changed the default sub-group size to 32 (#1406)

Co-authored-by: Praveen Kundurthy <u137620@s001-n067.aidevcloud>

Author: praveenkk123

DirectProgramming/C++SYCL/Jupyter/oneapi-essentials-training/04_SYCL_Sub_Groups/Sub_Groups.ipynb

@@ -357,7 +357,7 @@
     "    auto out = stream(1024, 768, h);\n",
     "\n",
     "    //# nd-range kernel with user specified sub_group size\n",
-    "    h.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(8)]] {\n",
+    "    h.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(32)]] {\n",
     "      //# get sub_group handle\n",
     "      auto sg = item.get_sub_group();\n",
     "\n",
@@ -783,7 +783,7 @@
     "  for(int i=0; i<N; i++) std::cout << input[i] << \" \"; std::cout << \"\\n\";  \n",
     "\n",
     "  //# use parallel_for and sub_groups\n",
-    "  q.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(8)]] {\n",
+    "  q.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(32)]] {\n",
     "    auto sg = item.get_sub_group();\n",
     "    auto i = item.get_global_id(0);\n",
     "\n",
@@ -976,7 +976,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3.7 (Intel® oneAPI)",
+   "display_name": "Python 3 (Intel® oneAPI 2023.0)",
    "language": "python",
    "name": "c009-intel_distribution_of_python_3_oneapi-beta05-python"
   },
@@ -990,7 +990,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.7"
+   "version": "3.9.15"
   },
   "toc": {
    "base_numbering": 1,

DirectProgramming/C++SYCL/Jupyter/oneapi-essentials-training/04_SYCL_Sub_Groups/lab/sub_group_reqd_size.cpp

@@ -27,7 +27,7 @@ int main() {
     auto out = stream(1024, 768, h);
 
     //# nd-range kernel with user specified sub_group size
-    h.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(8)]] {
+    h.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(32)]] {
       //# get sub_group handle
       auto sg = item.get_sub_group();
 

DirectProgramming/C++SYCL/Jupyter/oneapi-essentials-training/04_SYCL_Sub_Groups/lab/sub_group_votes.cpp

@@ -25,7 +25,7 @@ int main() {
   for(int i=0; i<N; i++) std::cout << input[i] << " "; std::cout << "\n";  
 
   //# use parallel_for and sub_groups
-  q.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(8)]] {
+  q.parallel_for(nd_range<1>(N, B), [=](nd_item<1> item)[[intel::reqd_sub_group_size(32)]] {
     auto sg = item.get_sub_group();
     auto i = item.get_global_id(0);