POC for JP

Signed-off-by: James Brodman <[email protected]>

Author: jbrodman

sycl/include/CL/sycl/handler.hpp

@@ -871,7 +871,25 @@ class __SYCL_EXPORT handler {
 
   template <typename KernelName = detail::auto_name, typename KernelType>
   void parallel_for(range<1> NumWorkItems, KernelType KernelFunc) {
-    parallel_for_lambda_impl<KernelName>(NumWorkItems, std::move(KernelFunc));
+    bool isPrime = true;
+    size_t R = NumWorkItems[0];
+    for (int i = 2; i <= R/2; i++) {
+      if (R % i == 0) {
+        isPrime = false;
+        break;
+      }
+    }
+    if (isPrime) {
+      size_t R64 = (R + 63) & -64;
+      parallel_for_lambda_impl<KernelName>(range<1>(R64), [=](id<1> ID) {
+        if (ID < NumWorkItems) {
+          KernelFunc(ID);
+        }
+      });
+    } else {
+      parallel_for_lambda_impl<KernelName>(NumWorkItems, std::move(KernelFunc));
+    }
+    
   }
 
   template <typename KernelName = detail::auto_name, typename KernelType>

sycl/test/usm/john.cpp

@@ -0,0 +1,57 @@
+// XFAIL: cuda
+// piextUSM*Alloc functions for CUDA are not behaving as described in
+// https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/USM/USM.adoc
+// https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/USM/cl_intel_unified_shared_memory.asciidoc
+//
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t1.out
+// RUN: env SYCL_DEVICE_TYPE=HOST %t1.out
+// RUN: %CPU_RUN_PLACEHOLDER %t1.out
+// RUN: %GPU_RUN_PLACEHOLDER %t1.out
+
+//==------------------- mixed.cpp - Mixed Memory test ---------------------==//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include <CL/sycl.hpp>
+
+using namespace sycl;
+
+int main() {
+  const int MAGIC_NUM = 42;
+  const int N = 17;
+  
+  queue q;
+  auto dev = q.get_device();
+  auto ctxt = q.get_context();
+
+  auto A = malloc_shared<int>(N, q);
+  auto B = malloc_shared<int>(N, q);
+  auto C = malloc_shared<int>(std::max(N,64), q);
+
+  for (int i = 0; i < N; i++) {
+    A[i] = 1;
+    B[i] = 2;
+  }
+
+  q.parallel_for(range<1>(N), [=](id<1> i) { C[i] = A[i] + B[i]; });
+  q.wait();
+
+  for (int i = 0; i < N; i++) {
+    assert(C[i] == 3);
+  }
+
+  for (int i = 0; i < 64; i++) {
+    std::cout << C[i] << " ";
+  }
+  std::cout << std::endl;
+  
+  free(A, q);
+  free(B, q);
+  free(C, q);
+
+  return 0;
+}