summaryrefslogtreecommitdiff
path: root/test/routines/level1
diff options
context:
space:
mode:
Diffstat (limited to 'test/routines/level1')
-rw-r--r--test/routines/level1/xamax.hpp20
-rw-r--r--test/routines/level1/xasum.hpp20
-rw-r--r--test/routines/level1/xaxpy.hpp20
-rw-r--r--test/routines/level1/xcopy.hpp20
-rw-r--r--test/routines/level1/xdot.hpp26
-rw-r--r--test/routines/level1/xdotc.hpp26
-rw-r--r--test/routines/level1/xdotu.hpp26
-rw-r--r--test/routines/level1/xnrm2.hpp20
-rw-r--r--test/routines/level1/xscal.hpp14
-rw-r--r--test/routines/level1/xswap.hpp22
10 files changed, 107 insertions, 107 deletions
diff --git a/test/routines/level1/xamax.hpp b/test/routines/level1/xamax.hpp
index a22f681f..faffff33 100644
--- a/test/routines/level1/xamax.hpp
+++ b/test/routines/level1/xamax.hpp
@@ -74,12 +74,12 @@ class TestXamax {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].scalar(), args.imax_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -87,12 +87,12 @@ class TestXamax {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXamax<T>(args.n,
- buffers.scalar, args.imax_offset,
- buffers.x_vec, args.x_offset, args.x_inc,
+ buffers[0].scalar, args.imax_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -101,15 +101,15 @@ class TestXamax {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
cblasXamax(args.n,
scalar_cpu, args.imax_offset,
x_vec_cpu, args.x_offset, args.x_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xasum.hpp b/test/routines/level1/xasum.hpp
index 64377189..fb2c9f1a 100644
--- a/test/routines/level1/xasum.hpp
+++ b/test/routines/level1/xasum.hpp
@@ -74,12 +74,12 @@ class TestXasum {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
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,
+ buffers[0].scalar(), args.asum_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -87,12 +87,12 @@ class TestXasum {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXasum<T>(args.n,
- buffers.scalar, args.asum_offset,
- buffers.x_vec, args.x_offset, args.x_inc,
+ buffers[0].scalar, args.asum_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -101,15 +101,15 @@ class TestXasum {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
cblasXasum(args.n,
scalar_cpu, args.asum_offset,
x_vec_cpu, args.x_offset, args.x_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xaxpy.hpp b/test/routines/level1/xaxpy.hpp
index eba067c0..1c74f67f 100644
--- a/test/routines/level1/xaxpy.hpp
+++ b/test/routines/level1/xaxpy.hpp
@@ -75,12 +75,12 @@ class TestXaxpy {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -88,12 +88,12 @@ class TestXaxpy {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXaxpy(args.n, args.alpha,
- buffers.x_vec, args.x_offset, args.x_inc,
- buffers.y_vec, args.y_offset, args.y_inc,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -102,15 +102,15 @@ class TestXaxpy {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXaxpy(args.n, args.alpha,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.y_vec.Write(queue, args.y_size, y_vec_cpu);
+ buffers[0].y_vec.Write(queue, args.y_size, y_vec_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xcopy.hpp b/test/routines/level1/xcopy.hpp
index 753f0da5..55980f30 100644
--- a/test/routines/level1/xcopy.hpp
+++ b/test/routines/level1/xcopy.hpp
@@ -74,12 +74,12 @@ class TestXcopy {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -87,12 +87,12 @@ class TestXcopy {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXcopy<T>(args.n,
- buffers.x_vec, args.x_offset, args.x_inc,
- buffers.y_vec, args.y_offset, args.y_inc,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -101,15 +101,15 @@ class TestXcopy {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXcopy(args.n,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.y_vec.Write(queue, args.y_size, y_vec_cpu);
+ buffers[0].y_vec.Write(queue, args.y_size, y_vec_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xdot.hpp b/test/routines/level1/xdot.hpp
index 8127247d..1ea69c17 100644
--- a/test/routines/level1/xdot.hpp
+++ b/test/routines/level1/xdot.hpp
@@ -78,13 +78,13 @@ class TestXdot {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].scalar(), args.dot_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -92,13 +92,13 @@ class TestXdot {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXdot<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,
+ buffers[0].scalar, args.dot_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -107,18 +107,18 @@ class TestXdot {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXdot(args.n,
scalar_cpu, args.dot_offset,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xdotc.hpp b/test/routines/level1/xdotc.hpp
index 96d97dc4..00dcf7c2 100644
--- a/test/routines/level1/xdotc.hpp
+++ b/test/routines/level1/xdotc.hpp
@@ -78,13 +78,13 @@ class TestXdotc {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].scalar(), args.dot_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -92,13 +92,13 @@ class TestXdotc {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXdotc<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,
+ buffers[0].scalar, args.dot_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -107,18 +107,18 @@ class TestXdotc {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXdotc(args.n,
scalar_cpu, args.dot_offset,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xdotu.hpp b/test/routines/level1/xdotu.hpp
index 70c7fceb..512de985 100644
--- a/test/routines/level1/xdotu.hpp
+++ b/test/routines/level1/xdotu.hpp
@@ -78,13 +78,13 @@ class TestXdotu {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].scalar(), args.dot_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -92,13 +92,13 @@ class TestXdotu {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXdotu<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,
+ buffers[0].scalar, args.dot_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -107,18 +107,18 @@ class TestXdotu {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXdotu(args.n,
scalar_cpu, args.dot_offset,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xnrm2.hpp b/test/routines/level1/xnrm2.hpp
index ce33fe59..20f75226 100644
--- a/test/routines/level1/xnrm2.hpp
+++ b/test/routines/level1/xnrm2.hpp
@@ -74,12 +74,12 @@ class TestXnrm2 {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].scalar(), args.nrm2_offset,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -87,12 +87,12 @@ class TestXnrm2 {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXnrm2<T>(args.n,
- buffers.scalar, args.nrm2_offset,
- buffers.x_vec, args.x_offset, args.x_inc,
+ buffers[0].scalar, args.nrm2_offset,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -101,15 +101,15 @@ class TestXnrm2 {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> scalar_cpu(args.scalar_size, static_cast<T>(0));
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
- buffers.scalar.Read(queue, args.scalar_size, scalar_cpu);
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].scalar.Read(queue, args.scalar_size, scalar_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
cblasXnrm2(args.n,
scalar_cpu, args.nrm2_offset,
x_vec_cpu, args.x_offset, args.x_inc);
- buffers.scalar.Write(queue, args.scalar_size, scalar_cpu);
+ buffers[0].scalar.Write(queue, args.scalar_size, scalar_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xscal.hpp b/test/routines/level1/xscal.hpp
index d89688b4..e2600834 100644
--- a/test/routines/level1/xscal.hpp
+++ b/test/routines/level1/xscal.hpp
@@ -71,11 +71,11 @@ class TestXscal {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -83,11 +83,11 @@ class TestXscal {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXscal(args.n, args.alpha,
- buffers.x_vec, args.x_offset, args.x_inc,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -96,12 +96,12 @@ class TestXscal {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
cblasXscal(args.n, args.alpha,
x_vec_cpu, args.x_offset, args.x_inc);
- buffers.x_vec.Write(queue, args.x_size, x_vec_cpu);
+ buffers[0].x_vec.Write(queue, args.x_size, x_vec_cpu);
return StatusCode::kSuccess;
}
#endif
diff --git a/test/routines/level1/xswap.hpp b/test/routines/level1/xswap.hpp
index 49b0d3d0..b9f06eb7 100644
--- a/test/routines/level1/xswap.hpp
+++ b/test/routines/level1/xswap.hpp
@@ -74,12 +74,12 @@ class TestXswap {
std::vector<T>&, std::vector<T>&) {} // N/A for this routine
// Describes how to run the CLBlast routine
- static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunRoutine(const Arguments<T> &args, std::vector<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,
+ buffers[0].x_vec(), args.x_offset, args.x_inc,
+ buffers[0].y_vec(), args.y_offset, args.y_inc,
&queue_plain, &event);
if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
return status;
@@ -87,12 +87,12 @@ class TestXswap {
// Describes how to run the clBLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CLBLAS
- static StatusCode RunReference1(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference1(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
auto queue_plain = queue();
auto event = cl_event{};
auto status = clblasXswap<T>(args.n,
- buffers.x_vec, args.x_offset, args.x_inc,
- buffers.y_vec, args.y_offset, args.y_inc,
+ buffers[0].x_vec, args.x_offset, args.x_inc,
+ buffers[0].y_vec, args.y_offset, args.y_inc,
1, &queue_plain, 0, nullptr, &event);
clWaitForEvents(1, &event);
return static_cast<StatusCode>(status);
@@ -101,16 +101,16 @@ class TestXswap {
// Describes how to run the CPU BLAS routine (for correctness/performance comparison)
#ifdef CLBLAST_REF_CBLAS
- static StatusCode RunReference2(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
+ static StatusCode RunReference2(const Arguments<T> &args, std::vector<Buffers<T>> &buffers, Queue &queue) {
std::vector<T> x_vec_cpu(args.x_size, static_cast<T>(0));
std::vector<T> y_vec_cpu(args.y_size, static_cast<T>(0));
- buffers.x_vec.Read(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Read(queue, args.y_size, y_vec_cpu);
+ buffers[0].x_vec.Read(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Read(queue, args.y_size, y_vec_cpu);
cblasXswap(args.n,
x_vec_cpu, args.x_offset, args.x_inc,
y_vec_cpu, args.y_offset, args.y_inc);
- buffers.x_vec.Write(queue, args.x_size, x_vec_cpu);
- buffers.y_vec.Write(queue, args.y_size, y_vec_cpu);
+ buffers[0].x_vec.Write(queue, args.x_size, x_vec_cpu);
+ buffers[0].y_vec.Write(queue, args.y_size, y_vec_cpu);
return StatusCode::kSuccess;
}
#endif
generated by cgit v1.2.3 (git 2.39.1) at 2024-08-10 00:44:43 +0000