Dynamitos/Seele
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Very strange mask bug

Browse Source
This commit is contained in:
Dynamitos
2024-10-30 10:28:10 +01:00
parent 289b759498
commit 2811757476
11 changed files with 276 additions and 950 deletions
Download Patch File Download Diff File Expand all files Collapse all files
res/shaders/terrain/CBT.slang
+71 -31
View File
@@ -60,7 +60,33 @@ static const uint32_t OCBT_depth_offset[18] = { 0, // Level 0
0, // Level 18
};
static const uint64_t OCBT_bit_mask[18] = { 0xffffffff, // Root 17
static const uint OCBT_bit_mask_high[18] = {
0x0, // Root 17
0x0, // Level 16
0x0, // level 15
0x0, // level 14
0x0, // level 13
0x0, // level 12
0x0, // level 11
0x0, // level 10
0x0, // level 9
0x0, // level 8
0x0, // level 8
0xffffffff, // level 7
0x0, // Level 6
0x0, // level 5
0x0, // level 4
0x0, // level 3
0x0, // level 2
0x0, // level 1
};
static const uint OCBT_bit_mask_low[18] = {
0xffffffff, // Root 17
0xffffffff, // Level 16
0xffffffff, // level 15
0xffffffff, // level 14
@@ -73,7 +99,7 @@ static const uint64_t OCBT_bit_mask[18] = { 0xffffffff, // Root 17
0xffff, // level 8
0xff, // level 8
0xffffffffffffffff, // level 7
0xffffffff, // level 7
0xffffffff, // Level 6
0xffff, // level 5
0xff, // level 4
@@ -172,14 +198,16 @@ uint get_heap_element(uint id, inout HeapDebug debug)
{
uint32_t slot = first_bit / 32;
uint32_t local_id = first_bit % 32;
uint32_t target_bits = (gs_cbtTree[slot] >> local_id) & uint32_t(OCBT_bit_mask[depth]);
return (gs_cbtTree[slot] >> local_id) & uint32_t(OCBT_bit_mask[depth]);
return (pParams.cbtBuffer[slot] >> local_id) & OCBT_bit_mask_low[depth];
}
else
{
uint32_t slot = first_bit / 64;
uint32_t local_id = first_bit % 64;
uint64_t target_bits = (pParams.bitFieldBuffer[slot] >> local_id) & OCBT_bit_mask[depth];
uint64_t mask = OCBT_bit_mask_high[depth];
mask = (mask << 32) | OCBT_bit_mask_low[depth];
uint64_t target_bits = (pParams.bitFieldBuffer[slot] >> local_id) & mask;
debug.target_bits = target_bits;
return countbits(uint(target_bits >> 32)) + countbits(uint(target_bits));
}
}
@@ -198,8 +226,8 @@ void set_heap_element(uint id, uint value)
uint local_id = first_bit % 32;
// Extract the relevant bits
gs_cbtTree[slot] &= ~(uint32_t(OCBT_bit_mask[depth]) << local_id);
gs_cbtTree[slot] |= ((uint32_t(OCBT_bit_mask[depth]) & value) << local_id);
gs_cbtTree[slot] &= ~(uint32_t(OCBT_bit_mask_low[depth]) << local_id);
gs_cbtTree[slot] |= ((uint32_t(OCBT_bit_mask_low[depth]) & value) << local_id);
}
// Should not be called if depth > TREE_LAST_LEVEL
@@ -216,8 +244,8 @@ void set_heap_element_atomic(uint id, uint value)
uint local_id = first_bit % 32;
// Extract the relevant bits
InterlockedAnd(gs_cbtTree[slot], ~(uint32_t(OCBT_bit_mask[depth]) << local_id));
InterlockedOr(gs_cbtTree[slot], ((uint32_t(OCBT_bit_mask[depth]) & value) << local_id));
InterlockedAnd(gs_cbtTree[slot], ~(uint32_t(OCBT_bit_mask_low[depth]) << local_id));
InterlockedOr(gs_cbtTree[slot], ((uint32_t(OCBT_bit_mask_low[depth]) & value) << local_id));
}
// Function that returns the number of active bits
@@ -233,37 +261,49 @@ uint bit_count(uint depth, uint element)
}
// decodes the position of the i-th zero in the bitfield
uint decode_bit_complement(uint handle, uint dispatchID)
uint decode_bit_complement(uint handle, inout DebugStruct debug)
{
uint temp = handle;
DebugStruct debug;
uint x = 0;
uint bitID = 1u;
uint c = OCBT_NUM_ELEMENTS / 2u;
while (bitID < OCBT_NUM_ELEMENTS) {
uint heapValue = c - get_heap_element(2u * bitID, debug.heapValues[x++]);
uint b = handle < heapValue ? 0u : 1u;
bitID = 2u * bitID + b;
handle -= heapValue * b;
uint b2 = 2u * bitID;
uint h = get_heap_element(b2, debug.heapValues[x]);
uint heapValue = c - h;
uint b;
if(handle < heapValue)
{
b = 0u;
}
else
{
b = 1u;
}
debug.heapValues[x].handle = handle;
debug.heapValues[x].heapValue = heapValue;
debug.heapValues[x].c = c;
debug.heapValues[x].h = h;
x++;
bitID = b2 + b;
uint t = heapValue * b;
handle = handle - t;
c /= 2u;
}
uint result = (bitID ^ OCBT_NUM_ELEMENTS);
handle = temp;
for(uint i = 0; i < OCBT_NUM_ELEMENTS; ++i)
{
if(get_bit(i) == 0)
{
if(handle == 0)
{
pParams.debugBuffer[dispatchID] = debug;
return i;
}
handle--;
}
}
return (bitID ^ OCBT_NUM_ELEMENTS);
//handle = temp;
//for(uint i = 0; i < OCBT_NUM_ELEMENTS; ++i)
//{
// if(get_bit(i) == 0)
// {
// if(handle == 0)
// {
// return i;
// }
// handle--;
// }
//}
}
void reduce(uint groupIndex)
res/shaders/terrain/CBTCompute.slang
+90 -88
View File
@@ -10,46 +10,48 @@ void Reset()
ResetBuffers();
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void Classify(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// This thread doesn't have any work to do, we're done
//if (dispatchID >= pParams.indirectDrawBuffer[9])
// return;
//for(uint i = 0; i < pParams.indirectDrawBuffer[9]; ++i)
//{
// This thread doesn't have any work to do, we're done
if (dispatchID >= pParams.indirectDrawBuffer[9])
if(dispatchID > 0)
return;
for(dispatchID = 0; dispatchID < pParams.indirectDrawBuffer[9]; ++dispatchID)
{
// Operate the indirection
uint currentID = pParams.indexedBisectorBuffer[dispatchID];
// Operate the indirection
uint currentID = pParams.indexedBisectorBuffer[dispatchID];
// Read the current geometry data
BisectorGeometry bis;
bis.p[0] = pParams.currentVertexBuffer[3 * currentID].xyz;
bis.p[1] = pParams.currentVertexBuffer[3 * currentID + 1].xyz;
bis.p[2] = pParams.currentVertexBuffer[3 * currentID + 2].xyz;
bis.p[3] = pParams.currentVertexBuffer[3 * pParams.geometry.totalNumElements + currentID].xyz;
// Read the current geometry data
BisectorGeometry bis;
bis.p[0] = pParams.currentVertexBuffer[3 * currentID].xyz;
bis.p[1] = pParams.currentVertexBuffer[3 * currentID + 1].xyz;
bis.p[2] = pParams.currentVertexBuffer[3 * currentID + 2].xyz;
bis.p[3] = pParams.currentVertexBuffer[3 * pParams.geometry.totalNumElements + currentID].xyz;
// Classify the element
ClassifyElement(currentID, bis, pParams.geometry.totalNumElements, pParams.geometry.baseDepth);
// Classify the element
ClassifyElement(currentID, bis, pParams.geometry.totalNumElements, pParams.geometry.baseDepth);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void Split(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
//if (dispatchID >= pParams.classificationBuffer[SPLIT_COUNTER])
// return;
//for(dispatchID = 0; dispatchID < pParams.classificationBuffer[SPLIT_COUNTER]; ++dispatchID)
//{
if (dispatchID >= pParams.classificationBuffer[SPLIT_COUNTER])
if(dispatchID > 0)
return;
for(dispatchID = 0; dispatchID < pParams.classificationBuffer[SPLIT_COUNTER]; ++dispatchID)
{
// Grab the real elementID
uint currentID = pParams.classificationBuffer[CLASSIFY_COUNTER_OFFSET + dispatchID];
// Grab the real elementID
uint currentID = pParams.classificationBuffer[CLASSIFY_COUNTER_OFFSET + dispatchID];
// Split the element
SplitElement(currentID, pParams.geometry.baseDepth);
}
// Split the element
SplitElement(currentID, pParams.geometry.baseDepth);
}
[numthreads(1, 1, 1)]
@@ -60,102 +62,102 @@ void PrepareIndirect(uint currentID : SV_DispatchThreadID)
pParams.indirectDispatchBuffer[currentID * 3 + 2] = 1;
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void Allocate(uint groupIndex : SV_GroupIndex, uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
//
//for(uint i = 0; i < 64; ++i)
//{
if(dispatchID > 0)
return;
for(uint i = 0; i < 64; ++i)
{
// Load the CBT to the LDS
load_buffer_to_shared_memory(groupIndex);
load_buffer_to_shared_memory(groupIndex);
}
// If this element doesn't need to be processed, we're done
if (dispatchID >= pParams.allocateBuffer[0])
return;
//for(uint i = 0; i < pParams.allocateBuffer[0]; ++i)
//{
for(dispatchID = 0; dispatchID < pParams.allocateBuffer[0]; ++dispatchID)
{
// Allocate the required bits
AllocateElement(pParams.allocateBuffer[1 + dispatchID], dispatchID);
//}
AllocateElement(pParams.allocateBuffer[1 + dispatchID], dispatchID);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void Bisect(uint groupIndex : SV_GroupIndex, uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
// If this element doesn't need to be processed, we're done
if (dispatchID >= pParams.allocateBuffer[0])
//if (dispatchID >= pParams.allocateBuffer[0])
// return;
if(dispatchID > 0)
return;
//for(uint i = 0; i < pParams.allocateBuffer[0]; ++i)
//{
for(dispatchID = 0; dispatchID < pParams.allocateBuffer[0]; ++dispatchID)
{
// Operation the bisection of this element
BisectElement(pParams.allocateBuffer[1 + dispatchID]);
//}
BisectElement(pParams.allocateBuffer[1 + dispatchID]);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void PropagateBisect(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
// If this element doesn't need to be processed, we're done
if (dispatchID >= pParams.propagateBuffer[0])
//if (dispatchID >= pParams.propagateBuffer[0])
// return;
if(dispatchID > 0)
return;
//for(uint i = 0; i < pParams.propagateBuffer[0]; ++i)
//{
PropagateBisectElement(pParams.propagateBuffer[2 + dispatchID]);
//}
for(dispatchID = 0; dispatchID < pParams.propagateBuffer[0]; ++dispatchID)
{
PropagateBisectElement(pParams.propagateBuffer[2 + dispatchID]);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void PrepareSimplify(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
// If this element doesn't need to be processed, we're done
if (dispatchID >= pParams.classificationBuffer[SIMPLIFY_COUNTER])
return;
//for(uint i = 0; i < pParams.classificationBuffer[SIMPLIFY_COUNTER]; ++i)
//{
// Grab the real elementID
uint currentID = pParams.classificationBuffer[CLASSIFY_COUNTER_OFFSET + pParams.geometry.totalNumElements + dispatchID];
if(dispatchID > 0)
return;
for(dispatchID = 0; dispatchID < pParams.classificationBuffer[SIMPLIFY_COUNTER]; ++dispatchID)
{
// Grab the real elementID
uint currentID = pParams.classificationBuffer[CLASSIFY_COUNTER_OFFSET + pParams.geometry.totalNumElements + dispatchID];
// Simplify an element
PrepareSimplifyElement(currentID);
//}
// Simplify an element
PrepareSimplifyElement(currentID);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void Simplify(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
// This thread doesn't have any work to do, we're done
if (dispatchID >= pParams.simplifyBuffer[0])
return;
//for(uint i = 0; i < pParams.simplifyBuffer[0]; ++i)
//{
// Simplify an element
SimplifyElement(pParams.simplifyBuffer[1 + dispatchID]);
//}
if(dispatchID > 0)
return;
for(dispatchID = 0; dispatchID < pParams.simplifyBuffer[0]; ++dispatchID)
{
// Simplify an element
SimplifyElement(pParams.simplifyBuffer[1 + dispatchID]);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void PropagateSimplify(uint dispatchID : SV_DispatchThreadID)
{
//if(dispatchID > 0)
// return;
// If this element doesn't need to be processed, we're done
if (dispatchID >= pParams.propagateBuffer[1])
return;
//for(uint i = 0; i < pParams.propagateBuffer[1]; ++i)
//{
PropagateElementSimplify(pParams.propagateBuffer[2 + dispatchID]);
//}
if(dispatchID > 0)
return;
for(dispatchID = 0; dispatchID < pParams.propagateBuffer[1]; ++dispatchID)
{
PropagateElementSimplify(pParams.propagateBuffer[2 + dispatchID]);
}
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
@@ -178,20 +180,20 @@ void ReduceSecondPass(uint groupIndex : SV_GroupIndex)
reduce_second_pass(groupIndex);
}
[numthreads(WORKGROUP_SIZE, 1, 1)]
[numthreads(1, 1, 1)]
void BisectorIndexation(uint currentID : SV_DispatchThreadID)
{
//if(currentID > 0)
// return;
// This thread doesn't have any work to do, we're done
if (currentID >= pParams.geometry.totalNumElements)
//if (currentID >= pParams.geometry.totalNumElements)
// return;
if(currentID > 0)
return;
//for(uint i = 0; i < pParams.geometry.totalNumElements; ++i)
//{
// Indexate this bisector
BisectorElementIndexation(currentID);
//}
for(currentID = 0; currentID < pParams.geometry.totalNumElements; ++currentID)
{
// Indexate this bisector
BisectorElementIndexation(currentID);
}
}
[numthreads(1, 1, 1)]
res/shaders/terrain/CompileTest.glsl
+9 -439
View File
@@ -1,455 +1,25 @@
#version 450
#extension GL_EXT_shader_explicit_arithmetic_types_int64 : require
#extension GL_KHR_memory_scope_semantics : require
layout(row_major) uniform;
layout(row_major) buffer;
#line 178 0
layout(std430, binding = 2) buffer StructuredBuffer_uint_t_0 {
uint _data[];
} pParams_classificationBuffer_0;
#line 178
layout(std430, binding = 9) buffer StructuredBuffer_uint_t_1 {
uint _data[];
} pParams_cbtBuffer_0;
#line 63
layout(std430, binding = 10) buffer StructuredBuffer_uint64_t_0 {
#line 35 0
layout(std430, binding = 0) buffer StructuredBuffer_uint64_t_0 {
uint64_t _data[];
} pParams_bitFieldBuffer_0;
#line 119
const uint64_t OCBT_bit_mask_0[18] = { 18446744073709551615UL, 18446744073709551615UL, 18446744073709551615UL, 18446744073709551615UL, 18446744073709551615UL, 18446744073709551615UL, 18446744073709551615UL, 65535UL, 65535UL, 65535UL, 255UL, 18446744073709551615UL, 18446744073709551615UL, 65535UL, 255UL, 15UL, 3UL, 1UL };
#line 8
const uint OCBT_bit_mask_0[18] = { 4294967295U, 4294967295U, 4294967295U, 4294967295U, 4294967295U, 4294967295U, 4294967295U, 65535U, 65535U, 65535U, 255U, 4294967295U, 4294967295U, 65535U, 255U, 15U, 3U, 1U };
#line 141
const uint OCBT_bit_count_0[18] = { 32U, 32U, 32U, 32U, 32U, 32U, 32U, 16U, 16U, 16U, 8U, 64U, 32U, 16U, 8U, 4U, 2U, 1U };
#line 97
const uint OCBT_depth_offset_0[18] = { 0U, 32U, 96U, 224U, 480U, 992U, 2016U, 4064U, 6112U, 10208U, 18400U, 0U, 0U, 0U, 0U, 0U, 0U, 0U };
#line 163
shared uint gs_cbtTree_0[831];
#line 171
void load_buffer_to_shared_memory_0(uint groupIndex_0)
{
#line 171
uint e_0 = 0U;
for(;;)
{
#line 174
if(int(e_0) < 13)
{
}
else
{
#line 174
break;
}
uint target_element_0 = uint(13 * int(groupIndex_0) + int(e_0));
if(int(target_element_0) < 831)
{
#line 178
gs_cbtTree_0[target_element_0] = pParams_cbtBuffer_0._data[uint(target_element_0)];
#line 177
}
#line 174
e_0 = e_0 + 1U;
#line 174
}
#line 180
controlBarrier(gl_ScopeWorkgroup, gl_ScopeWorkgroup, gl_StorageSemanticsShared, gl_SemanticsAcquireRelease);
return;
}
#line 183
uint get_heap_element_0(uint id_0)
{
uint real_heap_id_0 = id_0 - 1U;
uint depth_0 = uint(log2(float(real_heap_id_0 + 1U)));
uint first_bit_0 = OCBT_depth_offset_0[depth_0] + OCBT_bit_count_0[depth_0] * (real_heap_id_0 - ((1U << depth_0) - 1U));
if(depth_0 < 11U)
{
return gs_cbtTree_0[first_bit_0 / 32U] >> first_bit_0 % 32U & uint(OCBT_bit_mask_0[depth_0]);
}
else
{
uint64_t target_bits_0 = pParams_bitFieldBuffer_0._data[uint(first_bit_0 / 64U)] >> first_bit_0 % 64U & OCBT_bit_mask_0[depth_0];
return bitCount(uint(target_bits_0 >> 32)) + bitCount(uint(target_bits_0));
}
#line 205
}
uint decode_bit_0(uint handle_0)
{
#line 210
uint b_0;
#line 210
uint heapElementID_0 = 1U;
#line 210
uint currentDepth_0 = 0U;
#line 210
uint bitCount_0 = handle_0;
#line 227
for(;;)
{
#line 227
if(currentDepth_0 < 11U)
{
}
else
{
#line 227
break;
}
uint heapValue_0 = get_heap_element_0(2U * heapElementID_0);
if(bitCount_0 < heapValue_0)
{
#line 233
b_0 = 0U;
#line 233
}
else
{
#line 233
b_0 = 1U;
#line 233
}
uint _S1 = 2U * heapElementID_0 + b_0;
uint _S2 = bitCount_0 - heapValue_0 * b_0;
#line 227
uint currentDepth_1 = currentDepth_0 + 1U;
#line 227
heapElementID_0 = _S1;
#line 227
currentDepth_0 = currentDepth_1;
#line 227
bitCount_0 = _S2;
#line 227
}
#line 243
uint _S3 = currentDepth_0 + 1U;
uint64_t _S4 = pParams_bitFieldBuffer_0._data[uint(int(heapElementID_0) - 2048)];
#line 210
uint _S5 = bitCount_0;
#line 210
uint64_t mask_0 = 18446744073709551615UL;
#line 210
currentDepth_0 = _S3;
#line 210
bitCount_0 = 32U;
#line 210
uint _S6 = _S5;
#line 249
for(;;)
{
#line 249
if(currentDepth_0 < uint(log2(1.31072e+05)) + 1U)
{
}
else
{
#line 249
break;
}
#line 257
uint64_t target_bits_1 = _S4 >> bitCount_0 * (2U * heapElementID_0 - 1U - ((1U << currentDepth_0) - 1U)) % 64U & mask_0;
uint heapValue_1 = bitCount(uint(target_bits_1 >> 32)) + bitCount(uint(target_bits_1));
if(_S6 < heapValue_1)
{
#line 261
b_0 = 0U;
#line 261
}
else
{
#line 261
b_0 = 1U;
#line 261
}
uint _S7 = 2U * heapElementID_0 + b_0;
uint _S8 = _S6 - heapValue_1 * b_0;
uint bitCount_1 = bitCount_0 / 2U;
uint64_t _S9 = mask_0 >> bitCount_1;
#line 249
uint currentDepth_2 = currentDepth_0 + 1U;
#line 249
heapElementID_0 = _S7;
#line 249
mask_0 = _S9;
#line 249
currentDepth_0 = currentDepth_2;
#line 249
bitCount_0 = bitCount_1;
#line 249
_S6 = _S8;
#line 249
}
#line 273
return heapElementID_0 ^ 131072U;
}
uint decode_bit_complement_0(uint handle_1)
{
#line 278
uint b_1;
#line 278
uint heapElementID_1 = 1U;
#line 278
uint currentDepth_3 = 0U;
#line 278
uint c_0 = 65536U;
#line 278
uint bitCount_2 = handle_1;
#line 299
for(;;)
{
#line 299
if(currentDepth_3 < 11U)
{
}
else
{
#line 299
break;
}
uint heapValue_2 = c_0 - get_heap_element_0(2U * heapElementID_1);
if(bitCount_2 < heapValue_2)
{
#line 302
b_1 = 0U;
#line 302
}
else
{
#line 302
b_1 = 1U;
#line 302
}
uint _S10 = 2U * heapElementID_1 + b_1;
uint _S11 = bitCount_2 - heapValue_2 * b_1;
uint c_1 = c_0 / 2U;
#line 299
uint currentDepth_4 = currentDepth_3 + 1U;
#line 299
heapElementID_1 = _S10;
#line 299
currentDepth_3 = currentDepth_4;
#line 299
c_0 = c_1;
#line 299
bitCount_2 = _S11;
#line 299
}
#line 310
uint _S12 = currentDepth_3 + 1U;
uint64_t _S13 = pParams_bitFieldBuffer_0._data[uint(int(heapElementID_1) - 2048)];
#line 278
uint _S14 = bitCount_2;
#line 278
uint64_t mask_1 = 18446744073709551615UL;
#line 278
currentDepth_3 = _S12;
#line 278
bitCount_2 = 32U;
#line 278
uint _S15 = _S14;
#line 316
for(;;)
{
#line 316
if(currentDepth_3 < uint(log2(1.31072e+05)) + 1U)
{
}
else
{
#line 316
break;
}
#line 324
uint64_t target_bits_2 = _S13 >> bitCount_2 * (2U * heapElementID_1 - 1U - ((1U << currentDepth_3) - 1U)) % 64U & mask_1;
uint heapValue_3 = c_0 - bitCount(uint(target_bits_2 >> 32)) + bitCount(uint(target_bits_2));
if(_S15 < heapValue_3)
{
#line 327
b_1 = 0U;
#line 327
}
else
{
#line 327
b_1 = 1U;
#line 327
}
uint _S16 = 2U * heapElementID_1 + b_1;
uint _S17 = _S15 - heapValue_3 * b_1;
uint c_2 = c_0 / 2U;
uint bitCount_3 = bitCount_2 / 2U;
uint64_t _S18 = mask_1 >> bitCount_3;
#line 316
uint currentDepth_5 = currentDepth_3 + 1U;
#line 316
heapElementID_1 = _S16;
#line 316
mask_1 = _S18;
#line 316
currentDepth_3 = currentDepth_5;
#line 316
bitCount_2 = bitCount_3;
#line 316
c_0 = c_2;
#line 316
_S15 = _S17;
#line 316
}
#line 338
return heapElementID_1 ^ 131072U;
}
layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
#line 33
layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
void main()
{
#line 343
uint _S19 = gl_GlobalInvocationID.x;
#line 33
uint _S1 = gl_GlobalInvocationID.x;
load_buffer_to_shared_memory_0(gl_LocalInvocationIndex);
pParams_classificationBuffer_0._data[uint(_S19)] = decode_bit_0(_S19);
pParams_classificationBuffer_0._data[uint(_S19 * 2U)] = decode_bit_complement_0(_S19);
pParams_bitFieldBuffer_0._data[uint(_S1)] = uint64_t(OCBT_bit_mask_0[_S1]);
return;
}
res/shaders/terrain/CompileTest.slang
+7 -319
View File
@@ -1,122 +1,12 @@
// Pointer to an invalid neighbor or index
const static int INVALID_POINTER = 4294967295;
// Possible culling state
const static int BACK_FACE_CULLED = -3;
const static int FRUSTUM_CULLED = -2;
const static int TOO_SMALL = -1;
const static int UNCHANGED_ELEMENT = 0;
const static int BISECT_ELEMENT = 1;
const static int SIMPLIFY_ELEMENT = 2;
const static int MERGED_ELEMENT = 3;
// Bisector flags
const static int VISIBLE_BISECTOR = 0x1;
const static int MODIFIED_BISECTOR = 0x2;
// Possible splits
static const uint64_t NO_SPLIT = 0x00;
static const uint64_t CENTER_SPLIT = 0x01;
static const uint64_t RIGHT_SPLIT = 0x02;
static const uint64_t LEFT_SPLIT = 0x04;
static const uint64_t RIGHT_DOUBLE_SPLIT = (CENTER_SPLIT | RIGHT_SPLIT);
static const uint64_t LEFT_DOUBLE_SPLIT = (CENTER_SPLIT | LEFT_SPLIT);
static const uint64_t TRIPLE_SPLIT = (CENTER_SPLIT | RIGHT_SPLIT | LEFT_SPLIT);
// Split buffer slots
static const uint64_t SPLIT_COUNTER = 0;
static const uint64_t SIMPLIFY_COUNTER = 1;
static const uint64_t CLASSIFY_COUNTER_OFFSET = 2;
struct BisectorData
{
// Allocated indices for this bisector
uint32_t indices[3];
// Subvision that should be applied to this bisector
uint32_t subdivisionPattern;
// Neighbor that should be processed
uint32_t problematicNeighbor;
// State of this bisector (split, merge, etc)
uint32_t bisectorState;
// Visibility and modification flags of a bisector
uint32_t flags;
// ID used for the propagation
uint32_t propagationID;
};
uint HeapIDDepth(uint64_t x)
{
uint depth = 0;
while (x > 0u) {
++depth;
x >>= 1u;
}
return depth;
}
struct ComputeParams
{
RWStructuredBuffer<uint> indirectDrawBuffer;
RWStructuredBuffer<uint64_t> heapIDBuffer;
RWStructuredBuffer<uint> classificationBuffer;
RWStructuredBuffer<int> allocateBuffer;
RWStructuredBuffer<int> memoryBuffer;
RWStructuredBuffer<uint4> neighboursBuffer;
RWStructuredBuffer<BisectorData> bisectorDataBuffer;
RWStructuredBuffer<int> propagateBuffer;
RWStructuredBuffer<uint> simplifyBuffer;
RWStructuredBuffer<uint> cbtBuffer;
RWStructuredBuffer<uint64_t> bitFieldBuffer;
};
ParameterBlock<ComputeParams> pParams;
#define WORKGROUP_SIZE 64
// Num elements
#define OCBT_NUM_ELEMENTS 131072
// Tree sizes
#define OCBT_TREE_SIZE_BITS (32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32 + 32 * 64 + 16 * 128 + 16 * 256 + 16 * 512 + 8 * 1024)
#define OCBT_TREE_NUM_SLOTS (OCBT_TREE_SIZE_BITS / 32)
#define OCBT_BITFIELD_NUM_SLOTS (OCBT_NUM_ELEMENTS / 64)
#define OCBT_LAST_LEVEL_SIZE 1024
// Tree last level
#define TREE_LAST_LEVEL 10
// First virtual level
#define FIRST_VIRTUAL_LEVEL 11
// Leaf level
#define LEAF_LEVEL 17
// per level offset
static const uint32_t OCBT_depth_offset[18] = { 0, // Level 0
32 * 1, // level 1
32 * 1 + 32 * 2, // level 2
32 * 1 + 32 * 2 + 32 * 4, // level 3
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8, // Level 4
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16, // Level 5
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32, // Level 6
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32 + 32 * 64, // Level 7
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32 + 32 * 64 + 16 * 128, // Level 8
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32 + 32 * 64 + 16 * 128 + 16 * 256, // Level 9
32 * 1 + 32 * 2 + 32 * 4 + 32 * 8 + 32 * 16 + 32 * 32 + 32 * 64 + 16 * 128 + 16 * 256 + 16 * 512, // Level 10
0, // Level 12
0, // Level 13
0, // Level 14
0, // Level 15
0, // Level 16
0, // Level 17
0, // Level 18
};
static const uint64_t OCBT_bit_mask[18] = { 0xffffffff, // Root 17
static const uint32_t OCBT_bit_mask[18] = {
0xffffffff, // Root 17
0xffffffff, // Level 16
0xffffffff, // level 15
0xffffffff, // level 14
@@ -129,7 +19,7 @@ static const uint64_t OCBT_bit_mask[18] = { 0xffffffff, // Root 17
0xffff, // level 8
0xff, // level 8
0xffffffffffffffff, // level 7
0xffffffff, // level 7
0xffffffff, // Level 6
0xffff, // level 5
0xff, // level 4
@@ -138,211 +28,9 @@ static const uint64_t OCBT_bit_mask[18] = { 0xffffffff, // Root 17
0x1, // level 1
};
static const uint32_t OCBT_bit_count[18] = { 32, // Root 17
32, // Level 16
32, // level 15
32, // level 14
32, // level 13
32, // level 12
32, // level 11
16, // level 10
16, // level 9
16, // level 8
8, // level 8
64, // Level 5
32, // Level 5
16, // Level 4
8, // level 3
4, // level 2
2, // level 1
1, // level 0
};
groupshared uint gs_cbtTree[OCBT_TREE_NUM_SLOTS];
// Define the remaining values
#define BUFFER_ELEMENT_PER_LANE ((OCBT_TREE_NUM_SLOTS + WORKGROUP_SIZE - 1) / WORKGROUP_SIZE)
#define BUFFER_ELEMENT_PER_LANE_NO_BITFIELD ((OCBT_TREE_NUM_SLOTS + WORKGROUP_SIZE - 1) / WORKGROUP_SIZE)
#define BITFIELD_ELEMENT_PER_LANE ((OCBT_BITFIELD_NUM_SLOTS + WORKGROUP_SIZE - 1) / WORKGROUP_SIZE)
#define WAVE_TREE_DEPTH uint(log2(OCBT_NUM_ELEMENTS))
void load_buffer_to_shared_memory(uint groupIndex)
[numthreads(1, 1, 1)]
[shader("compute")]
void TestHeap(uint dispatchID: SV_DispatchThreadID)
{
// Load the bitfield to the LDS
for (uint e = 0; e < BUFFER_ELEMENT_PER_LANE; ++e)
{
uint target_element = BUFFER_ELEMENT_PER_LANE * groupIndex + e;
if (target_element < OCBT_TREE_NUM_SLOTS)
gs_cbtTree[target_element] = pParams.cbtBuffer[target_element];
}
GroupMemoryBarrierWithGroupSync();
pParams.bitFieldBuffer[dispatchID] = OCBT_bit_mask[dispatchID];
}
uint get_heap_element(uint id)
{
// Figure out the location of the first bit of this element
uint32_t real_heap_id = id - 1;
uint32_t depth = uint32_t(log2(real_heap_id + 1));
uint32_t level_first_element = (1u << depth) - 1;
uint32_t id_in_level = real_heap_id - level_first_element;
uint32_t first_bit = OCBT_depth_offset[depth] + OCBT_bit_count[depth] * id_in_level;
if (depth < FIRST_VIRTUAL_LEVEL)
{
uint32_t slot = first_bit / 32;
uint32_t local_id = first_bit % 32;
uint32_t target_bits = (gs_cbtTree[slot] >> local_id) & uint32_t(OCBT_bit_mask[depth]);
return (gs_cbtTree[slot] >> local_id) & uint32_t(OCBT_bit_mask[depth]);
}
else
{
uint32_t slot = first_bit / 64;
uint32_t local_id = first_bit % 64;
uint64_t target_bits = (pParams.bitFieldBuffer[slot] >> local_id) & OCBT_bit_mask[depth];
uint32_t high = uint(target_bits >> 32);
uint32_t low = uint(target_bits);
return countbits(high) + countbits(low);
}
}
// decodes the position of the i-th one in the bitfield
uint decode_bit(uint handle)
{
#if defined(NAIVE_DECODE)
uint bitID = 1;
for (uint currentDepth = 0; currentDepth < WAVE_TREE_DEPTH; ++currentDepth)
{
uint heapValue = get_heap_element(2 * bitID);
uint b = handle < heapValue ? 0 : 1;
bitID = 2 * bitID + b;
handle -= heapValue * b;
}
return (bitID ^ OCBT_NUM_ELEMENTS);
#else
uint currentDepth = 0;
uint heapElementID = 1u;
for (currentDepth = 0; currentDepth < FIRST_VIRTUAL_LEVEL; ++currentDepth)
{
// Read the left element
uint heapValue = get_heap_element(2u * heapElementID);
// Does it fall in the right or left subtree?
uint b = handle < heapValue ? 0u : 1u;
// Pick a subtree
heapElementID = 2u * heapElementID + b;
// Move the iterator to exclude the right subtree if required
handle -= heapValue * b;
}
// Align with the internal depth
currentDepth++;
// Ok we have our subtree, now we need to pick the right bit
uint64_t heapValue = pParams.bitFieldBuffer[heapElementID - OCBT_LAST_LEVEL_SIZE * 2];
uint64_t mask = 0xffffffff;
uint32_t bitCount = 32;
for (; currentDepth < (WAVE_TREE_DEPTH + 1); ++currentDepth)
{
// Figure out the location of the first bit of this element
uint real_heap_id = 2 * heapElementID - 1;
uint level_first_element = (1u << currentDepth) - 1;
uint id_in_level = real_heap_id - level_first_element;
uint first_bit = bitCount * id_in_level;
uint local_id = first_bit % 64;
uint64_t target_bits = (heapValue >> local_id) & mask;
uint heapValue = countbits(uint(target_bits >> 32)) + countbits(uint(target_bits));
// Does it fall in the right or left subtree?
uint b = handle < heapValue ? 0u : 1u;
// Pick a subtree
heapElementID = 2u * heapElementID + b;
// Move the iterator to exclude the right subtree if required
handle -= heapValue * b;
// Adjust the mask and bitcount
bitCount /= 2;
mask = mask >> bitCount;
}
return (heapElementID ^ OCBT_NUM_ELEMENTS);
#endif
}
// decodes the position of the i-th zero in the bitfield
uint decode_bit_complement(uint handle)
{
#if defined(NAIVE_DECODE)
uint bitID = 1u;
uint c = OCBT_NUM_ELEMENTS / 2u;
while (bitID < OCBT_NUM_ELEMENTS) {
uint heapValue = c - get_heap_element(2u * bitID);
uint b = handle < heapValue ? 0u : 1u;
bitID = 2u * bitID + b;
handle -= heapValue * b;
c /= 2u;
}
return (bitID ^ OCBT_NUM_ELEMENTS);
#else
uint heapElementID = 1u;
uint c = OCBT_NUM_ELEMENTS / 2u;
uint currentDepth = 0;
for (currentDepth = 0; currentDepth < FIRST_VIRTUAL_LEVEL; ++currentDepth)
{
uint heapValue = c - get_heap_element(2u * heapElementID);
uint b = handle < heapValue ? 0u : 1u;
heapElementID = 2u * heapElementID + b;
handle -= heapValue * b;
c /= 2u;
}
// Align with the internal depth
currentDepth++;
// Ok we have our subtree, now we need to pick the right bit
uint64_t heapValue = pParams.bitFieldBuffer[heapElementID - OCBT_LAST_LEVEL_SIZE * 2];
uint64_t mask = 0xffffffff;
uint32_t bitCount = 32;
for (; currentDepth < (WAVE_TREE_DEPTH + 1); ++currentDepth)
{
// Figure out the location of the first bit of this element
uint real_heap_id = 2 * heapElementID - 1;
uint level_first_element = (1u << currentDepth) - 1;
uint id_in_level = real_heap_id - level_first_element;
uint first_bit = bitCount * id_in_level;
uint local_id = first_bit % 64;
uint64_t target_bits = (heapValue >> local_id) & mask;
uint heapValue = c - countbits(uint(target_bits >> 32)) + countbits(uint(target_bits));;
uint b = handle < heapValue ? 0u : 1u;
heapElementID = 2u * heapElementID + b;
handle -= heapValue * b;
c /= 2u;
// Adjust the mask and bitcount
bitCount /= 2;
mask = mask >> bitCount;
}
return (heapElementID ^ OCBT_NUM_ELEMENTS);
#endif
}
[numthreads(64, 1, 1)]
void GetHeap(uint groupIndex : SV_GroupIndex, uint dispatchID : SV_DispatchThreadID)
{
load_buffer_to_shared_memory(groupIndex);
pParams.classificationBuffer[dispatchID] = decode_bit(dispatchID);
pParams.classificationBuffer[dispatchID * 2] = decode_bit_complement(dispatchID);
}
res/shaders/terrain/CompileTest.spv
BIN
View File
Binary file not shown.
res/shaders/terrain/LEB.slang
+4 -4
View File
@@ -310,9 +310,9 @@ void EvaluateLEB(uint currentID : SV_DispatchThreadID, uint groupIndex: SV_Group
EvaluateElementPosition(cHeapID, 0, pParams.geometry.baseDepth, pParams.currentVertexBuffer, parentTri, childTri);
// Export the child
pParams.lebPositionBuffer[3 * currentID + 0] = float4((childTri.p[0]), 1.0f);
pParams.lebPositionBuffer[3 * currentID + 1] = float4((childTri.p[1]), 1.0f);
pParams.lebPositionBuffer[3 * currentID + 2] = float4((childTri.p[2]), 1.0f);
pParams.lebPositionBuffer[3 * currentID + 0] = float4(normalize(childTri.p[0]), 1.0f);
pParams.lebPositionBuffer[3 * currentID + 1] = float4(normalize(childTri.p[1]), 1.0f);
pParams.lebPositionBuffer[3 * currentID + 2] = float4(normalize(childTri.p[2]), 1.0f);
// Export the fourth element
if (pParams.geometry.baseDepth < depth)
@@ -321,6 +321,6 @@ void EvaluateLEB(uint currentID : SV_DispatchThreadID, uint groupIndex: SV_Group
const uint parentOffset = 3 * pParams.geometry.totalNumElements;
// Transform the coordinate to planet space
pParams.lebPositionBuffer[parentOffset + currentID] = float4((cHeapID % 2 == 0 ? parentTri.p[0] : parentTri.p[2]), 1.0f);
pParams.lebPositionBuffer[parentOffset + currentID] = float4(normalize(cHeapID % 2 == 0 ? parentTri.p[0] : parentTri.p[2]), 1.0f);
}
}
res/shaders/terrain/Parameters.slang
+6 -1
View File
@@ -34,11 +34,16 @@ struct HeapDebug
uint level_first_element;
uint id_in_level;
uint first_bit;
uint handle;
uint heapValue;
uint h;
uint c;
uint64_t target_bits;
};
struct DebugStruct
{
HeapDebug heapValues[18];
HeapDebug heapValues[17];
};
res/shaders/terrain/update_utils.slang
+7 -2
View File
@@ -44,7 +44,7 @@ bool FrustumAABBIntersect(in Frustum frustum, float3 aabbMin, float3 aabbMax)
int ClassifyBisector(in BisectorGeometry tri, uint depth)
{
// Check the triangle's visibility
float3 triNormal = normalize(cross(tri.p[2] - tri.p[0], tri.p[1] - tri.p[0]));
float3 triNormal = normalize(cross(tri.p[2] - tri.p[1], tri.p[0] - tri.p[1]));
float3 triCenter = (tri.p[0] + tri.p[1] + tri.p[2]) / 3.0;
float3 viewDir = normalize(-triCenter);
float FdotV = dot(viewDir, -pViewParams.cameraForward_WS.xyz);
@@ -147,6 +147,9 @@ void ClassifyElement(uint currentID, BisectorGeometry bis, uint totalNumElements
BisectorData cbisectorData = pParams.bisectorDataBuffer[currentID];
// Reset some values
cbisectorData.indices[0] = 0;
cbisectorData.indices[1] = 0;
cbisectorData.indices[2] = 0;
cbisectorData.subdivisionPattern = 0;
cbisectorData.bisectorState = UNCHANGED_ELEMENT;
cbisectorData.problematicNeighbor = INVALID_POINTER;
@@ -325,6 +328,7 @@ void AllocateElement(uint currentID, uint dispatchID)
// Load the bisector for this element
BisectorData bisectorData = pParams.bisectorDataBuffer[currentID];
DebugStruct debug;
// Does this guy need to be subdivided
if (bisectorData.subdivisionPattern != 0)
{
@@ -338,12 +342,13 @@ void AllocateElement(uint currentID, uint dispatchID)
// llocate the bits we need
for (uint bitId = 0; bitId < numSlots; ++bitId)
{
bisectorData.indices[bitId] = decode_bit_complement(firstBitIndex + bitId, dispatchID);
bisectorData.indices[bitId] = decode_bit_complement(firstBitIndex + bitId, debug);
}
// Output
pParams.bisectorDataBuffer[currentID] = bisectorData;
}
pParams.debugBuffer[dispatchID] = debug;
}
#define SUBLING0_ID 0
src/Engine/Graphics/CBT/CBT.cpp
+62 -62
View File
@@ -1,16 +1,16 @@
#include "CBT.h"
#include "Graphics/Shader.h"
#include "Asset/AssetRegistry.h"
#include "Graphics/Shader.h"
using namespace Seele;
Array<Vector4> basePoints = {
//Vector4(0.0f / 3, 0, 0.0f / 3, 1), Vector4(0.0f / 3, 0, 1.0f / 3, 1), Vector4(0.0f / 3, 0, 2.0f / 3, 1),
//Vector4(0.0f / 3, 0, 3.0f / 3, 1), Vector4(1.0f / 3, 0, 0.0f / 3, 1), Vector4(1.0f / 3, 0, 1.0f / 3, 1),
//Vector4(1.0f / 3, 0, 2.0f / 3, 1), Vector4(1.0f / 3, 0, 3.0f / 3, 1), Vector4(2.0f / 3, 0, 0.0f / 3, 1),
//Vector4(2.0f / 3, 0, 1.0f / 3, 1), Vector4(2.0f / 3, 0, 2.0f / 3, 1), Vector4(2.0f / 3, 0, 3.0f / 3, 1),
//Vector4(3.0f / 3, 0, 0.0f / 3, 1), Vector4(3.0f / 3, 0, 1.0f / 3, 1), Vector4(3.0f / 3, 0, 2.0f / 3, 1),
//Vector4(3.0f / 3, 0, 3.0f / 3, 1),
// Vector4(0.0f / 3, 0, 0.0f / 3, 1), Vector4(0.0f / 3, 0, 1.0f / 3, 1), Vector4(0.0f / 3, 0, 2.0f / 3, 1),
// Vector4(0.0f / 3, 0, 3.0f / 3, 1), Vector4(1.0f / 3, 0, 0.0f / 3, 1), Vector4(1.0f / 3, 0, 1.0f / 3, 1),
// Vector4(1.0f / 3, 0, 2.0f / 3, 1), Vector4(1.0f / 3, 0, 3.0f / 3, 1), Vector4(2.0f / 3, 0, 0.0f / 3, 1),
// Vector4(2.0f / 3, 0, 1.0f / 3, 1), Vector4(2.0f / 3, 0, 2.0f / 3, 1), Vector4(2.0f / 3, 0, 3.0f / 3, 1),
// Vector4(3.0f / 3, 0, 0.0f / 3, 1), Vector4(3.0f / 3, 0, 1.0f / 3, 1), Vector4(3.0f / 3, 0, 2.0f / 3, 1),
// Vector4(3.0f / 3, 0, 3.0f / 3, 1),
Vector4(0, -0.525731, 0.850651, 1), Vector4(0.850651, 0, 0.525731, 1), Vector4(0.850651, 0, -0.525731, 1),
Vector4(-0.850651, 0, -0.525731, 1), Vector4(-0.850651, 0, 0.525731, 1), Vector4(-0.525731, 0.850651, 0, 1),
Vector4(0.525731, 0.850651, 0, 1), Vector4(0.525731, -0.850651, 0, 1), Vector4(-0.525731, -0.850651, 0, 1),
@@ -23,77 +23,77 @@ struct Halfedge {
uint32 twinID;
};
Array<Halfedge> edges = {
//Halfedge{0, 1, 2, 4294967295},
//Halfedge{4, 2, 0, 3},
//Halfedge{1, 0, 1, 4294967295},
// Halfedge{0, 1, 2, 4294967295},
// Halfedge{4, 2, 0, 3},
// Halfedge{1, 0, 1, 4294967295},
//
//Halfedge{4, 4, 5, 1},
//Halfedge{5, 5, 3, 8},
//Halfedge{1, 3, 4, 18},
// Halfedge{4, 4, 5, 1},
// Halfedge{5, 5, 3, 8},
// Halfedge{1, 3, 4, 18},
//
//Halfedge{1, 7, 8, 4294967295},
//Halfedge{5, 8, 6, 9},
//Halfedge{2, 6, 7, 4},
// Halfedge{1, 7, 8, 4294967295},
// Halfedge{5, 8, 6, 9},
// Halfedge{2, 6, 7, 4},
//
//Halfedge{5, 10, 11, 7},
//Halfedge{6, 11, 9, 14},
//Halfedge{2, 9, 10, 24},
// Halfedge{5, 10, 11, 7},
// Halfedge{6, 11, 9, 14},
// Halfedge{2, 9, 10, 24},
//
//Halfedge{2, 13, 14, 4294967295},
//Halfedge{6, 14, 12, 15},
//Halfedge{3, 12, 13, 10},
// Halfedge{2, 13, 14, 4294967295},
// Halfedge{6, 14, 12, 15},
// Halfedge{3, 12, 13, 10},
//
//Halfedge{6, 16, 17, 13},
//Halfedge{7, 17, 15, 4294967295},
//Halfedge{3, 15, 16, 30},
// Halfedge{6, 16, 17, 13},
// Halfedge{7, 17, 15, 4294967295},
// Halfedge{3, 15, 16, 30},
//
//Halfedge{4, 19, 20, 5},
//Halfedge{8, 20, 18, 21},
//Halfedge{5, 18, 19, 4294967295},
// Halfedge{4, 19, 20, 5},
// Halfedge{8, 20, 18, 21},
// Halfedge{5, 18, 19, 4294967295},
//
//Halfedge{8, 22, 23, 19},
//Halfedge{9, 23, 21, 26},
//Halfedge{5, 21, 22, 36},
// Halfedge{8, 22, 23, 19},
// Halfedge{9, 23, 21, 26},
// Halfedge{5, 21, 22, 36},
//
//Halfedge{5, 25, 26, 11},
//Halfedge{9, 26, 24, 27},
//Halfedge{6, 24, 25, 22},
// Halfedge{5, 25, 26, 11},
// Halfedge{9, 26, 24, 27},
// Halfedge{6, 24, 25, 22},
//
//Halfedge{9, 28, 29, 25},
//Halfedge{10, 29, 27, 32},
//Halfedge{6, 27, 28, 42},
// Halfedge{9, 28, 29, 25},
// Halfedge{10, 29, 27, 32},
// Halfedge{6, 27, 28, 42},
//
//Halfedge{6, 31, 32, 17},
//Halfedge{10, 32, 30, 33},
//Halfedge{7, 30, 31, 28},
// Halfedge{6, 31, 32, 17},
// Halfedge{10, 32, 30, 33},
// Halfedge{7, 30, 31, 28},
//
//Halfedge{10, 34, 35, 31},
//Halfedge{11, 35, 33, 4294967295},
//Halfedge{7, 33, 34, 48},
// Halfedge{10, 34, 35, 31},
// Halfedge{11, 35, 33, 4294967295},
// Halfedge{7, 33, 34, 48},
//
//Halfedge{8, 37, 38, 23},
//Halfedge{12, 38, 36, 39},
//Halfedge{9, 36, 37, 4294967295},
// Halfedge{8, 37, 38, 23},
// Halfedge{12, 38, 36, 39},
// Halfedge{9, 36, 37, 4294967295},
//
//Halfedge{12, 40, 41, 37},
//Halfedge{13, 41, 39, 44},
//Halfedge{9, 39, 40, 4294967295},
// Halfedge{12, 40, 41, 37},
// Halfedge{13, 41, 39, 44},
// Halfedge{9, 39, 40, 4294967295},
//
//Halfedge{9, 43, 44, 29},
//Halfedge{13, 44, 42, 45},
//Halfedge{10, 42, 43, 40},
// Halfedge{9, 43, 44, 29},
// Halfedge{13, 44, 42, 45},
// Halfedge{10, 42, 43, 40},
//
//Halfedge{13, 46, 47, 43},
//Halfedge{14, 47, 45, 50},
//Halfedge{10, 45, 46, 4294967295},
// Halfedge{13, 46, 47, 43},
// Halfedge{14, 47, 45, 50},
// Halfedge{10, 45, 46, 4294967295},
//
//Halfedge{10, 49, 50, 35},
//Halfedge{14, 50, 48, 51},
//Halfedge{11, 48, 49, 46},
// Halfedge{10, 49, 50, 35},
// Halfedge{14, 50, 48, 51},
// Halfedge{11, 48, 49, 46},
//
//Halfedge{14, 52, 53, 49},
//Halfedge{15, 53, 51, 4294967295},
//Halfedge{11, 51, 52, 4294967295},
// Halfedge{14, 52, 53, 49},
// Halfedge{15, 53, 51, 4294967295},
// Halfedge{11, 51, 52, 4294967295},
Halfedge{1, 1, 2, 5}, Halfedge{2, 2, 0, 36}, Halfedge{6, 0, 1, 39}, Halfedge{1, 4, 5, 51}, Halfedge{7, 5, 3, 48},
Halfedge{2, 3, 4, 0}, Halfedge{3, 7, 8, 9}, Halfedge{4, 8, 6, 45}, Halfedge{5, 6, 7, 42}, Halfedge{4, 10, 11, 6},
@@ -609,7 +609,7 @@ void MeshUpdater::update(CBTMesh& mesh, Gfx::PDescriptorSet viewParamsSet, Gfx::
mesh.propagateBuffer->pipelineBarrier(Gfx::SE_ACCESS_SHADER_WRITE_BIT, Gfx::SE_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
Gfx::SE_ACCESS_SHADER_READ_BIT, Gfx::SE_PIPELINE_STAGE_COMPUTE_SHADER_BIT);
mesh.gpuCBT.bufferArray[1]->pipelineBarrier(Gfx::SE_ACCESS_SHADER_WRITE_BIT, Gfx::SE_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
Gfx::SE_ACCESS_SHADER_READ_BIT, Gfx::SE_PIPELINE_STAGE_COMPUTE_SHADER_BIT);
Gfx::SE_ACCESS_SHADER_READ_BIT, Gfx::SE_PIPELINE_STAGE_COMPUTE_SHADER_BIT);
}
graphics->endDebugRegion();
src/Engine/Graphics/RenderPass/TerrainRenderer.cpp
+2 -2
View File
@@ -13,7 +13,7 @@ TerrainRenderer::TerrainRenderer(Gfx::PGraphics graphics, PScene scene, Gfx::PDe
Gfx::OPipelineLayout test = graphics->createPipelineLayout();
graphics->beginShaderCompilation(ShaderCompilationInfo{
.modules = {"CompileTest"},
.entryPoints = {{"GetHeap", "CompileTest"}},
.entryPoints = {{"TestHeap", "CompileTest"}},
.rootSignature = test,
});
graphics->createComputeShader({0});
@@ -302,7 +302,7 @@ void TerrainRenderer::setViewport(Gfx::PViewport _viewport, Gfx::PRenderPass ren
.rasterizationState =
{
.polygonMode = Gfx::SE_POLYGON_MODE_LINE,
.cullMode = Gfx::SE_CULL_MODE_BACK_BIT,
.cullMode = Gfx::SE_CULL_MODE_NONE,
},
.colorBlend =
{
src/Engine/Graphics/Vulkan/Graphics.cpp
+18 -2
View File
@@ -35,6 +35,8 @@ PFN_vkCmdDrawMeshTasksIndirectEXT cmdDrawMeshTasksIndirect;
PFN_vkSetDebugUtilsObjectNameEXT setDebugUtilsObjectName;
PFN_vkQueueBeginDebugUtilsLabelEXT queueBeginDebugUtilsLabelEXT;
PFN_vkQueueEndDebugUtilsLabelEXT queueEndDebugUtilsLabelEXT;
PFN_vkCmdBeginDebugUtilsLabelEXT cmdBeginDebugUtilsLabelEXT;
PFN_vkCmdEndDebugUtilsLabelEXT cmdEndDebugUtilsLabelEXT;
PFN_vkCreateAccelerationStructureKHR createAccelerationStructure;
PFN_vkCmdBuildAccelerationStructuresKHR cmdBuildAccelerationStructures;
PFN_vkGetAccelerationStructureBuildSizesKHR getAccelerationStructureBuildSize;
@@ -71,6 +73,18 @@ void vkQueueEndDebugUtilsLabelEXT(VkQueue queue) {
}
}
void vkCmdBeginDebugUtilsLabelEXT(VkCommandBuffer commandBuffer, const VkDebugUtilsLabelEXT* pLabelInfo) {
if (cmdBeginDebugUtilsLabelEXT != nullptr) {
cmdBeginDebugUtilsLabelEXT(commandBuffer, pLabelInfo);
}
}
void vkCmdEndDebugUtilsLabelEXT(VkCommandBuffer commandBuffer) {
if (cmdEndDebugUtilsLabelEXT != nullptr) {
cmdEndDebugUtilsLabelEXT(commandBuffer);
}
}
VkResult vkCreateAccelerationStructureKHR(VkDevice device, const VkAccelerationStructureCreateInfoKHR* pCreateInfo,
const VkAllocationCallbacks* pAllocator, VkAccelerationStructureKHR* pAccelerationStructure) {
return createAccelerationStructure(device, pCreateInfo, pAllocator, pAccelerationStructure);
@@ -314,10 +328,10 @@ void Graphics::beginDebugRegion(const std::string& name) {
.pNext = nullptr,
.pLabelName = name.c_str(),
};
vkQueueBeginDebugUtilsLabelEXT(queues[graphicsQueue]->getHandle(), &label);
vkCmdBeginDebugUtilsLabelEXT(getGraphicsCommands()->getCommands()->getHandle(), &label);
}
void Graphics::endDebugRegion() { vkQueueEndDebugUtilsLabelEXT(queues[graphicsQueue]->getHandle()); }
void Graphics::endDebugRegion() { vkCmdEndDebugUtilsLabelEXT(getGraphicsCommands()->getCommands()->getHandle()); }
void Graphics::resolveTexture(Gfx::PTexture source, Gfx::PTexture destination) {
PTextureBase sourceTex = source.cast<TextureBase>();
@@ -952,6 +966,8 @@ void Graphics::createDevice(GraphicsInitializer initializer) {
setDebugUtilsObjectName = (PFN_vkSetDebugUtilsObjectNameEXT)vkGetInstanceProcAddr(instance, "vkSetDebugUtilsObjectNameEXT");
queueBeginDebugUtilsLabelEXT = (PFN_vkQueueBeginDebugUtilsLabelEXT)vkGetInstanceProcAddr(instance, "vkQueueBeginDebugUtilsLabelEXT");
queueEndDebugUtilsLabelEXT = (PFN_vkQueueEndDebugUtilsLabelEXT)vkGetInstanceProcAddr(instance, "vkQueueEndDebugUtilsLabelEXT");
cmdBeginDebugUtilsLabelEXT = (PFN_vkCmdBeginDebugUtilsLabelEXT)vkGetInstanceProcAddr(instance, "vkCmdBeginDebugUtilsLabelEXT");
cmdEndDebugUtilsLabelEXT = (PFN_vkCmdEndDebugUtilsLabelEXT)vkGetInstanceProcAddr(instance, "vkCmdEndDebugUtilsLabelEXT");
createAccelerationStructure = (PFN_vkCreateAccelerationStructureKHR)vkGetDeviceProcAddr(handle, "vkCreateAccelerationStructureKHR");
cmdBuildAccelerationStructures =