-
Notifications
You must be signed in to change notification settings - Fork 28
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #340 from lks1248/gpu-observables
gpu observables
- Loading branch information
Showing
12 changed files
with
451 additions
and
107 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,8 +1,8 @@ | ||
| if(CMAKE_HIP_COMPILER) | ||
| set_source_files_properties(conserved_gpu.cu PROPERTIES LANGUAGE HIP) | ||
| set_source_files_properties(conserved_gpu.cu gpu_reductions.cu PROPERTIES LANGUAGE HIP) | ||
| endif() | ||
|
|
||
| if(CMAKE_CUDA_COMPILER OR CMAKE_HIP_COMPILER) | ||
| add_library(observables_gpu conserved_gpu.cu) | ||
| target_include_directories(observables_gpu PRIVATE ${CSTONE_DIR}) | ||
| add_library(observables_gpu conserved_gpu.cu gpu_reductions.cu) | ||
| target_include_directories(observables_gpu PRIVATE ${CSTONE_DIR} ${SPH_DIR}) | ||
| endif() |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,183 @@ | ||
| /* | ||
| * MIT License | ||
| * | ||
| * Copyright (c) 2022 CSCS, ETH Zurich | ||
| * 2021 University of Basel | ||
| * | ||
| * Permission is hereby granted, free of charge, to any person obtaining a copy | ||
| * of this software and associated documentation files (the "Software"), to deal | ||
| * in the Software without restriction, including without limitation the rights | ||
| * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell | ||
| * copies of the Software, and to permit persons to whom the Software is | ||
| * furnished to do so, subject to the following conditions: | ||
| * | ||
| * The above copyright notice and this permission notice shall be included in all | ||
| * copies or substantial portions of the Software. | ||
| * | ||
| * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR | ||
| * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, | ||
| * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE | ||
| * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER | ||
| * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, | ||
| * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE | ||
| * SOFTWARE. | ||
| */ | ||
|
|
||
| /*! @file | ||
| * @brief reductions for different observables on the GPU | ||
| * | ||
| * @author Lukas Schmidt | ||
| */ | ||
|
|
||
| #include <thrust/execution_policy.h> | ||
| #include <thrust/iterator/zip_iterator.h> | ||
| #include <thrust/transform_reduce.h> | ||
|
|
||
| #include "cstone/util/tuple.hpp" | ||
| #include "gpu_reductions.h" | ||
|
|
||
| namespace sphexa | ||
| { | ||
|
|
||
| using cstone::Box; | ||
| using cstone::Vec3; | ||
| using thrust::get; | ||
|
|
||
| //!@brief functor for Kelvin-Helmholtz growth rate | ||
| template<class Tc, class Tv, class Tm> | ||
| struct GrowthRate | ||
| { | ||
| HOST_DEVICE_FUN | ||
| thrust::tuple<double, double, double> operator()(const thrust::tuple<Tc, Tc, Tv, Tm, Tm>& p) | ||
| { | ||
| Tc xi = get<0>(p); | ||
| Tc yi = get<1>(p); | ||
| Tv vyi = get<2>(p); | ||
| Tm xmi = get<3>(p); | ||
| Tm kxi = get<4>(p); | ||
| Tc voli = xmi / kxi; | ||
| Tc aux; | ||
|
|
||
| if (yi < ybox * Tc(0.5)) { aux = std::exp(-4.0 * PI * std::abs(yi - 0.25)); } | ||
| else { aux = std::exp(-4.0 * PI * std::abs(ybox - yi - 0.25)); } | ||
| Tc si = vyi * voli * std::sin(4.0 * PI * xi) * aux; | ||
| Tc ci = vyi * voli * std::cos(4.0 * PI * xi) * aux; | ||
| Tc di = voli * aux; | ||
|
|
||
| return thrust::make_tuple(si, ci, di); | ||
| } | ||
|
|
||
| Tc ybox; | ||
| }; | ||
|
|
||
| template<class Tc, class Tv, class Tm> | ||
| std::tuple<double, double, double> gpuGrowthRate(const Tc* x, const Tc* y, const Tv* vy, const Tm* xm, const Tm* kx, | ||
| const cstone::Box<Tc>& box, size_t startIndex, size_t endIndex) | ||
| { | ||
| auto it1 = thrust::make_zip_iterator( | ||
| thrust::make_tuple(x + startIndex, y + startIndex, vy + startIndex, xm + startIndex, kx + startIndex)); | ||
|
|
||
| auto it2 = thrust::make_zip_iterator( | ||
| thrust::make_tuple(x + endIndex, y + endIndex, vy + endIndex, xm + endIndex, kx + endIndex)); | ||
|
|
||
| auto plus = util::TuplePlus<thrust::tuple<double, double, double>>{}; | ||
| auto init = thrust::make_tuple(0.0, 0.0, 0.0); | ||
|
|
||
| auto result = thrust::transform_reduce(thrust::device, it1, it2, GrowthRate<Tc, Tm, Tv>{box.ly()}, init, plus); | ||
|
|
||
| return {get<0>(result), get<1>(result), get<2>(result)}; | ||
| } | ||
|
|
||
| #define GROWTH_GPU(Tc, Tv, Tm) \ | ||
| template std::tuple<double, double, double> gpuGrowthRate(const Tc* x, const Tc* y, const Tv* vy, const Tm* xm, \ | ||
| const Tm* kx, const cstone::Box<Tc>& box, \ | ||
| size_t startIndex, size_t endIndex) | ||
|
|
||
| GROWTH_GPU(double, double, double); | ||
| GROWTH_GPU(double, double, float); | ||
| GROWTH_GPU(double, float, float); | ||
| GROWTH_GPU(float, float, float); | ||
|
|
||
| //!@brief functor for the machSquare summing | ||
| template<class Tc, class Tv> | ||
| struct MachSquareSum | ||
| { | ||
| HOST_DEVICE_FUN | ||
| double operator()(const thrust::tuple<Tv, Tv, Tv, Tc>& p) | ||
| { | ||
| Vec3<double> V{get<0>(p), get<1>(p), get<2>(p)}; | ||
| Tc c = get<3>(p); | ||
| return norm2(V) / (double(c) * c); | ||
| } | ||
| }; | ||
|
|
||
| template<class Tc, class Tv> | ||
| double machSquareSumGpu(const Tv* vx, const Tv* vy, const Tv* vz, const Tc* c, size_t first, size_t last) | ||
| { | ||
| auto it1 = thrust::make_zip_iterator(thrust::make_tuple(vx + first, vy + first, vz + first, c + first)); | ||
| auto it2 = thrust::make_zip_iterator(thrust::make_tuple(vx + last, vy + last, vz + last, c + last)); | ||
|
|
||
| auto plus = thrust::plus<double>{}; | ||
| double localMachSquareSum = 0.0; | ||
|
|
||
| localMachSquareSum = | ||
| thrust::transform_reduce(thrust::device, it1, it2, MachSquareSum<Tc, Tv>{}, localMachSquareSum, plus); | ||
|
|
||
| return localMachSquareSum; | ||
| } | ||
|
|
||
| #define MACH_GPU(Tc, Tv) \ | ||
| template double machSquareSumGpu(const Tv* vx, const Tv* vy, const Tv* vz, const Tc* c, size_t, size_t) | ||
|
|
||
| MACH_GPU(double, double); | ||
| MACH_GPU(double, float); | ||
| MACH_GPU(float, float); | ||
|
|
||
| //!@brief functor to count particles that belong to the cloud | ||
| template<class Tc, class Tt, class Tm> | ||
| struct Survivors | ||
| { | ||
| HOST_DEVICE_FUN | ||
| double operator()(const thrust::tuple<Tt, Tc, Tc, Tm>& p) | ||
| { | ||
| size_t isCloud; | ||
| Tt temp = get<0>(p); | ||
| Tc kx = get<1>(p); | ||
| Tc xm = get<2>(p); | ||
| Tm m = get<3>(p); | ||
| Tc rhoi = kx / xm * m; | ||
| if (rhoi >= 0.64 * rhoBubble && temp <= 0.9 * tempWind) { isCloud = 1; } | ||
| else { isCloud = 0; } | ||
| return isCloud; | ||
| } | ||
|
|
||
| double rhoBubble; | ||
| double tempWind; | ||
| }; | ||
| template<class Tc, class Tt, class Tm> | ||
| size_t survivorsGpu(const Tt* temp, const Tt* kx, const Tc* xmass, const Tm* m, double rhoBubble, double tempWind, | ||
| size_t first, size_t last) | ||
| { | ||
| auto it1 = thrust::make_zip_iterator(thrust::make_tuple(temp + first, kx + first, xmass + first, m + first)); | ||
| auto it2 = thrust::make_zip_iterator(thrust::make_tuple(temp + last, kx + last, xmass + last, m + last)); | ||
|
|
||
| auto plus = thrust::plus<size_t>{}; | ||
|
|
||
| size_t localSurvivors = 0; | ||
|
|
||
| localSurvivors = thrust::transform_reduce(thrust::device, it1, it2, Survivors<Tc, Tt, Tm>{rhoBubble, tempWind}, | ||
| localSurvivors, plus); | ||
|
|
||
| return localSurvivors; | ||
| } | ||
|
|
||
| #define SURVIVORS(Tc, Tt, Tm) \ | ||
| template size_t survivorsGpu(const Tt* temp, const Tt* kx, const Tc* xmass, const Tm* m, double, double, size_t, \ | ||
| size_t) | ||
|
|
||
| SURVIVORS(double, double, double); | ||
| SURVIVORS(double, double, float); | ||
| SURVIVORS(double, float, float); | ||
| SURVIVORS(float, float, float); | ||
|
|
||
| } // namespace sphexa |
Oops, something went wrong.