diff options
Diffstat (limited to 'APEX_1.4/module/basicios/cuda')
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/common.h | 63 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/compact.h | 17 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/histogram.h | 23 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/moduleList.h | 16 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/reduce.h | 19 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/scan.h | 25 | ||||
| -rw-r--r-- | APEX_1.4/module/basicios/cuda/include/simulate.h | 93 |
7 files changed, 256 insertions, 0 deletions
diff --git a/APEX_1.4/module/basicios/cuda/include/common.h b/APEX_1.4/module/basicios/cuda/include/common.h new file mode 100644 index 00000000..e530b9bc --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/common.h @@ -0,0 +1,63 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +#ifndef __COMMON_H__ +#define __COMMON_H__ + +#define APEX_CUDA_MODULE_PREFIX BasicIOS_ + +#include "ApexCuda.h" +#include "InplaceTypes.h" +#include "IofxManagerIntl.h" +#include <float.h> + +#if PX_WINDOWS_FAMILY +#pragma warning(push) +#pragma warning(disable:4201) +#pragma warning(disable:4408) +#endif + +#include <vector_types.h> + +#if PX_WINDOWS_FAMILY +#pragma warning(pop) +#endif + +const unsigned int HISTOGRAM_BIN_COUNT = 256; +const unsigned int HISTOGRAM_SIMULATE_BIN_COUNT = 512; + + +#define COMPACT_KERNEL_CONFIG (0, WARP_SIZE * 3) +#define HISTOGRAM_KERNEL_CONFIG (0, HISTOGRAM_BIN_COUNT) +#define REDUCE_KERNEL_CONFIG (0, WARP_SIZE * 4) +#define SCAN_KERNEL_CONFIG (0, WARP_SIZE * 4) +#define SIMULATE_KERNEL_CONFIG (0, HISTOGRAM_SIMULATE_BIN_COUNT) + + +const unsigned int HOLE_SCAN_FLAG_BIT = 31; +const unsigned int HOLE_SCAN_FLAG = (1U << HOLE_SCAN_FLAG_BIT); +const unsigned int HOLE_SCAN_MASK = (HOLE_SCAN_FLAG - 1); + +// mTmpOutput +const unsigned int STATUS_LAST_ACTIVE_COUNT = 0; +const unsigned int STATUS_LAST_BENEFIT_SUM = 1; +const unsigned int STATUS_LAST_BENEFIT_MIN = 2; +const unsigned int STATUS_LAST_BENEFIT_MAX = 3; + +namespace nvidia +{ +namespace basicios +{ + +} +} // namespace nvidia + +#endif diff --git a/APEX_1.4/module/basicios/cuda/include/compact.h b/APEX_1.4/module/basicios/cuda/include/compact.h new file mode 100644 index 00000000..f0313c09 --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/compact.h @@ -0,0 +1,17 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +APEX_CUDA_TEXTURE_1D(texRefCompactScanSum, unsigned int) + +APEX_CUDA_BOUND_KERNEL(COMPACT_KERNEL_CONFIG, compactKernel, + ((unsigned int, targetCount))((unsigned int, totalCount))((unsigned int, injectorCount)) + ((APEX_MEM_BLOCK(unsigned int), g_outIndices))((APEX_MEM_BLOCK(unsigned int), g_outCount))((APEX_MEM_BLOCK(unsigned int), g_injCounters)) + ) diff --git a/APEX_1.4/module/basicios/cuda/include/histogram.h b/APEX_1.4/module/basicios/cuda/include/histogram.h new file mode 100644 index 00000000..ca45aa5e --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/histogram.h @@ -0,0 +1,23 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +APEX_CUDA_SYNC_KERNEL(HISTOGRAM_KERNEL_CONFIG, histogramSyncKernel, ((unsigned int, count)) + ((const float*, g_data))((unsigned int, bound)) + ((float, dataMin))((float, dataMax))((unsigned int*, g_boundParams)) + ((unsigned int*, g_tmpHistograms)) + ) + +APEX_CUDA_BOUND_KERNEL(HISTOGRAM_KERNEL_CONFIG, histogramKernel, + ((APEX_MEM_BLOCK(float), g_data))((unsigned int, bound)) + ((float, dataMin))((float, dataMax))((APEX_MEM_BLOCK(unsigned int), g_boundParams)) + ((APEX_MEM_BLOCK(unsigned int), g_tmpHistograms)) + ((unsigned int, phase))((unsigned int, gridSize)) + ) diff --git a/APEX_1.4/module/basicios/cuda/include/moduleList.h b/APEX_1.4/module/basicios/cuda/include/moduleList.h new file mode 100644 index 00000000..8a8749ce --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/moduleList.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +#include "compact.h" +#include "histogram.h" +#include "reduce.h" +#include "scan.h" +#include "simulate.h" diff --git a/APEX_1.4/module/basicios/cuda/include/reduce.h b/APEX_1.4/module/basicios/cuda/include/reduce.h new file mode 100644 index 00000000..adde0ad7 --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/reduce.h @@ -0,0 +1,19 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +APEX_CUDA_SYNC_KERNEL(REDUCE_KERNEL_CONFIG, reduceSyncKernel, ((unsigned int, count)) + ((float*, g_benefit))((float4*, g_output))((unsigned int*, g_tmp)) + ) + +APEX_CUDA_BOUND_KERNEL(REDUCE_KERNEL_CONFIG, reduceKernel, + ((APEX_MEM_BLOCK(float), g_benefit))((APEX_MEM_BLOCK(float4), g_output))((APEX_MEM_BLOCK(unsigned int), g_tmp)) + ((unsigned int, phase))((unsigned int, gridSize)) + ) diff --git a/APEX_1.4/module/basicios/cuda/include/scan.h b/APEX_1.4/module/basicios/cuda/include/scan.h new file mode 100644 index 00000000..12553c5f --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/scan.h @@ -0,0 +1,25 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +APEX_CUDA_SYNC_KERNEL(SCAN_KERNEL_CONFIG, scanSyncKernel, ((unsigned int, count)) + ((float, benefitMin))((float, benefitMax)) + ((unsigned int*, g_indices))((const float*, g_benefits)) + ((unsigned int*, g_boundParams)) + ((unsigned int*, g_tmpCounts))((unsigned int*, g_tmpCounts1)) + ) + +APEX_CUDA_BOUND_KERNEL(SCAN_KERNEL_CONFIG, scanKernel, + ((float, benefitMin))((float, benefitMax)) + ((APEX_MEM_BLOCK(unsigned int), g_indices))((APEX_MEM_BLOCK(float), g_benefits)) + ((APEX_MEM_BLOCK(unsigned int), g_boundParams)) + ((APEX_MEM_BLOCK(unsigned int), g_tmpCounts))((APEX_MEM_BLOCK(unsigned int), g_tmpCounts1)) + ((unsigned int, phase))((unsigned int, gridSize)) + ) diff --git a/APEX_1.4/module/basicios/cuda/include/simulate.h b/APEX_1.4/module/basicios/cuda/include/simulate.h new file mode 100644 index 00000000..5ade0039 --- /dev/null +++ b/APEX_1.4/module/basicios/cuda/include/simulate.h @@ -0,0 +1,93 @@ +/* + * Copyright (c) 2008-2015, NVIDIA CORPORATION. All rights reserved. + * + * NVIDIA CORPORATION and its licensors retain all intellectual property + * and proprietary rights in and to this software, related documentation + * and any modifications thereto. Any use, reproduction, disclosure or + * distribution of this software and related documentation without an express + * license agreement from NVIDIA CORPORATION is strictly prohibited. + */ + + +APEX_CUDA_STORAGE(simulateStorage) + + +APEX_CUDA_TEXTURE_1D(texRefPositionMass, float4) +APEX_CUDA_TEXTURE_1D(texRefVelocityLife, float4) +APEX_CUDA_TEXTURE_1D(texRefIofxActorIDs, unsigned int) +APEX_CUDA_TEXTURE_1D(texRefLifeSpan, float) +APEX_CUDA_TEXTURE_1D(texRefLifeTime, float) +APEX_CUDA_TEXTURE_1D(texRefInjector, unsigned int) + +APEX_CUDA_TEXTURE_1D(texRefUserData, unsigned int) + +APEX_CUDA_TEXTURE_1D(texRefConvexPlanes, float4) +APEX_CUDA_TEXTURE_1D(texRefConvexVerts, float4) +APEX_CUDA_TEXTURE_1D(texRefConvexPolygonsData, unsigned int) + +APEX_CUDA_TEXTURE_1D(texRefTrimeshIndices, unsigned int) +APEX_CUDA_TEXTURE_1D(texRefTrimeshVerts, float4) + +APEX_CUDA_TEXTURE_1D(texRefHoleScanSum, unsigned int) +APEX_CUDA_TEXTURE_1D(texRefMoveIndices, unsigned int) + +APEX_CUDA_TEXTURE_1D(texRefField, float4) + + +APEX_CUDA_BOUND_KERNEL(SIMULATE_KERNEL_CONFIG, simulateKernel, + ((unsigned int, lastCount))((float, deltaTime))((physx::PxVec3, gravity))((physx::PxVec3, eyePos)) + ((InplaceHandle<InjectorParamsArray>, injectorParamsArrayHandle))((unsigned int, injectorCount)) + ((unsigned int*, g_holeScanSum))((unsigned int*, g_moveCount))((unsigned int*, g_tmpHistogram))((unsigned int*, g_InjectorsCounters)) + ((float4*, g_positionMass))((float4*, g_velocityLife))((float4*, g_collisionNormalFlags))((unsigned int*, g_userData)) + ((float*, g_lifeSpan))((float*, g_lifeTime))((unsigned int*, g_injector))((IofxActorIDIntl*, g_iofxActorIDs)) + ((float*, g_benefit))((InplaceHandle<SimulationParams>, paramsHandle)) + ) + +APEX_CUDA_BOUND_KERNEL(SIMULATE_KERNEL_CONFIG, simulateApplyFieldKernel, + ((unsigned int, lastCount))((float, deltaTime))((physx::PxVec3, gravity))((physx::PxVec3, eyePos)) + ((InplaceHandle<InjectorParamsArray>, injectorParamsArrayHandle))((unsigned int, injectorCount)) + ((APEX_MEM_BLOCK(unsigned int), g_holeScanSum))((APEX_MEM_BLOCK(unsigned int), g_moveCount)) + ((APEX_MEM_BLOCK(unsigned int), g_tmpHistogram))((APEX_MEM_BLOCK(unsigned int), g_InjectorsCounters)) + ((APEX_MEM_BLOCK(float4), g_positionMass))((APEX_MEM_BLOCK(float4), g_velocityLife)) + ((APEX_MEM_BLOCK(float4), g_collisionNormalFlags))((APEX_MEM_BLOCK(unsigned int), g_userData)) + ((APEX_MEM_BLOCK(float), g_lifeSpan))((APEX_MEM_BLOCK(float), g_lifeTime)) + ((APEX_MEM_BLOCK(unsigned int), g_injector))((APEX_MEM_BLOCK(IofxActorIDIntl), g_iofxActorIDs)) + ((APEX_MEM_BLOCK(float), g_benefit))((InplaceHandle<SimulationParams>, paramsHandle)) + ) + + +APEX_CUDA_BOUND_KERNEL((), mergeHistogramKernel, + ((APEX_MEM_BLOCK(unsigned int), g_InjectorsCounters)) + ((APEX_MEM_BLOCK(unsigned int), g_tmpHistograms)) + ((unsigned int, gridSize))((unsigned int, injectorCount)) + ) + +APEX_CUDA_BOUND_KERNEL((), stateKernel, + ((unsigned int, lastCount))((unsigned int, targetCount)) + ((APEX_MEM_BLOCK(unsigned int), g_moveCount)) + ((APEX_MEM_BLOCK(unsigned int), g_inStateToInput))((APEX_MEM_BLOCK(unsigned int), g_outStateToInput)) + ) + +APEX_CUDA_BOUND_KERNEL((), gridDensityGridClearKernel, + ((float*, gridDensityGrid))((GridDensityParams, params)) + ) + +APEX_CUDA_BOUND_KERNEL((), gridDensityGridFillKernel, + ((float4*, positionMass))((float*, gridDensityGrid))((GridDensityParams, params)) + ) + +APEX_CUDA_BOUND_KERNEL((), gridDensityGridApplyKernel, + ((float*, density))((float4*, positionMass))((float*, gridDensityGrid))((GridDensityParams, params)) + ) + +APEX_CUDA_BOUND_KERNEL((), gridDensityGridFillFrustumKernel, + ((float4*, positionMass))((float*, gridDensityGrid))((GridDensityParams, params))((::physx::PxMat44,mat))((GridDensityFrustumParams,frustum)) + ) + +APEX_CUDA_BOUND_KERNEL((), gridDensityGridApplyFrustumKernel, + ((float*, density))((float4*, positionMass))((float*, gridDensityGrid))((GridDensityParams, params))((::physx::PxMat44,mat))((GridDensityFrustumParams,frustum)) + ) + +APEX_CUDA_FREE_KERNEL((), gridDensityGridLowPassKernel, + ((float*, gridDensityGridIn))((float*, gridDensityGridOut))((GridDensityParams, params)) + )
\ No newline at end of file |