summaryrefslogtreecommitdiff
path: root/test/routines/level1
diff options
context:
space:
mode:
Diffstat (limited to 'test/routines/level1')
-rw-r--r--test/routines/level1/xamax.hpp22
-rw-r--r--test/routines/level1/xasum.hpp18
-rw-r--r--test/routines/level1/xaxpy.hpp22
-rw-r--r--test/routines/level1/xcopy.hpp22
-rw-r--r--test/routines/level1/xdot.hpp25
-rw-r--r--test/routines/level1/xdotc.hpp25
-rw-r--r--test/routines/level1/xdotu.hpp25
-rw-r--r--test/routines/level1/xnrm2.hpp22
-rw-r--r--test/routines/level1/xscal.hpp19
-rw-r--r--test/routines/level1/xswap.hpp22
10 files changed, 152 insertions, 70 deletions
diff --git a/test/routines/level1/xamax.hpp b/test/routines/level1/xamax.hpp
index 868a79ed..d74807c9 100644
--- a/test/routines/level1/xamax.hpp
+++ b/test/routines/level1/xamax.hpp
@@ -69,13 +69,21 @@ class TestXamax {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Amax<T>(args.n,
- buffers.scalar(), args.imax_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Amax<T>(args.n,
+ buffers.scalar(), args.imax_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Amax<T>(args.n,
+ buffers.scalar(), args.imax_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xasum.hpp b/test/routines/level1/xasum.hpp
index 6add9c64..573f1223 100644
--- a/test/routines/level1/xasum.hpp
+++ b/test/routines/level1/xasum.hpp
@@ -69,13 +69,21 @@ class TestXasum {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ #ifdef OPENCL_API
auto queue_plain = queue();
auto event = cl_event{};
- auto status = Asum<T>(args.n,
- buffers.scalar(), args.asum_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ auto status = Asum<T>(args.n,
+ buffers.scalar(), args.asum_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Asum<T>(args.n,
+ buffers.scalar(), args.asum_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xaxpy.hpp b/test/routines/level1/xaxpy.hpp
index 17cae6ad..7491a9e8 100644
--- a/test/routines/level1/xaxpy.hpp
+++ b/test/routines/level1/xaxpy.hpp
@@ -70,13 +70,21 @@ class TestXaxpy {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Axpy(args.n, args.alpha,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Axpy(args.n, args.alpha,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Axpy(args.n, args.alpha,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xcopy.hpp b/test/routines/level1/xcopy.hpp
index 7a5c99b8..58abdbf4 100644
--- a/test/routines/level1/xcopy.hpp
+++ b/test/routines/level1/xcopy.hpp
@@ -69,13 +69,21 @@ class TestXcopy {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Copy<T>(args.n,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Copy<T>(args.n,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Copy<T>(args.n,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xdot.hpp b/test/routines/level1/xdot.hpp
index 1ea25994..229d18c9 100644
--- a/test/routines/level1/xdot.hpp
+++ b/test/routines/level1/xdot.hpp
@@ -73,14 +73,23 @@ class TestXdot {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Dot<T>(args.n,
- buffers.scalar(), args.dot_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Dot<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Dot<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xdotc.hpp b/test/routines/level1/xdotc.hpp
index c800c1f5..9a1dc33a 100644
--- a/test/routines/level1/xdotc.hpp
+++ b/test/routines/level1/xdotc.hpp
@@ -73,14 +73,23 @@ class TestXdotc {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Dotc<T>(args.n,
- buffers.scalar(), args.dot_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Dotc<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Dotc<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xdotu.hpp b/test/routines/level1/xdotu.hpp
index 3545a3a6..4b2c7647 100644
--- a/test/routines/level1/xdotu.hpp
+++ b/test/routines/level1/xdotu.hpp
@@ -73,14 +73,23 @@ class TestXdotu {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Dotu<T>(args.n,
- buffers.scalar(), args.dot_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Dotu<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Dotu<T>(args.n,
+ buffers.scalar(), args.dot_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xnrm2.hpp b/test/routines/level1/xnrm2.hpp
index 1db70537..f3a789b5 100644
--- a/test/routines/level1/xnrm2.hpp
+++ b/test/routines/level1/xnrm2.hpp
@@ -69,13 +69,21 @@ class TestXnrm2 {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Nrm2<T>(args.n,
- buffers.scalar(), args.nrm2_offset,
- buffers.x_vec(), args.x_offset, args.x_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Nrm2<T>(args.n,
+ buffers.scalar(), args.nrm2_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Nrm2<T>(args.n,
+ buffers.scalar(), args.nrm2_offset,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xscal.hpp b/test/routines/level1/xscal.hpp
index efa0988d..95038032 100644
--- a/test/routines/level1/xscal.hpp
+++ b/test/routines/level1/xscal.hpp
@@ -66,12 +66,19 @@ class TestXscal {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Scal(args.n, args.alpha,
- buffers.x_vec(), args.x_offset, args.x_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Scal(args.n, args.alpha,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Scal(args.n, args.alpha,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
diff --git a/test/routines/level1/xswap.hpp b/test/routines/level1/xswap.hpp
index d778cc23..58310698 100644
--- a/test/routines/level1/xswap.hpp
+++ b/test/routines/level1/xswap.hpp
@@ -69,13 +69,21 @@ class TestXswap {
// Describes how to run the CLBlast routine
static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
- auto queue_plain = queue();
- auto event = cl_event{};
- auto status = Swap<T>(args.n,
- buffers.x_vec(), args.x_offset, args.x_inc,
- buffers.y_vec(), args.y_offset, args.y_inc,
- &queue_plain, &event);
- if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #ifdef OPENCL_API
+ auto queue_plain = queue();
+ auto event = cl_event{};
+ auto status = Swap<T>(args.n,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ &queue_plain, &event);
+ if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
+ #elif CUDA_API
+ auto status = Swap<T>(args.n,
+ buffers.x_vec(), args.x_offset, args.x_inc,
+ buffers.y_vec(), args.y_offset, args.y_inc,
+ queue.GetContext()(), queue.GetDevice()());
+ cuStreamSynchronize(queue());
+ #endif
return status;
}
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-12 15:54:04 +0000