diff --git a/coreblas/compute/core_ztpmlqt.c b/coreblas/compute/core_ztpmlqt.c
index 162375f966f1076860acd44854cacaff9ea1381f..bd5e29973702c9a5d739de15e82756f3469d9562 100644
--- a/coreblas/compute/core_ztpmlqt.c
+++ b/coreblas/compute/core_ztpmlqt.c
@@ -112,8 +112,8 @@
  *        V = [V1] [V2].
  *
  *  The size of the trapezoidal block V2 is determined by the parameter L,
- *  where 0 <= L <= K; V2 is upper trapezoidal, consisting of the first L
- *  rows of a K-by-K upper triangular matrix.  If L=K, V2 is upper triangular;
+ *  where 0 <= L <= K; V2 is lower trapezoidal, consisting of the first L
+ *  rows of a K-by-K upper triangular matrix.  If L=K, V2 is lower triangular;
  *  if L=0, there is no trapezoidal block, hence V = V1 is rectangular.
  *
  *  If side = MorseLeft:  C = [A]  where A is K-by-N,  B is M-by-N and V is K-by-M.
diff --git a/runtime/starpu/codelets/codelet_ztpmlqt.c b/runtime/starpu/codelets/codelet_ztpmlqt.c
index 2363d67c0cbd74f7ab954db90bfa8e3f6817b906..090ce92f2af8af120e3f8e08e4de28f022026551 100644
--- a/runtime/starpu/codelets/codelet_ztpmlqt.c
+++ b/runtime/starpu/codelets/codelet_ztpmlqt.c
@@ -54,7 +54,7 @@ static void cl_ztpmlqt_cpu_func(void *descr[], void *cl_arg)
                   V, ldv, T, ldt, A, lda, B, ldb, WORK );
 }
 
-#if defined(CHAMELEON_USE_CUDA)
+#if defined(CHAMELEON_USE_CUDA) && 0
 static void cl_ztpmlqt_cuda_func(void *descr[], void *cl_arg)
 {
     MORSE_enum side;
@@ -100,7 +100,8 @@ static void cl_ztpmlqt_cuda_func(void *descr[], void *cl_arg)
 /*
  * Codelet definition
  */
-CODELETS(ztpmlqt, 5, cl_ztpmlqt_cpu_func, cl_ztpmlqt_cuda_func, STARPU_CUDA_ASYNC)
+CODELETS_CPU(ztpmlqt, 5, cl_ztpmlqt_cpu_func)
+//CODELETS(ztpmlqt, 5, cl_ztpmlqt_cpu_func, cl_ztpmlqt_cuda_func, STARPU_CUDA_ASYNC)
 
 void
 MORSE_TASK_ztpmlqt( const MORSE_option_t *options,