adding tensor regression test.

tests/regression/tensor/kernel.cpp

@@ -0,0 +1,41 @@
+#include <stdint.h>
+#include <vx_intrinsics.h>
+#include <vx_spawn.h>
+#include "common.h"
+
+inline char is_log2(uint32_t x) {
+    return ((x & (x-1)) == 0);
+}
+
+inline uint32_t log2_fast(uint32_t x) { 
+    return 31 - __builtin_clz (x);
+}
+
+void kernel_body(uint32_t task_id, kernel_arg_t* __UNIFORM__ arg) {
+	auto size  = arg->size;
+    auto A = reinterpret_cast<TYPE*>(arg->A_addr);
+	auto B = reinterpret_cast<TYPE*>(arg->B_addr);
+	auto C = reinterpret_cast<TYPE*>(arg->C_addr);
+
+    uint32_t row, col;
+    if (is_log2(size)) {
+        uint32_t log_size = log2_fast(size);
+        row = task_id >> log_size;
+        col = task_id & (size-1);
+    } else {
+        row = task_id / size;
+        col = task_id % size;
+    }
+
+    TYPE sum (0);
+    for (int e = 0; e < size; ++e) {
+        sum += A[row * size + e] * B[e * size + col];
+    }
+    C[row * size + col] = sum;
+}
+
+int main() {
+	kernel_arg_t* arg = (kernel_arg_t*)KERNEL_ARG_DEV_MEM_ADDR;
+	vx_spawn_tasks(arg->num_tasks, (vx_spawn_tasks_cb)kernel_body, arg);
+	return 0;
+}