GrayScottSeq/src/gray-scott.cpp

+
 #include <vector>
 #include <iostream>
 #include <iomanip>
@@ -6,26 +7,24 @@
 
 using real = double ;
 
-const real KILL_RATE { 0.062f };
-const real FEED_RATE { 0.03f };
-const real DT { 1.0f };
-const real DIFFUSION_RATE_U { 0.1f };
-const real DIFFUSION_RATE_V { 0.05f };
+constexpr real KILL_RATE { 0.062l };
+constexpr real FEED_RATE { 0.03l };
+constexpr real DT { 1.0l };
+constexpr real DIFFUSION_RATE_U { 0.1l };
+constexpr real DIFFUSION_RATE_V { 0.05l };
 
 template<typename Collection>
 void drop( Collection & data, std::size_t nb_rows, std::size_t nb_cols, real value ) {
-
     const std::size_t center_row { nb_rows/2ul };
     const std::size_t center_col { nb_cols/2ul };
-    for (std::size_t i = (center_row-2ul) ; i < (center_row+2ul); i++) {
-        for (std::size_t j = (center_col-3ul); j < (center_col+3ul); j++) {
-            data[i*nb_cols+j] = value ;
+    for (std::size_t row = (center_row-2ul) ; row < (center_row+2ul); row++) {
+        for (std::size_t col = (center_col-3ul); col < (center_col+3ul); col++) {
+            data[row*nb_cols+col] = value ;
         }
     }
-
 }
 
-void process(
+void transform(
     std::vector<real> const & iu, std::vector<real> const & iv,
     std::vector<real> & ou, std::vector<real> & ov,
     std::size_t nb_rows, std::size_t nb_cols ) {
@@ -37,16 +36,16 @@ void process(
             real v = iv[(r+1)*nb_cols+c+1];
             real uvv = u*v*v;
 
-            real full_u = 0.0f;
-            real full_v = 0.0f;
-            for(std::size_t k = 0ul; k < 3ul; ++k){
-                for(std::size_t l = 0ul; l < 3ul; ++l){
-                    full_u += iu[(r+k)*nb_cols+c+l] - u;
-                    full_v += iv[(r+k)*nb_cols+c+l] - v;
+            real full_u = 0.l;
+            real full_v = 0.l;
+            for(std::size_t dr = 0ul; dr < 3ul; ++dr){
+                for(std::size_t dc = 0ul; dc < 3ul; ++dc){
+                    full_u += iu[(r+dr)*nb_cols+c+dc] - u;
+                    full_v += iv[(r+dr)*nb_cols+c+dc] - v;
                 }
             }
 
-            real du = DIFFUSION_RATE_U*full_u - uvv + FEED_RATE*(1.0f - u);
+            real du = DIFFUSION_RATE_U*full_u - uvv + FEED_RATE*(static_cast<real>(1.0l) - u);
             real dv = DIFFUSION_RATE_V*full_v + uvv - (FEED_RATE + KILL_RATE)*v;
 
             ou[(r+1)*nb_cols+c+1] = u + du*DT;
@@ -65,17 +64,18 @@ int main( int argc, char * argv[] ) {
     std::size_t nb_images {std::stoul(argv[3])};
     std::size_t nb_iterations {std::stoul(argv[4])};
     std::cout << std::fixed << std::setprecision(6);
+    assert(nb_iterations%2==0);
     assert(nb_images<=1000);
 
     try {
 
         // Temporary work images
-        std::vector<real> u1(nb_rows*nb_cols,1.f);
-        std::vector<real> v1(nb_rows*nb_cols,0.f);
-        std::vector<real> u2(nb_rows*nb_cols,1.f);
-        std::vector<real> v2(nb_rows*nb_cols,0.f);
-        drop(u1, nb_rows, nb_cols, 0.f);
-        drop(v1, nb_rows, nb_cols, 1.f);
+        std::vector<real> u1(nb_rows*nb_cols,1.l);
+        std::vector<real> v1(nb_rows*nb_cols,0.l);
+        std::vector<real> u2(nb_rows*nb_cols,1.l);
+        std::vector<real> v2(nb_rows*nb_cols,0.l);
+        drop(u1, nb_rows, nb_cols, 0.l);
+        drop(v1, nb_rows, nb_cols, 1.l);
 
         // Final images sequence
         std::vector<std::vector<real>> images;
@@ -85,7 +85,7 @@ int main( int argc, char * argv[] ) {
         std::cout<<std::endl ;
         for ( std::size_t image = 0 ; image < nb_images ; ++image ) {
             for ( std::size_t iter = 0 ; iter < nb_iterations ; ++iter ) {
-                process( u1, v1, u2, v2, nb_rows, nb_cols );
+                transform( u1, v1, u2, v2, nb_rows, nb_cols );
                 std::swap(u1,u2) ; std::swap (v1,v2) ;
             }
             std::cout << '.' << std::flush;
@@ -95,7 +95,7 @@ int main( int argc, char * argv[] ) {
         
         // Final product
         std::cout << std::endl;
-        real product = std::inner_product(u1.begin(), u1.end(), v1.begin(),0.f);
+        real product = std::inner_product(u1.begin(), u1.end(), v1.begin(),static_cast<real>(0.l));
         std::cout
           << "product : " << product
           << std::endl;

SetupInstructions/intel-devcloud.md

@@ -32,5 +32,5 @@ gpusub intel.bash 2
 
 The helper function `qclean` will remove all the output of the previous jobs you have submitted.
 
-Have a look at `intel.bash` to see the various compile and run steps, and optionally adapt it to your own needs.
+Have a look at `intel.bash` to see the various compile/run steps, and optionally adapt it to your own needs.
 

SetupInstructions/intel-docker.md

@@ -23,22 +23,20 @@ cd GrayScottSyclSetup/
 
 # check CPU nodes
 IMG=gitlab-registry.in2p3.fr/codeursintensifs/grayscott/grayscottsyclsetup/jfalcou:2024.1
-docker run --gpus all --device=/dev/dri --network host -it --rm -v ${PWD}:/work -w /work ${IMG}
-cd CheckOneApiCpu
-./sycl-ls.bash
-./cmake.bash
-./make.bash
-./run.bash
+docker run --network host -it --rm -v ${PWD}:/work -w /work ${IMG}
+cd CheckOneApi
+./intel.bash   # give the list of available devices
+./intel.bash 1 # check the results of device 1
 exit
 
 # Check GPU nodes
 IMG=gitlab-registry.in2p3.fr/codeursintensifs/grayscott/grayscottsyclsetup/dchamont:2024.1
 docker run --gpus all --device=/dev/dri --network host -it --rm -v ${PWD}:/work -w /work ${IMG}
-cd CheckOneApiCuda
+cd CheckOneApi
 nvidia-smi
-./sycl-ls.bash
-./cmake.bash
-./make.bash
-./run.bash
+./cuda.bash   # give the list of available devices
+./cuda.bash 4 # give the list of available devices
 exit
 ```
+
+Have a look at `intel.bash` to see the various compile and run steps, and optionally adapt it to your own needs.

docker/jfalcou/Dockerfile

@@ -10,7 +10,7 @@ RUN pacman-key --init
 RUN pacman -Syu --noconfirm --needed                                                        \
     gcc nano cmake git intel-oneapi-compiler-dpcpp-cpp-runtime-libs intel-oneapi-dpcpp-cpp
 
-RUN pacman -Syu --noconfirm --needed ninja
+RUN pacman -Syu --noconfirm --needed make
 
 RUN rm -rf /var/cache/pacman/pkg/