summaryrefslogtreecommitdiff
path: root/src/routines
diff options
context:
space:
mode:
authorCedric Nugteren <web@cedricnugteren.nl>2016-05-18 21:32:56 +0200
committerCedric Nugteren <web@cedricnugteren.nl>2016-05-18 21:32:56 +0200
commit489c5d76cfe95a97542dfeaa6d8b19cd9100919a (patch)
tree31a7082f5847f3bd21af1f2aa5a7d1eb68d188db /src/routines
parent7a3b695db70810595ae17d9d753c3b926aa738c0 (diff)
Merged in latest changes from 0.7.1 release
Diffstat (limited to 'src/routines')
-rw-r--r--src/routines/level3/xgemm.cc5
1 files changed, 3 insertions, 2 deletions
diff --git a/src/routines/level3/xgemm.cc b/src/routines/level3/xgemm.cc
index 5395667a..ab36076c 100644
--- a/src/routines/level3/xgemm.cc
+++ b/src/routines/level3/xgemm.cc
@@ -191,12 +191,13 @@ StatusCode Xgemm<T>::DoGemm(const Layout layout,
// Launches the kernel
auto eventKernel = Event();
- status = RunKernel(kernel, global, local, eventKernel.pointer(), eventWaitList);
+ auto eventPointer = (!c_no_temp) ? eventKernel.pointer() : event_;
+ status = RunKernel(kernel, global, local, eventPointer, eventWaitList);
if (ErrorIn(status)) { return status; }
- eventWaitList.push_back(eventKernel);
// Runs the post-processing kernel if needed
if (!c_no_temp) {
+ eventWaitList.push_back(eventKernel);
status = PadCopyTransposeMatrix(event_, eventWaitList,
m_ceiled, n_ceiled, m_ceiled, 0, c_temp,
c_one, c_two, c_ld, c_offset, c_buffer,