From 52ccaf5b25e14c9ce032315e5e96b1f27886d481 Mon Sep 17 00:00:00 2001 From: Cedric Nugteren Date: Thu, 16 Jun 2016 18:07:46 +0200 Subject: Added XOMATCOPY routines to perform out-of-place matrix scaling, copying, and/or transposing --- src/tuning/copy_fast.cc | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'src/tuning/copy_fast.cc') diff --git a/src/tuning/copy_fast.cc b/src/tuning/copy_fast.cc index 2da707be..09fdbaba 100644 --- a/src/tuning/copy_fast.cc +++ b/src/tuning/copy_fast.cc @@ -37,7 +37,7 @@ class TuneCopy { } // The list of arguments relevant for this routine - static std::vector GetOptions() { return {kArgM, kArgN}; } + static std::vector GetOptions() { return {kArgM, kArgN, kArgAlpha}; } // Tests for valid arguments static void TestValidArguments(const Arguments &) { } @@ -86,9 +86,11 @@ class TuneCopy { std::vector &, std::vector &, std::vector &a_mat, std::vector &b_mat, std::vector &, std::vector &) { + auto alpha_buffer = std::vector{args.alpha}; tuner.AddArgumentScalar(static_cast(args.m)); tuner.AddArgumentInput(a_mat); tuner.AddArgumentOutput(b_mat); + tuner.AddArgumentInput(alpha_buffer); } // Describes how to compute the performance metrics -- cgit v1.2.3