From e8880c1699816483090aa5574cf9b3322943831f Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 7 Jun 2018 10:26:55 +0200
Subject: [PATCH] Use a single thread for small input size

copies daxpy improvement from #27, see #1560
---
 interface/zaxpy.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/interface/zaxpy.c b/interface/zaxpy.c
index fbb830ffb..529e78e79 100644
--- a/interface/zaxpy.c
+++ b/interface/zaxpy.c
@@ -41,7 +41,11 @@
 #ifdef FUNCTION_PROFILE
 #include "functable.h"
 #endif
-
+#if  defined(Z13)
+#define MULTI_THREAD_MINIMAL  200000
+#else
+#define MULTI_THREAD_MINIMAL  10000
+#endif
 #ifndef CBLAS
 
 void NAME(blasint *N, FLOAT *ALPHA, FLOAT *x, blasint *INCX, FLOAT *y, blasint *INCY){
@@ -69,7 +73,7 @@ void CNAME(blasint n, FLOAT *ALPHA, FLOAT *x, blasint incx, FLOAT *y, blasint in
 #endif
 
 #ifndef CBLAS
-  PRINT_DEBUG_CNAME;
+  PRINT_DEBUG_NAME;
 #else
   PRINT_DEBUG_CNAME;
 #endif
@@ -93,6 +97,11 @@ void CNAME(blasint n, FLOAT *ALPHA, FLOAT *x, blasint incx, FLOAT *y, blasint in
   if (incx == 0 || incy == 0)
 	  nthreads = 1;
 
+  //Work around the low performance issue with small imput size &
+  //multithreads.
+  if (n <= MULTI_THREAD_MINIMAL) {
+	  nthreads = 1;
+  }
   if (nthreads == 1) {
 #endif