
const uint OPModVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPModVec4Float=__LINE__-1; //V4 F //V4
const uint OPPowFloatFloat=__LINE__-1; //F F //F
const uint OPPowVec2Vec2=__LINE__-1; //V2 V2 //V2
const uint OPPowVec3Vec3=__LINE__-1; //V3 V3 //V3
const uint OPPowVec4Vec4=__LINE__-1; //V4 V4 //V4
const uint OPCrossVec3=__LINE__-1; //V3 V3 //V3
const uint OPDotVec2=__LINE__-1; //V2 V2 //F
const uint OPDotVec3=__LINE__-1; //V3 V3 //F
const uint OPDotVec4=__LINE__-1; //V4 V4 //F
const uint OPLengthVec2=__LINE__-1; //V2 //F
const uint OPLengthVec3=__LINE__-1; //V3 //F
const uint OPLengthVec4=__LINE__-1; //V4 //F
const uint OPDistanceVec2=__LINE__-1; //V2 V2 //F
const uint OPDistanceVec3=__LINE__-1; //V3 V3 //F
const uint OPDistanceVec4=__LINE__-1; //V4 V4 //F
const uint OPNormalizeVec2=__LINE__-1; //V2 //V2
const uint OPNormalizeVec3=__LINE__-1; //V2 //V2
const uint OPNormalizeVec4=__LINE__-1; //V4 //V4
const uint OPAbsFloat=__LINE__-1; //F //F
const uint OPSignFloat=__LINE__-1; //F //F
const uint OPFloorFloat=__LINE__-1; //F //F
const uint OPCeilFloat=__LINE__-1; //F //F
const uint OPFractFloat=__LINE__-1; //F //F
const uint OPSqrtFloat=__LINE__-1; //F //F
const uint OPInverseSqrtFloat=__LINE__-1; //F //F
const uint OPExpFloat=__LINE__-1; //F //F
const uint OPExp2Float=__LINE__-1; //F //F
const uint OPLogFloat=__LINE__-1; //F //F
const uint OPLog2Float=__LINE__-1; //F //F
const uint OPSinFloat=__LINE__-1; //F //F
const uint OPCosFloat=__LINE__-1; //F //F
const uint OPTanFloat=__LINE__-1; //F //F
const uint OPAsinFloat=__LINE__-1; //F //F
const uint OPAcosFloat=__LINE__-1; //F //F
const uint OPAtanFloat=__LINE__-1; //F //F
const uint OPMinFloat=__LINE__-1; //F F //F
const uint OPMaxFloat=__LINE__-1; //F F //F
const uint OPSmoothMinFloat=__LINE__-1; //F F F //F
const uint OPSmoothMaxFloat=__LINE__-1; //F F F //F
const uint OPMinMaterialFloat=__LINE__-1; //F F //F
const uint OPMaxMaterialFloat=__LINE__-1; //F F //F
const uint OPSmoothMinMaterialFloat=__LINE__-1; //F F F //F
const uint OPSmoothMaxMaterialFloat=__LINE__-1; //F F F //F
const uint OPDupFloat=__LINE__-1; //F //F F
const uint OPDup2Float=__LINE__-1; //F //F F
const uint OPDup3Float=__LINE__-1; //F //F F
const uint OPDup4Float=__LINE__-1; //F //F F
const uint OPAbsVec2=__LINE__-1; //V2 //V2
const uint OPSignVec2=__LINE__-1; //V2 //V2
const uint OPFloorVec2=__LINE__-1; //V2 //V2
const uint OPCeilVec2=__LINE__-1; //V2 //V2
const uint OPFractVec2=__LINE__-1; //V2 //V2
const uint OPSqrtVec2=__LINE__-1; //V2 //V2
const uint OPInverseSqrtVec2=__LINE__-1; //V2 //V2
const uint OPExpVec2=__LINE__-1; //V2 //V2
const uint OPExp2Vec2=__LINE__-1; //V2 //V2
const uint OPLogVec2=__LINE__-1; //V2 //V2
const uint OPLog2Vec2=__LINE__-1; //V2 //V2
const uint OPSinVec2=__LINE__-1; //V2 //V2
const uint OPCosVec2=__LINE__-1; //V2 //V2
const uint OPTanVec2=__LINE__-1; //V2 //V2
const uint OPAsinVec2=__LINE__-1; //V2 //V2
const uint OPAcosVec2=__LINE__-1; //V2 //V2
const uint OPAtanVec2=__LINE__-1; //V2 //V2
const uint OPMinVec2=__LINE__-1; //V2 V2 //V2
const uint OPMaxVec2=__LINE__-1; //V2 V2 //V2
const uint OPDupVec2=__LINE__-1; //V2 //V2 V2
const uint OPDup2Vec2=__LINE__-1; //V2 //V2 V2
const uint OPDup3Vec2=__LINE__-1; //V2 //V2 V2
const uint OPDup4Vec2=__LINE__-1; //V2 //V2 V2
const uint OPAbsVec3=__LINE__-1; //V3 //V3
const uint OPSignVec3=__LINE__-1; //V3 //V3
const uint OPFloorVec3=__LINE__-1; //V3 //V3
const uint OPCeilVec3=__LINE__-1; //V3 //V3
const uint OPFractVec3=__LINE__-1; //V3 //V3
const uint OPSqrtVec3=__LINE__-1; //V3 //V3
const uint OPInverseSqrtVec3=__LINE__-1; //V3 //V3
const uint OPExpVec3=__LINE__-1; //V3 //V3
const uint OPExp2Vec3=__LINE__-1; //V3 //V3
const uint OPLogVec3=__LINE__-1; //V3 //V3
const uint OPLog2Vec3=__LINE__-1; //V3 //V3
const uint OPSinVec3=__LINE__-1; //V3 //V3
const uint OPCosVec3=__LINE__-1; //V3 //V3
const uint OPTanVec3=__LINE__-1; //V3 //V3
const uint OPAsinVec3=__LINE__-1; //V3 //V3
const uint OPAcosVec3=__LINE__-1; //V3 //V3
const uint OPAtanVec3=__LINE__-1; //V3 //V3
const uint OPMinVec3=__LINE__-1; //V3 V3 //V3
const uint OPMaxVec3=__LINE__-1; //V3 V3 //V3
const uint OPDupVec3=__LINE__-1; //V3 //V3 V3
const uint OPDup2Vec3=__LINE__-1; //V3 //V3 V3
const uint OPDup3Vec3=__LINE__-1; //V3 //V3 V3
const uint OPDup4Vec3=__LINE__-1; //V3 //V3 V3
const uint OPAbsVec4=__LINE__-1; //V4 //V4
const uint OPSignVec4=__LINE__-1; //V4 //V4
const uint OPFloorVec4=__LINE__-1; //V4 //V4
const uint OPCeilVec4=__LINE__-1; //V4 //V4
const uint OPFractVec4=__LINE__-1; //V4 //V4
const uint OPSqrtVec4=__LINE__-1; //V4 //V4
const uint OPInverseSqrtVec4=__LINE__-1; //V4 //V4
const uint OPExpVec4=__LINE__-1; //V4 //V4
const uint OPExp2Vec4=__LINE__-1; //V4 //V4
const uint OPLogVec4=__LINE__-1; //V4 //V4
const uint OPLog2Vec4=__LINE__-1; //V4 //V4
const uint OPSinVec4=__LINE__-1; //V4 //V4
const uint OPCosVec4=__LINE__-1; //V4 //V4
const uint OPTanVec4=__LINE__-1; //V4 //V4
const uint OPAsinVec4=__LINE__-1; //V4 //V4
const uint OPAcosVec4=__LINE__-1; //V4 //V4
const uint OPAtanVec4=__LINE__-1; //V4 //V4
const uint OPMinVec4=__LINE__-1; //V4 V4 //V4
const uint OPMaxVec4=__LINE__-1; //V4 V4 //V4
const uint OPDupVec4=__LINE__-1; //V4 //V4 V4
const uint OPDup2Vec4=__LINE__-1; //V4 //V4 V4
const uint OPDup3Vec4=__LINE__-1; //V4 //V4 V4
const uint OPDup4Vec4=__LINE__-1; //V4 //V4 V4
const uint OPPromoteFloatFloatVec2=__LINE__-1; //F F //V2
const uint OPPromoteFloatFloatFloatVec3=__LINE__-1; //F F F //V3
const uint OPPromoteFloatFloatFloatFloatVec4=__LINE__-1; //F F F F //V4
const uint OPPromoteVec2FloatVec3=__LINE__-1; //V2 F //V3
const uint OPPromoteVec2FloatFloatVec4=__LINE__-1; //V2 F F //V4
const uint OPPromoteVec2Vec2Vec4=__LINE__-1; //V2 V2 //V4
const uint OPPromoteVec3FloatVec4=__LINE__-1; //V3 F //V4
const uint OPAcoshFloat=__LINE__-1; //F //F
const uint OPAcoshVec2=__LINE__-1; //V2 //V2
const uint OPAcoshVec3=__LINE__-1; //V3 //V3
const uint OPAcoshVec4=__LINE__-1; //V4 //V4
const uint OPAsinhFloat=__LINE__-1; //F //F
const uint OPAsinhVec2=__LINE__-1; //V2 //V2
const uint OPAsinhVec3=__LINE__-1; //V3 //V3
const uint OPAsinhVec4=__LINE__-1; //V4 //V4
const uint OPAtanhFloat=__LINE__-1; //F //F
const uint OPAtanhVec2=__LINE__-1; //V2 //V2
const uint OPAtanhVec3=__LINE__-1; //V3 //V3
const uint OPAtanhVec4=__LINE__-1; //V4 //V4
const uint OPCoshFloat=__LINE__-1; //F //F
const uint OPCoshVec2=__LINE__-1; //V2 //V2
const uint OPCoshVec3=__LINE__-1; //V3 //V3
const uint OPCoshVec4=__LINE__-1; //V4 //V4
const uint OPSinhFloat=__LINE__-1; //F //F
const uint OPSinhVec2=__LINE__-1; //V2 //V2
const uint OPSinhVec3=__LINE__-1; //V3 //V3
const uint OPSinhVec4=__LINE__-1; //V4 //V4
const uint OPTanhFloat=__LINE__-1; //F //F
const uint OPTanhVec2=__LINE__-1; //V2 //V2
const uint OPTanhVec3=__LINE__-1; //V3 //V3
const uint OPTanhVec4=__LINE__-1; //V4 //V4
const uint OPRoundFloat=__LINE__-1; //F //F
const uint OPRoundVec2=__LINE__-1; //V2 //V2
const uint OPRoundVec3=__LINE__-1; //V3 //V3
const uint OPRoundVec4=__LINE__-1; //V4 //V4
const uint OPTruncFloat=__LINE__-1; //F //F
const uint OPTruncVec2=__LINE__-1; //V2 //V2
const uint OPTruncVec3=__LINE__-1; //V3 //V3
const uint OPTruncVec4=__LINE__-1; //V4 //V4
const uint OPFMAFloat=__LINE__-1; //F //F
const uint OPFMAVec2=__LINE__-1; //V2 //V2
const uint OPFMAVec3=__LINE__-1; //V3 //V3
const uint OPFMAVec4=__LINE__-1; //V4 //V4
const uint OPClampFloatFloat=__LINE__-1; //F F F //F
const uint OPClampVec2Vec2=__LINE__-1; //V2 V2 V2 //V2
const uint OPClampVec2Float=__LINE__-1; //V2 F F //V2
const uint OPClampVec3Vec3=__LINE__-1; //V3 V3 V3 //V3
const uint OPClampVec3Float=__LINE__-1; //V3 F F //V3
const uint OPClampVec4Vec4=__LINE__-1; //V4 V4 V4 //V4
const uint OPClampVec4Float=__LINE__-1; //V4 F F //V4
const uint OPMixFloatFloat=__LINE__-1; //F F F //F
const uint OPMixVec2Vec2=__LINE__-1; //V2 V2 V2 //V2
const uint OPMixVec2Float=__LINE__-1; //V2 V2 F //V2
const uint OPMixVec3Vec3=__LINE__-1; //V3 V3 V3 //V3
const uint OPMixVec3Float=__LINE__-1; //V3 V3 F //V3
const uint OPMixVec4Vec4=__LINE__-1; //V4 V4 V4 //V4
const uint OPMixVec4Float=__LINE__-1; //V4 V4 F //V4
const uint OPSquareFloat=__LINE__-1; //F //F
const uint OPCubeFloat=__LINE__-1; //F //F
const uint OPSquareVec2=__LINE__-1; //V2 //V2
const uint OPCubeVec2=__LINE__-1; //V2 //V2
const uint OPSquareVec3=__LINE__-1; //V3 //V3
const uint OPCubeVec3=__LINE__-1; //V3 //V3
const uint OPSquareVec4=__LINE__-1; //V4 //V4
const uint OPCubeVec4=__LINE__-1; //V4 //V4
const uint OPSDFSphere=__LINE__-1; //F V3 //F
const uint OPInvalid=__LINE__-1; // //
+3092
View File
File diff suppressed because it is too large Load Diff
+48 -30
View File
@@ -17,8 +17,10 @@ struct Description{
uint mat2s;
uint mat3s;
uint mat4s;
uint mats;
uint dependencies;
};
Description desc;
layout(set=0,binding=2)restrict readonly buffer SceneDescription{
Description desc[];
@@ -56,7 +58,7 @@ layout(set=0,binding=12)restrict readonly buffer DepInfo{
}depinfo;
// unpack integers
#define get_caches u32vec4 major_unpack=scenes.opcodes[major_position+desc.scene];\
#define get_caches u32vec4 major_unpack=scenes.opcodes[desc.scene+major_position];\
minor_integer_cache[0]=major_unpack.x&65535;\
minor_integer_cache[1]=major_unpack.x>>16;\
minor_integer_cache[2]=major_unpack.y&65535;\
@@ -88,6 +90,7 @@ uint vec4_const_head=0;
uint mat2_const_head=0;
uint mat3_const_head=0;
uint mat4_const_head=0;
uint mat_const_head=0;
void push_float(float f[2]){
float_stack[float_stack_head++]=f;
@@ -95,7 +98,7 @@ void push_float(float f[2]){
float[2]pull_float(bool c){
if (c) {
float f = fconst.floats[float_const_head++];
float f = fconst.floats[desc.floats+float_const_head++];
return float[2](f,f);
}
else {
@@ -104,7 +107,7 @@ float[2]pull_float(bool c){
}
float cpull_float(){
return fconst.floats[float_const_head++];
return fconst.floats[desc.floats+float_const_head++];
}
void push_vec2(vec2 f[2]){
@@ -113,7 +116,7 @@ void push_vec2(vec2 f[2]){
vec2[2]pull_vec2(bool c){
if (c) {
vec2 f = v2const.vec2s[vec2_const_head++];
vec2 f = v2const.vec2s[desc.vec2s+vec2_const_head++];
return vec2[2](f,f);
}
else {
@@ -122,7 +125,7 @@ vec2[2]pull_vec2(bool c){
}
vec2 cpull_vec2(){
return v2const.vec2s[vec2_const_head++];
return v2const.vec2s[desc.vec2s+vec2_const_head++];
}
void push_vec3(vec3 f[2]){
@@ -131,7 +134,7 @@ void push_vec3(vec3 f[2]){
vec3[2]pull_vec3(bool c){
if (c) {
vec3 f = v3const.vec3s[vec3_const_head++];
vec3 f = v3const.vec3s[desc.vec3s+vec3_const_head++];
return vec3[2](f,f);
}
else {
@@ -140,7 +143,7 @@ vec3[2]pull_vec3(bool c){
}
vec3 cpull_vec3(){
return v3const.vec3s[vec3_const_head++];
return v3const.vec3s[desc.vec3s+vec3_const_head++];
}
void push_vec4(vec4 f[2]){
@@ -149,7 +152,7 @@ void push_vec4(vec4 f[2]){
vec4[2]pull_vec4(bool c){
if (c) {
vec4 f = v4const.vec4s[vec4_const_head++];
vec4 f = v4const.vec4s[desc.vec4s+vec4_const_head++];
return vec4[2](f,f);
}
else {
@@ -158,7 +161,7 @@ vec4[2]pull_vec4(bool c){
}
vec4 cpull_vec4(){
return v4const.vec4s[vec4_const_head++];
return v4const.vec4s[desc.vec4s+vec4_const_head++];
}
void push_mat2(mat2 f[2]){
@@ -167,7 +170,7 @@ void push_mat2(mat2 f[2]){
mat2[2]pull_mat2(bool c){
if (c) {
mat2 f = m2const.mat2s[mat2_const_head++];
mat2 f = m2const.mat2s[desc.mat2s+mat2_const_head++];
return mat2[2](f,f);
}
else {
@@ -176,7 +179,7 @@ mat2[2]pull_mat2(bool c){
}
mat2 cpull_mat2(){
return m2const.mat2s[mat2_const_head++];
return m2const.mat2s[desc.mat2s+mat2_const_head++];
}
void push_mat3(mat3 f[2]){
@@ -185,7 +188,7 @@ void push_mat3(mat3 f[2]){
mat3[2]pull_mat3(bool c){
if (c) {
mat3 f = m3const.mat3s[mat3_const_head++];
mat3 f = m3const.mat3s[desc.mat3s+mat3_const_head++];
return mat3[2](f,f);
}
else {
@@ -194,7 +197,7 @@ mat3[2]pull_mat3(bool c){
}
mat3 cpull_mat3(){
return m3const.mat3s[mat3_const_head++];
return m3const.mat3s[desc.mat3s+mat3_const_head++];
}
void push_mat4(mat4 f[2]){
@@ -203,7 +206,7 @@ void push_mat4(mat4 f[2]){
mat4[2]pull_mat4(bool c){
if (c) {
mat4 f = m4const.mat4s[mat4_const_head++];
mat4 f = m4const.mat4s[desc.mat4s+mat4_const_head++];
return mat4[2](f,f);
}
else {
@@ -212,7 +215,7 @@ mat4[2]pull_mat4(bool c){
}
mat4 cpull_mat4(){
return m4const.mat4s[mat4_const_head++];
return m4const.mat4s[desc.mat4s+mat4_const_head++];
}
void clear_stacks()
@@ -231,6 +234,7 @@ void clear_stacks()
mat2_const_head=0;
mat3_const_head=0;
mat4_const_head=0;
mat_const_head=0;
}
const int masklen = 29;
@@ -554,7 +558,6 @@ void default_mask()
in1[1]=trunc(in1[1]);\
}
Description desc;
//0 - prune nothing
//1 - prune myself
//2 - prune myself and children
@@ -694,21 +697,21 @@ void pruneself (int pos) {
break;\
}
#define minpruning if (all(lessThan(in1[1],in2[0]))) {\
#define minpruning if (prune) { if (all(lessThan(in1[1],in2[0]))) {\
prunesome(OPPos,bool[6](false,true,false,false,false,false));\
passthroughself(OPPos);\
} else if (all(lessThan(in2[1],in1[0]))) {\
prunesome(OPPos,bool[6](true,false,false,false,false,false));\
passthroughself(OPPos);\
}
}}
#define maxpruning if (all(greaterThan(in1[0],in2[1]))) {\
#define maxpruning if (prune) { if (all(greaterThan(in1[0],in2[1]))) {\
prunesome(OPPos,bool[6](false,true,false,false,false,false));\
passthroughself(OPPos);\
} else if (all(greaterThan(in2[0],in1[1]))) {\
prunesome(OPPos,bool[6](true,false,false,false,false,false));\
passthroughself(OPPos);\
}
}}
#ifdef debug
vec3 scene(vec3 p[2], bool prune)
@@ -721,7 +724,7 @@ float[2]scene(vec3 p[2], bool prune)
uint minor_integer_cache[8];
desc = scene_description.desc[gl_GlobalInvocationID.x];
desc = scene_description.desc[(DescriptionIndex)+1];
clear_stacks();
push_vec3(p);
@@ -732,15 +735,22 @@ float[2]scene(vec3 p[2], bool prune)
if(minor_position==0){
get_caches;
}
/*#ifdef debug
if((minor_integer_cache[minor_position]&1023)==OPStop) {
#ifdef debug
/*if((minor_integer_cache[minor_position]&1023)==OPStop) {
return vec3(0.,0.,1.);
}
if((minor_integer_cache[minor_position]&1023)==OPSDFSphere) {
return vec3(1.,0.,0.);
}*/
/*if((minor_integer_cache[minor_position] & (1 << (15 - 1))) > 0) {
return vec3(1.,0.,0.);
}
return vec3(0.,1.,0.);
#endif*/
if((minor_integer_cache[minor_position] & (1 << (15 - 0))) > 0) {
return vec3(0.,0.,1.);
}*/
//return vec3(0.,1.,0.);
return vec3(desc.floats,desc.scene,DescriptionIndex);
#endif
switch(minor_integer_cache[minor_position]&1023)
{
@@ -1552,6 +1562,7 @@ float[2]scene(vec3 p[2], bool prune)
inputmask2(float_const_head,float_const_head);
float[2]in1=pull_float(ifconst(0));
float[2]in2=pull_float(ifconst(1));
if (prune) {
if (in1[1] < in2[0])
{
prunesome(OPPos,bool[6](false,true,false,false,false,false));
@@ -1560,7 +1571,7 @@ float[2]scene(vec3 p[2], bool prune)
{
prunesome(OPPos,bool[6](true,false,false,false,false,false));
passthroughself(OPPos);
}
}}
float[2]temp;
minimum;
push_float(in1);
@@ -1571,6 +1582,7 @@ float[2]scene(vec3 p[2], bool prune)
inputmask2(float_const_head,float_const_head);
float[2]in1=pull_float(ifconst(0));
float[2]in2=pull_float(ifconst(1));
if (prune) {
if (in1[0] > in2[1])
{
prunesome(OPPos,bool[6](false,true,false,false,false,false));
@@ -1579,7 +1591,7 @@ float[2]scene(vec3 p[2], bool prune)
{
prunesome(OPPos,bool[6](true,false,false,false,false,false));
passthroughself(OPPos);
}
}}
float[2]temp;
maximum;
push_float(in1);
@@ -3072,9 +3084,13 @@ float[2]scene(vec3 p[2], bool prune)
{
inputmask2(float_const_head,vec3_const_head);
float[2] in2=pull_float(ifconst(0));
/*#ifdef debug
return vec3(in2[0],in2[1],0.);
#endif*/
#ifdef debug
if (in2[0] == in2[1] && in2[0] == 0.2)
{return vec3(in2[0],in2[1],0.);
}else{
return vec3(in2[0],in2[1],1.);
}
#endif
vec3[2] in1=pull_vec3(ifconst(1));
/*#ifdef debug
return in1[0];
@@ -3096,6 +3112,7 @@ float[2]scene(vec3 p[2], bool prune)
case OPNop:
break;
case OPStop:
if (prune)
pruneall(uint8_t((major_position<<3)|minor_position));
#ifdef debug
return vec3(pull_float(ifconst(0))[0]);
@@ -3118,6 +3135,7 @@ float[2]scene(vec3 p[2], bool prune)
major_position++;
if(major_position==masklen)
{
if (prune)
pruneall(uint8_t((masklen*8)));
#ifdef debug
return vec3(pull_float(false)[0]);
+414 -142
View File
@@ -3,6 +3,8 @@ use cgmath::{
Deg, EuclideanSpace, Euler, Matrix2, Matrix3, Matrix4, Point3, Rad, SquareMatrix, Vector2,
Vector3, Vector4,
};
use instruction_set::InputTypes;
use vulkano::command_buffer::{CopyBufferInfo, PrimaryCommandBufferAbstract};
use std::io::Cursor;
use std::{sync::Arc, time::Instant};
use vulkano::buffer::allocator::{SubbufferAllocator, SubbufferAllocatorCreateInfo};
@@ -10,7 +12,7 @@ use vulkano::buffer::{Buffer, BufferAllocateInfo, Subbuffer};
use vulkano::command_buffer::allocator::StandardCommandBufferAllocator;
use vulkano::descriptor_set::allocator::StandardDescriptorSetAllocator;
use vulkano::descriptor_set::{PersistentDescriptorSet, WriteDescriptorSet};
use vulkano::device::{DeviceOwned, Features, QueueFlags};
use vulkano::device::{DeviceOwned, Features, QueueFlags, Queue};
use vulkano::format::Format;
use vulkano::half::f16;
use vulkano::image::view::ImageViewCreateInfo;
@@ -105,7 +107,7 @@ fn main() {
..DeviceExtensions::empty()
};
let (physical_device, queue_family_index) = instance
let (physical_device, (queue_family_index, transfer_index)) = instance
.enumerate_physical_devices()
.unwrap()
.filter(|p| p.supported_extensions().contains(&device_extensions))
@@ -116,8 +118,21 @@ fn main() {
.position(|(i, q)| {
q.queue_flags.intersects(QueueFlags::GRAPHICS)
&& p.surface_support(i as u32, &surface).unwrap_or(false)
})
.map(|i| (p, i as u32))
}).and_then( | graphics| {
p.queue_family_properties()
.iter()
.enumerate()
.position(|(i, q)| {
q.queue_flags.intersects(QueueFlags::TRANSFER) && i != graphics
}).or_else( || {
p.queue_family_properties()
.iter()
.enumerate()
.position(|(i, q)| {
q.queue_flags.intersects(QueueFlags::TRANSFER)
})}
).map(|i| (graphics as u32, i as u32))}).map(|i| (p, i))
})
.min_by_key(|(p, _)| {
// We assign a lower score to device types that are likely to be faster/better.
@@ -146,6 +161,10 @@ fn main() {
queue_create_infos: vec![QueueCreateInfo {
queue_family_index,
..Default::default()
},
QueueCreateInfo {
queue_family_index: transfer_index,
..Default::default()
}],
enabled_features: Features {
mesh_shader: true,
@@ -165,6 +184,7 @@ fn main() {
.expect("Unable to initialize device");
let queue = queues.next().expect("Unable to retrieve queues");
let transfer_queue = queues.next().expect("Unable to retrieve queues");
let (mut swapchain, images) = {
let surface_capabilities = device
@@ -251,6 +271,20 @@ fn main() {
}
}
mod implicit_ts {
vulkano_shaders::shader! {
ty: "task",
path: "src/implicit.task.glsl",
types_meta: {
use bytemuck::{Pod, Zeroable};
#[derive(Clone, Copy, Zeroable, Pod, Debug)]
},
vulkan_version: "1.2",
spirv_version: "1.5"
}
}
mod implicit_fs {
vulkano_shaders::shader! {
ty: "fragment",
@@ -295,6 +329,13 @@ fn main() {
::std::sync::Arc<::vulkano::shader::ShaderModule>,
::vulkano::shader::ShaderCreationError,
>),
((implicit_ts::load)
as fn(
::std::sync::Arc<::vulkano::device::Device>,
) -> Result<
::std::sync::Arc<::vulkano::shader::ShaderModule>,
::vulkano::shader::ShaderCreationError,
>),
((implicit_fs::load)
as fn(
::std::sync::Arc<::vulkano::device::Device>,
@@ -313,7 +354,8 @@ fn main() {
let mesh_fs = pariter[1].clone();
let implicit_ms = pariter[2].clone();
let implicit_fs = pariter[3].clone();
let implicit_ts = pariter[3].clone();
let implicit_fs = pariter[4].clone();
drop(pariter);
@@ -367,6 +409,7 @@ fn main() {
&mesh_vs,
&mesh_fs,
&implicit_ms,
&implicit_ts,
&implicit_fs,
&images,
render_pass.clone(),
@@ -448,7 +491,7 @@ fn main() {
.lights
.push(Light::new([-4., 6., -8.], [8., 4., 1.], 0.05));
let fragment_masks_buffer = object_size_dependent_setup(&memory_allocator, &gstate);
let subbuffers = object_size_dependent_setup(memory_allocator.clone(), &gstate, &command_buffer_allocator, transfer_queue.clone());
let mut render_start = Instant::now();
@@ -547,6 +590,7 @@ fn main() {
&mesh_vs,
&mesh_fs,
&implicit_ms,
&implicit_ts,
&implicit_fs,
&new_images,
render_pass.clone(),
@@ -652,125 +696,6 @@ fn main() {
sub
};
let mut data = [[0u32; 4]; 29];
let parts = vec![
//CSGPart::opcode(InstructionSet::OPDupVec3, 0b000000),
//CSGPart::opcode(InstructionSet::OPSubVec3Vec3, 0b010000),
CSGPart::opcode(InstructionSet::OPSDFSphere, 0b100000),
//CSGPart::opcode(InstructionSet::OPAddVec3Vec3, 0b010000),
//CSGPart::opcode(InstructionSet::OPSDFSphere, 0b100000),
//CSGPart::opcode(InstructionSet::OPSmoothMinFloat, 0b000000),
CSGPart::opcode(InstructionSet::OPStop, 0b000000),
];
let dependencies: Vec<[u8; 2]> = vec![[1, 255], [255, 255]];
let floats: Vec<f32> = vec![0.2, 0.2, 0.05];
let vec2s: Vec<[f32; 2]> = vec![[0.; 2]];
let vec3s: Vec<[f32; 3]> = vec![[0., 0.2, 0.], [0., 0.2, 0.]];
let vec4s: Vec<[f32; 4]> = vec![[0.; 4]];
let mat2s: Vec<[f32; 4]> = vec![[0.; 4]];
let mat3s: Vec<[f32; 9]> = vec![[0.; 9]];
let mat4s: Vec<[f32; 16]> = vec![[0.; 16]];
let mats: Vec<[f32; 16]> = vec![[0.; 16]];
let mut lower = true;
let mut minor = 0;
let mut major = 0;
for part in parts {
data[major][minor] |= (part.code as u32) << (if lower { 0 } else { 16 });
lower = !lower;
if lower {
minor += 1;
if minor == 4 {
minor = 0;
major += 1;
if major == 29 {
panic!("CSGParts Too full!");
}
}
}
}
//println!("data: {:?}, {:?}", data[0][0], data[0][1]);
let desc = implicit_fs::ty::Description {
scene: 0,
floats: 0,
vec2s: 0,
vec3s: 0,
vec4s: 0,
mat2s: 0,
mat3s: 0,
mat4s: 0,
dependencies: 0,
};
let descvec = vec![desc];
fn new_desc(
input: &[implicit_fs::ty::Description],
) -> &implicit_fs::ty::SceneDescription {
unsafe { ::std::mem::transmute(input) }
}
let csg_object = uniform_buffer
.allocate_slice((descvec.len() as u64).max(1))
.unwrap();
csg_object.write().unwrap().copy_from_slice(&descvec[..]);
let csg_opcodes = uniform_buffer
.allocate_slice((data.len() as u64).max(1))
.unwrap();
csg_opcodes.write().unwrap().copy_from_slice(&data[..]);
let csg_floats = uniform_buffer
.allocate_slice((floats.len() as u64).max(1))
.unwrap();
csg_floats.write().unwrap().copy_from_slice(&floats[..]);
let csg_vec2s = uniform_buffer
.allocate_slice((vec2s.len() as u64).max(1))
.unwrap();
csg_vec2s.write().unwrap().copy_from_slice(&vec2s[..]);
let csg_vec3s = uniform_buffer
.allocate_slice((vec3s.len() as u64).max(1))
.unwrap();
csg_vec3s.write().unwrap().copy_from_slice(&vec3s[..]);
let csg_vec4s = uniform_buffer
.allocate_slice((vec4s.len() as u64).max(1))
.unwrap();
csg_vec4s.write().unwrap().copy_from_slice(&vec4s[..]);
let csg_mat2s = uniform_buffer
.allocate_slice((mat2s.len() as u64).max(1))
.unwrap();
csg_mat2s.write().unwrap().copy_from_slice(&mat2s[..]);
let csg_mat3s = uniform_buffer
.allocate_slice((mat3s.len() as u64).max(1))
.unwrap();
csg_mat3s.write().unwrap().copy_from_slice(&mat3s[..]);
let csg_mat4s = uniform_buffer
.allocate_slice((mat4s.len() as u64).max(1))
.unwrap();
csg_mat4s.write().unwrap().copy_from_slice(&mat4s[..]);
let csg_mats = uniform_buffer
.allocate_slice((mats.len() as u64).max(1))
.unwrap();
csg_mats.write().unwrap().copy_from_slice(&mats[..]);
let csg_depends = uniform_buffer
.allocate_slice((dependencies.len() as u64).max(1))
.unwrap();
csg_depends
.write()
.unwrap()
.copy_from_slice(&dependencies[..]);
let mesh_layout = mesh_pipeline.layout().set_layouts().get(0).unwrap();
let mesh_set = PersistentDescriptorSet::new(
&descriptor_set_allocator,
@@ -789,18 +714,18 @@ fn main() {
[
WriteDescriptorSet::buffer(0, uniform_buffer_subbuffer.clone()),
WriteDescriptorSet::buffer(1, cam_set.clone()),
WriteDescriptorSet::buffer(2, csg_object.clone()),
WriteDescriptorSet::buffer(3, csg_opcodes.clone()),
WriteDescriptorSet::buffer(4, csg_floats.clone()),
WriteDescriptorSet::buffer(5, csg_vec2s.clone()),
WriteDescriptorSet::buffer(6, csg_vec3s.clone()),
WriteDescriptorSet::buffer(7, csg_vec4s.clone()),
//WriteDescriptorSet::buffer(8, csg_mat2s.clone()),
//WriteDescriptorSet::buffer(9, csg_mat3s.clone()),
//WriteDescriptorSet::buffer(10, csg_mat4s.clone()),
//WriteDescriptorSet::buffer(11, csg_mats.clone()),
WriteDescriptorSet::buffer(12, csg_depends.clone()),
WriteDescriptorSet::buffer(20, fragment_masks_buffer.clone()),
WriteDescriptorSet::buffer(2, subbuffers.desc.clone()),
WriteDescriptorSet::buffer(3, subbuffers.scene.clone()),
WriteDescriptorSet::buffer(4, subbuffers.floats.clone()),
WriteDescriptorSet::buffer(5, subbuffers.vec2s.clone()),
WriteDescriptorSet::buffer(6, subbuffers.vec3s.clone()),
WriteDescriptorSet::buffer(7, subbuffers.vec4s.clone()),
//WriteDescriptorSet::buffer(8, subbuffers.mat2s.clone()),
//WriteDescriptorSet::buffer(9, subbuffers.mat3s.clone()),
//WriteDescriptorSet::buffer(10, subbuffers.mat4s.clone()),
//WriteDescriptorSet::buffer(11, subbuffers.mats.clone()),
WriteDescriptorSet::buffer(12, subbuffers.deps.clone()),
WriteDescriptorSet::buffer(20, subbuffers.masks.clone()),
],
)
.unwrap();
@@ -944,6 +869,7 @@ fn window_size_dependent_setup<Mms>(
mesh_vs: &ShaderModule,
mesh_fs: &ShaderModule,
implicit_ms: &ShaderModule,
implicit_ts: &ShaderModule,
implicit_fs: &ShaderModule,
images: &[Arc<SwapchainImage>],
render_pass: Arc<RenderPass>,
@@ -1049,6 +975,7 @@ where
},
]))
.fragment_shader(implicit_fs.entry_point("main").unwrap(), specs)
.task_shader(implicit_ts.entry_point("main").unwrap(), ())
.mesh_shader(implicit_ms.entry_point("main").unwrap(), ())
.depth_stencil_state(DepthStencilState::simple_depth_test())
.rasterization_state(RasterizationState {
@@ -1072,20 +999,365 @@ where
([mesh_pipeline, implicit_pipeline], framebuffers)
}
fn object_size_dependent_setup(
struct Subbuffers {
masks: Subbuffer<[[u8; 29]]>,
floats: Subbuffer<[f32]>,
vec2s: Subbuffer<[[f32; 2]]>,
vec3s: Subbuffer<[[f32; 3]]>,
vec4s: Subbuffer<[[f32; 4]]>,
mat2s: Subbuffer<[[[f32; 2]; 2]]>,
mat3s: Subbuffer<[[[f32; 3]; 3]]>,
mat4s: Subbuffer<[[[f32; 4]; 4]]>,
mats: Subbuffer<[[[f32; 4]; 4]]>,
scene: Subbuffer<[[u32; 4]]>,
deps: Subbuffer<[[u8; 2]]>,
desc: Subbuffer<[[u32; 10]]>,
}
impl PartialEq<InputTypes> for Inputs {
fn eq(&self, other: &InputTypes) -> bool {
match self {
&Inputs::Variable => true,
&Inputs::Float(_) => *other == InputTypes::Float,
&Inputs::Vec2(_) => *other == InputTypes::Vec2,
&Inputs::Vec3(_) => *other == InputTypes::Vec3,
&Inputs::Vec4(_) => *other == InputTypes::Vec4,
&Inputs::Mat2(_) => *other == InputTypes::Mat2,
&Inputs::Mat3(_) => *other == InputTypes::Mat3,
&Inputs::Mat4(_) => *other == InputTypes::Mat4,
}
}
}
fn gpu_buffer<T>(
input: Vec<T>,
allocator: &StandardMemoryAllocator,
sub_allocator: &SubbufferAllocator,
command_allocator: &StandardCommandBufferAllocator,
transfer_queue: Arc<Queue>,
) -> Subbuffer<[T]>
where
T: bytemuck::Pod + Send + Sync
{
let buffer = Buffer::new_slice(
allocator,
BufferAllocateInfo {
buffer_usage: BufferUsage::STORAGE_BUFFER | BufferUsage::TRANSFER_DST,
memory_usage: MemoryUsage::GpuOnly,
allocate_preference: MemoryAllocatePreference::AlwaysAllocate,
..Default::default()
},
(input.len()) as u64,
)
.unwrap();
let staging = sub_allocator.allocate_slice(input.len() as u64).unwrap();
staging.write().unwrap().copy_from_slice(&input[..]);
let mut builder = AutoCommandBufferBuilder::primary(
command_allocator,
transfer_queue.queue_family_index(),
CommandBufferUsage::OneTimeSubmit,
)
.unwrap();
builder.copy_buffer(CopyBufferInfo::buffers(
staging,
buffer.clone())
).unwrap();
let commands = builder.build().unwrap();
commands.execute(transfer_queue.clone())
.unwrap()
.then_signal_fence_and_flush()
.unwrap()
.wait(None)
.unwrap();
buffer
}
fn object_size_dependent_setup(
allocator: Arc<StandardMemoryAllocator>,
state: &GState,
) -> Subbuffer<[[u8; 29]]> {
command_allocator: &StandardCommandBufferAllocator,
queue: Arc<Queue>,
) -> Subbuffers {
let mut floats: Vec<f32> = vec![Default::default()];
let mut vec2s: Vec<[f32; 2]> = vec![Default::default()];
let mut vec3s: Vec<[f32; 3]> = vec![Default::default()];
let mut vec4s: Vec<[f32; 4]> = vec![Default::default()];
let mut mat2s: Vec<[[f32; 2]; 2]> = vec![Default::default()];
let mut mat3s: Vec<[[f32; 3]; 3]> = vec![Default::default()];
let mut mat4s: Vec<[[f32; 4]; 4]> = vec![Default::default()];
let mut mats: Vec<[[f32; 4]; 4]> = vec![Default::default()];
let mut scene: Vec<[u32; 4]> = vec![Default::default()];
let mut deps: Vec<[u8; 2]> = vec![Default::default()];
let mut desc: Vec<[u32; 10]> = vec![Default::default()];
'nextcsg: for csg in &state.csg {
let mut data: Vec<[u32; 4]> = vec![];
let to_push = [
scene.len() as u32,
floats.len() as u32,
vec2s.len() as u32,
vec3s.len() as u32,
vec4s.len() as u32,
mat2s.len() as u32,
mat3s.len() as u32,
mat4s.len() as u32,
mats.len() as u32,
deps.len() as u32,
];
let parts = vec![
//CSGPart::opcode(InstructionSet::OPDupVec3, vec![Inputs::Variable]),
//CSGPart::opcode(InstructionSet::OPSubVec3Vec3, vec![Inputs::Variable, Inputs::Vec3([0., 0.2, 0.].into())]),
CSGPart::opcode(
InstructionSet::OPSDFSphere,
vec![Inputs::Float(1.0), Inputs::Variable],
),
//CSGPart::opcode(InstructionSet::OPAddVec3Vec3, 0b010000),
//CSGPart::opcode(InstructionSet::OPSDFSphere, 0b100000),
//CSGPart::opcode(InstructionSet::OPSmoothMinFloat, 0b000000),
CSGPart::opcode(InstructionSet::OPStop, vec![Inputs::Variable]),
];
let mut dependencies: Vec<[u8; 2]> = vec![];
for _ in 0..parts.len() {
dependencies.push([u8::MAX, u8::MAX]);
}
let mut runtime_floats: Vec<usize> = vec![];
let mut runtime_vec2s: Vec<usize> = vec![];
let mut runtime_vec3s: Vec<usize> = vec![usize::MAX];
let mut runtime_vec4s: Vec<usize> = vec![];
let mut runtime_mat2s: Vec<usize> = vec![];
let mut runtime_mat3s: Vec<usize> = vec![];
let mut runtime_mat4s: Vec<usize> = vec![];
for (index, part) in parts.iter().enumerate() {
let inputs = part.opcode.input();
for (expected, actual) in inputs.iter().zip(part.constants.iter()) {
if actual != expected {
eprintln!(
"csg {} is invalid ({:?} != {:?})",
csg.name, actual, expected
);
continue 'nextcsg;
}
if actual == &Inputs::Variable {
match expected {
&InputTypes::Float => match runtime_floats.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on floats", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Vec2 => match runtime_vec2s.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on vec2s", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Vec3 => match runtime_vec3s.pop() {
Some(u) => {
if u != usize::MAX {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
}
None => {
eprintln!("csg {} underflowed on vec3s", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Vec4 => match runtime_vec4s.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on vec4s", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Mat2 => match runtime_mat2s.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on mat2s", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Mat3 => match runtime_mat3s.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on mat3s", csg.name);
continue 'nextcsg;
}
},
&InputTypes::Mat4 => match runtime_mat4s.pop() {
Some(u) => {
if dependencies[u][0] != u8::MAX {
dependencies[u][1] = index as u8
} else {
dependencies[u][0] = index as u8
}
}
None => {
eprintln!("csg {} underflowed on mat4s", csg.name);
continue 'nextcsg;
}
},
}
} else {
match actual {
&Inputs::Float(f) => floats.push(f),
&Inputs::Vec2(f) => vec2s.push(f.into()),
&Inputs::Vec3(f) => vec3s.push(f.into()),
&Inputs::Vec4(f) => vec4s.push(f.into()),
&Inputs::Mat2(f) => mat2s.push(f.into()),
&Inputs::Mat3(f) => mat3s.push(f.into()),
&Inputs::Mat4(f) => mat4s.push(f.into()),
&Inputs::Variable => unreachable!(),
}
}
}
let outputs = part.opcode.output();
for output in outputs {
match output {
InputTypes::Float => runtime_floats.push(index),
InputTypes::Vec2 => runtime_vec2s.push(index),
InputTypes::Vec3 => runtime_vec3s.push(index),
InputTypes::Vec4 => runtime_vec4s.push(index),
InputTypes::Mat2 => runtime_mat2s.push(index),
InputTypes::Mat3 => runtime_mat3s.push(index),
InputTypes::Mat4 => runtime_mat4s.push(index),
}
}
}
let mut lower = true;
let mut minor = 0;
let mut major = 0;
for part in parts {
if major == data.len() {
data.push([0; 4]);
}
data[major][minor] |= (part.code as u32) << (if lower { 0 } else { 16 });
lower = !lower;
if lower {
minor += 1;
if minor == 4 {
minor = 0;
major += 1;
if major == 29 {
panic!("CSGParts Too full!");
}
}
}
}
desc.push(to_push);
scene.append(&mut data);
deps.append(&mut dependencies);
}
println!("floats: {:?}", floats);
println!("vec2s: {:?}", vec2s);
println!("vec3s: {:?}", vec3s);
println!("vec4s: {:?}", vec4s);
println!("mat2s: {:?}", mat2s);
println!("mat3s: {:?}", mat3s);
println!("mat4s: {:?}", mat4s);
println!("mats: {:?}", mats);
println!("scene: {:?}", scene);
println!("deps: {:?}", deps);
println!("desc: {:?}", desc);
let fragment_masks_buffer = Buffer::new_slice(
allocator.clone(),
&allocator,
BufferAllocateInfo {
buffer_usage: BufferUsage::STORAGE_BUFFER,
memory_usage: MemoryUsage::GpuOnly,
allocate_preference: MemoryAllocatePreference::AlwaysAllocate,
..Default::default()
},
(state.csg.len() * (4 * 4 * 4) * (3 * 3 * 3) * 29) as u64,
((desc.len()-1) * (4 * 4 * 4) * (4 * 4 * 2) * 29) as u64,
)
.unwrap();
fragment_masks_buffer
let staging = SubbufferAllocator::new(
allocator.clone(),
SubbufferAllocatorCreateInfo {
buffer_usage: BufferUsage::TRANSFER_SRC,
memory_usage: MemoryUsage::Upload,
..Default::default()
},
);
let csg_scene = gpu_buffer(scene, &allocator, &staging, command_allocator, queue.clone());
let csg_desc = gpu_buffer(desc, &allocator, &staging, command_allocator, queue.clone());
let csg_floats = gpu_buffer(floats, &allocator, &staging, command_allocator, queue.clone());
let csg_vec2s = gpu_buffer(vec2s, &allocator, &staging, command_allocator, queue.clone());
let csg_vec3s = gpu_buffer(vec3s, &allocator, &staging, command_allocator, queue.clone());
let csg_vec4s = gpu_buffer(vec4s, &allocator, &staging, command_allocator, queue.clone());
let csg_mat2s = gpu_buffer(mat2s, &allocator, &staging, command_allocator, queue.clone());
let csg_mat3s = gpu_buffer(mat3s, &allocator, &staging, command_allocator, queue.clone());
let csg_mat4s = gpu_buffer(mat4s, &allocator, &staging, command_allocator, queue.clone());
let csg_mats = gpu_buffer(mats, &allocator, &staging, command_allocator, queue.clone());
let csg_deps = gpu_buffer(deps, &allocator, &staging, command_allocator, queue.clone());
Subbuffers {
masks: fragment_masks_buffer,
floats: csg_floats,
vec2s: csg_vec2s,
vec3s: csg_vec3s,
vec4s: csg_vec4s,
mat2s: csg_mat2s,
mat3s: csg_mat3s,
mat4s: csg_mat4s,
mats: csg_mats,
scene: csg_scene,
deps: csg_deps,
desc: csg_desc,
}
}
+49 -22
View File
@@ -1,11 +1,15 @@
use std::{
collections::HashMap,
default,
io::{Cursor, Read},
mem,
};
use bytemuck::{Pod, Zeroable};
use cgmath::{Deg, EuclideanSpace, Euler, Matrix3, Point3, SquareMatrix, Vector3};
use cgmath::{
Deg, EuclideanSpace, Euler, Matrix2, Matrix3, Matrix4, Point3, SquareMatrix, Vector2, Vector3,
Vector4,
};
use obj::{LoadConfig, ObjData, ObjError};
use serde::{Deserialize, Serialize};
use vulkano::{
@@ -45,30 +49,63 @@ pub struct Mesh {
#[derive(Debug)]
pub struct CSG {
pub name: String,
pub parts: Subbuffer<[CSGPart]>,
pub parts: Vec<CSGPart>,
pub pos: Point3<f32>,
pub rot: Euler<Deg<f32>>,
pub scale: Vector3<f32>,
}
#[derive(Clone, Debug, Default, PartialEq)]
pub enum Inputs {
#[default]
Variable,
Float(f32),
Vec2(Vector2<f32>),
Vec3(Vector3<f32>),
Vec4(Vector4<f32>),
Mat2(Matrix2<f32>),
Mat3(Matrix3<f32>),
Mat4(Matrix4<f32>),
}
#[repr(C)]
#[derive(Clone, Copy, Debug, Default, Zeroable, Pod, Vertex)]
#[derive(Clone, Debug)]
pub struct CSGPart {
#[format(R16_SFLOAT)]
pub code: u16,
pub opcode: InstructionSet,
pub constants: Vec<Inputs>,
pub material: Option<Matrix4<f32>>,
}
impl CSGPart {
pub fn opcode(opcode: InstructionSet, flags: u16) -> CSGPart {
pub fn opcode(opcode: InstructionSet, inputs: Vec<Inputs>) -> CSGPart {
CSGPart {
code: (opcode as u16 & 1023) | flags << 10,
code: (opcode as u16 & 1023)
| inputs
.iter()
.enumerate()
.map(|(i, n)| {
if n != &Inputs::Variable {
1 << (15 - i)
} else {
0
}
})
.fold(0, |a, i| a | i),
opcode,
constants: inputs,
material: None,
}
}
pub fn literal(literal: f16) -> CSGPart {
CSGPart {
code: literal.to_bits(),
}
pub fn opcode_with_material(
opcode: InstructionSet,
inputs: Vec<Inputs>,
material: Matrix4<f32>,
) -> CSGPart {
let mut c = CSGPart::opcode(opcode, inputs);
c.material = Some(material);
c
}
}
@@ -326,7 +363,7 @@ pub fn load_csg(
.iter()
.map(|inpart| {
let ty = inpart.get("type").ok_or("no type!")?.as_str();
let mut csgpart = CSGPart::literal(0u8.into());
let mut csgpart = CSGPart::opcode(InstructionSet::OPNop, vec![]);
match ty {
"sphere" => {
let blend = get_f32(&inpart, "blend")?;
@@ -342,18 +379,8 @@ pub fn load_csg(
})
.collect::<Result<Vec<CSGPart>, String>>()?;
let parts_buffer = Buffer::from_iter(
memory_allocator,
BufferAllocateInfo {
buffer_usage: BufferUsage::VERTEX_BUFFER,
..Default::default()
},
parts,
)
.unwrap();
Ok(CSG {
parts: parts_buffer,
parts,
pos: Point3 {
x: 0.,
y: 0.,