Added an extra level of indirection for the GPU benchmark to work

hpvm/test/benchmarks/dhpvm/benchmarks/sgemm/src/hpvm/main.cc

@@ -105,7 +105,7 @@ void basicSgemmLvl2(float *A, size_t bytes_A, int lda, float *B, size_t bytes_B,
                     int ldb, float *C, size_t bytes_C, int ldc, int k,
                     float alpha, float beta, size_t dim_X1, size_t dim_Y1,
                     size_t dim_X2, size_t dim_Y2) {
-  __hpvm__hint(hpvm::REMOTE_TARGET);
+  __hpvm__hint(hpvm::CPU_TARGET);
   __hpvm__attributes(3, A, B, C, 1, C);
   void *sgemm_node =
       __hpvm__createNodeND(2, basicSgemmLvl1, (size_t)dim_X2, (size_t)dim_Y2);
@@ -130,7 +130,7 @@ void basicSgemmLvl3(float *A, size_t bytes_A, int lda, float *B, size_t bytes_B,
                     int ldb, float *C, size_t bytes_C, int ldc, int k,
                     float alpha, float beta, size_t dim_X1, size_t dim_Y1,
                     size_t dim_X2, size_t dim_Y2) {
-  __hpvm__hint(hpvm::CPU_TARGET);
+  __hpvm__hint(hpvm::REMOTE_TARGET);
   __hpvm__attributes(3, A, B, C, 1, C);
   void *sgemm_node = __hpvm__createNodeND(0, basicSgemmLvl2);
   __hpvm__bindIn(sgemm_node, 0, 0, 0);
@@ -151,6 +151,32 @@ void basicSgemmLvl3(float *A, size_t bytes_A, int lda, float *B, size_t bytes_B,
   __hpvm__bindIn(sgemm_node, 15, 15, 0);
 }
 
+// A wrapper level used in codegen for some backends
+void basicSgemmLvl4(float *A, size_t bytes_A, int lda, float *B, size_t bytes_B,
+                    int ldb, float *C, size_t bytes_C, int ldc, int k,
+                    float alpha, float beta, size_t dim_X1, size_t dim_Y1,
+                    size_t dim_X2, size_t dim_Y2) {
+  __hpvm__hint(hpvm::CPU_TARGET);
+  __hpvm__attributes(3, A, B, C, 1, C);
+  void *sgemm_node = __hpvm__createNodeND(0, basicSgemmLvl3);
+  __hpvm__bindIn(sgemm_node, 0, 0, 0);
+  __hpvm__bindIn(sgemm_node, 1, 1, 0);
+  __hpvm__bindIn(sgemm_node, 2, 2, 0);
+  __hpvm__bindIn(sgemm_node, 3, 3, 0);
+  __hpvm__bindIn(sgemm_node, 4, 4, 0);
+  __hpvm__bindIn(sgemm_node, 5, 5, 0);
+  __hpvm__bindIn(sgemm_node, 6, 6, 0);
+  __hpvm__bindIn(sgemm_node, 7, 7, 0);
+  __hpvm__bindIn(sgemm_node, 8, 8, 0);
+  __hpvm__bindIn(sgemm_node, 9, 9, 0);
+  __hpvm__bindIn(sgemm_node, 10, 10, 0);
+  __hpvm__bindIn(sgemm_node, 11, 11, 0);
+  __hpvm__bindIn(sgemm_node, 12, 12, 0);
+  __hpvm__bindIn(sgemm_node, 13, 13, 0);
+  __hpvm__bindIn(sgemm_node, 14, 14, 0);
+  __hpvm__bindIn(sgemm_node, 15, 15, 0);
+}
+
 __attribute__((noinline)) void basicSgemm(char transa, char transb, int m,
                                           int n, int k, float alpha, float *A,
                                           size_t bytesA, int lda, float *B,
@@ -194,7 +220,7 @@ __attribute__((noinline)) void basicSgemm(char transa, char transb, int m,
                           dg[0] / db[0],
                           dg[1] / db[1]};
   *(RootIn *)root_in = root_in_local;
-  void *sgemmDFG = __hpvm__launch(0, basicSgemmLvl3, root_in);
+  void *sgemmDFG = __hpvm__launch(0, basicSgemmLvl4, root_in);
   __hpvm__wait(sgemmDFG);
 }