bitonic sort sample

13_BitonicSort/CMakeLists.txt

@@ -0,0 +1,24 @@
+include(common RESULT_VARIABLE RES)
+if(NOT RES)
+	message(FATAL_ERROR "common.cmake not found. Should be in {repo_root}/cmake directory")
+endif()
+
+nbl_create_executable_project("" "" "" "" "${NBL_EXECUTABLE_PROJECT_CREATION_PCH_TARGET}")
+
+if(NBL_EMBED_BUILTIN_RESOURCES)
+	set(_BR_TARGET_ ${EXECUTABLE_NAME}_builtinResourceData)
+	set(RESOURCE_DIR "app_resources")
+
+	get_filename_component(_SEARCH_DIRECTORIES_ "${CMAKE_CURRENT_SOURCE_DIR}" ABSOLUTE)
+	get_filename_component(_OUTPUT_DIRECTORY_SOURCE_ "${CMAKE_CURRENT_BINARY_DIR}/src" ABSOLUTE)
+	get_filename_component(_OUTPUT_DIRECTORY_HEADER_ "${CMAKE_CURRENT_BINARY_DIR}/include" ABSOLUTE)
+
+    file(GLOB_RECURSE BUILTIN_RESOURCE_FILES RELATIVE "${CMAKE_CURRENT_SOURCE_DIR}/${RESOURCE_DIR}" "${CMAKE_CURRENT_SOURCE_DIR}/${RESOURCE_DIR}/*")
+    foreach(RES_FILE ${BUILTIN_RESOURCE_FILES})
+      LIST_BUILTIN_RESOURCE(RESOURCES_TO_EMBED "${RES_FILE}")
+    endforeach()
+
+	ADD_CUSTOM_BUILTIN_RESOURCES(${_BR_TARGET_} RESOURCES_TO_EMBED "${_SEARCH_DIRECTORIES_}" "${RESOURCE_DIR}" "nbl::this_example::builtin" "${_OUTPUT_DIRECTORY_HEADER_}" "${_OUTPUT_DIRECTORY_SOURCE_}")
+
+	LINK_BUILTIN_RESOURCES_TO_TARGET(${EXECUTABLE_NAME} ${_BR_TARGET_})
+endif()

13_BitonicSort/app_resources/bitonic_sort_shader.comp.hlsl

@@ -0,0 +1,112 @@
+#include "nbl/builtin/hlsl/bda/bda_accessor.hlsl"
+
+struct BitonicPushData
+{
+    uint64_t inputKeyAddress;
+    uint64_t inputValueAddress;
+    uint64_t outputKeyAddress;
+    uint64_t outputValueAddress;
+    uint32_t dataElementCount;
+};
+
+using namespace nbl::hlsl;
+
+[[vk::push_constant]] BitonicPushData pushData;
+
+using DataPtr = bda::__ptr<uint32_t>;
+using DataAccessor = BdaAccessor<uint32_t>;
+
+groupshared uint32_t sharedKeys[ElementCount];
+groupshared uint32_t sharedValues[ElementCount];
+
+[numthreads(WorkgroupSize, 1, 1)]
+[shader("compute")]
+void main(uint32_t3 dispatchId : SV_DispatchThreadID, uint32_t3 localId : SV_GroupThreadID)
+{
+    const uint32_t threadId = localId.x;
+    const uint32_t dataSize = pushData.dataElementCount;
+
+    DataAccessor inputKeys = DataAccessor::create(DataPtr::create(pushData.inputKeyAddress));
+    DataAccessor inputValues = DataAccessor::create(DataPtr::create(pushData.inputValueAddress));
+
+    for (uint32_t i = threadId; i < dataSize; i += WorkgroupSize)
+    {
+        inputKeys.get(i, sharedKeys[i]);
+        inputValues.get(i, sharedValues[i]);
+    }
+
+    // Synchronize all threads after loading
+    GroupMemoryBarrierWithGroupSync();
+
+
+    for (uint32_t stage = 0; stage < Log2ElementCount; stage++)
+    {
+        for (uint32_t pass = 0; pass <= stage; pass++)
+        {
+            const uint32_t compareDistance = 1 << (stage - pass);
+
+            for (uint32_t i = threadId; i < dataSize; i += WorkgroupSize)
+            {
+                const uint32_t partnerId = i ^ compareDistance;
+
+                if (partnerId >= dataSize)
+                    continue;
+
+                const uint32_t waveSize = WaveGetLaneCount();
+                const uint32_t myWaveId = i / waveSize;
+                const uint32_t partnerWaveId = partnerId / waveSize;
+                const bool sameWave = (myWaveId == partnerWaveId);
+
+                uint32_t myKey, myValue, partnerKey, partnerValue;
+                [branch]
+                if (sameWave && compareDistance < waveSize)
+                {
+                    // WAVE INTRINSIC
+                    myKey = sharedKeys[i];
+                    myValue = sharedValues[i];
+
+                    const uint32_t partnerLane = partnerId % waveSize;
+                    partnerKey = WaveReadLaneAt(myKey, partnerLane);
+                    partnerValue = WaveReadLaneAt(myValue, partnerLane);
+                }
+                else
+                {
+                    // SHARED MEM
+                    myKey = sharedKeys[i];
+                    myValue = sharedValues[i];
+                    partnerKey = sharedKeys[partnerId];
+                    partnerValue = sharedValues[partnerId];
+                }
+
+                const uint32_t sequenceSize = 1 << (stage + 1);
+                const uint32_t sequenceIndex = i / sequenceSize;
+                const bool sequenceAscending = (sequenceIndex % 2) == 0;
+                const bool ascending = true;
+                const bool finalDirection = sequenceAscending == ascending;
+
+                const bool swap = (myKey > partnerKey) == finalDirection;
+
+                // WORKGROUP COORDINATION: Only lower-indexed element writes both
+                if (i < partnerId && swap)
+                {
+                    sharedKeys[i] = partnerKey;
+                    sharedKeys[partnerId] = myKey;
+                    sharedValues[i] = partnerValue;
+                    sharedValues[partnerId] = myValue;
+                }
+            }
+
+            GroupMemoryBarrierWithGroupSync();
+        }
+    }
+
+
+    DataAccessor outputKeys = DataAccessor::create(DataPtr::create(pushData.outputKeyAddress));
+    DataAccessor outputValues = DataAccessor::create(DataPtr::create(pushData.outputValueAddress));
+
+    for (uint32_t i = threadId; i < dataSize; i += WorkgroupSize)
+    {
+        outputKeys.set(i, sharedKeys[i]);
+        outputValues.set(i, sharedValues[i]);
+    }
+}

13_BitonicSort/app_resources/common.hlsl

@@ -0,0 +1,17 @@
+// Copyright (C) 2018-2024 - DevSH Graphics Programming Sp. z O.O.
+// This file is part of the "Nabla Engine".
+// For conditions of distribution and use, see copyright notice in nabla.h
+#ifndef _BITONIC_SORT_COMMON_INCLUDED_
+#define _BITONIC_SORT_COMMON_INCLUDED_
+
+struct BitonicPushData
+{
+
+	uint64_t inputKeyAddress;
+	uint64_t inputValueAddress;
+	uint64_t outputKeyAddress;
+	uint64_t outputValueAddress;
+	uint32_t dataElementCount;
+};
+
+#endif

13_BitonicSort/config.json.template

@@ -0,0 +1,28 @@
+{
+  "enableParallelBuild": true,
+  "threadsPerBuildProcess" : 2,
+  "isExecuted": false,
+  "scriptPath": "",
+  "cmake": {
+    "configurations": [ "Release", "Debug", "RelWithDebInfo" ],
+    "buildModes": [],
+    "requiredOptions": []
+  }, 
+  "profiles": [
+    {
+      "backend": "vulkan", // should be none
+      "platform": "windows",
+      "buildModes": [],
+      "runConfiguration": "Release", // we also need to run in Debug nad RWDI because foundational example
+      "gpuArchitectures": []
+    }
+  ],
+  "dependencies": [],
+  "data": [
+    {
+      "dependencies": [],
+      "command": [""],
+      "outputs": []
+    }
+  ]
+}