Clean up canonical mt benchmarks and reorganize extra versions in /mt. All versions support support at least 1/2/4 threads.

1 files changed, 17 insertions, 0 deletions

diff --git a/mt/vvadd1.c b/mt/vvadd1.c
new file mode 100755
index 0000000..b7e44a7
--- /dev/null
+++ b/mt/vvadd1.c
@@ -0,0 +1,17 @@
+#include "stdlib.h"
+#include "dataset.h"
+
+void __attribute__((noinline)) vvadd(int coreid, int ncores, size_t n, const data_t* x, const data_t* y, data_t* z)
+{
+   size_t i;
+   for (i = coreid*4; i < n; i += 8*ncores) {
+      z[i]   = x[i]   + y[i];
+      z[i+1] = x[i+1] + y[i+1];
+      z[i+2] = x[i+2] + y[i+2];
+      z[i+3] = x[i+3] + y[i+3];
+      z[i+ncores*4]   = x[i+ncores*4]   + y[i+ncores*4];
+      z[i+ncores*4+1] = x[i+ncores*4+1] + y[i+ncores*4+1];
+      z[i+ncores*4+2] = x[i+ncores*4+2] + y[i+ncores*4+2];
+      z[i+ncores*4+3] = x[i+ncores*4+3] + y[i+ncores*4+3];
+   }
+}