summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--include/clblast.h7
-rw-r--r--include/clblast_c.h18
-rw-r--r--scripts/generator/generator.py1
-rw-r--r--scripts/generator/routine.py2
-rw-r--r--src/clblast.cc25
-rw-r--r--src/clblast_c.cc42
-rw-r--r--test/correctness/routines/level1/xasum.cc28
-rw-r--r--test/performance/routines/level1/xasum.cc35
-rw-r--r--test/wrapper_cblas.h26
-rw-r--r--test/wrapper_clblas.h68
10 files changed, 251 insertions, 1 deletions
diff --git a/include/clblast.h b/include/clblast.h
index 431f2510..cf55a256 100644
--- a/include/clblast.h
+++ b/include/clblast.h
@@ -181,6 +181,13 @@ StatusCode Nrm2(const size_t n,
const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
cl_command_queue* queue, cl_event* event = nullptr);
+// Absolute sum of values in a vector: SASUM/DASUM/ScASUM/DzASUM
+template <typename T>
+StatusCode Asum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event = nullptr);
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
diff --git a/include/clblast_c.h b/include/clblast_c.h
index f72cff3a..190e6a46 100644
--- a/include/clblast_c.h
+++ b/include/clblast_c.h
@@ -278,6 +278,24 @@ StatusCode PUBLIC_API CLBlastDznrm2(const size_t n,
const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
cl_command_queue* queue, cl_event* event);
+// Absolute sum of values in a vector: SASUM/DASUM/ScASUM/DzASUM
+StatusCode PUBLIC_API CLBlastSasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event);
+StatusCode PUBLIC_API CLBlastDasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event);
+StatusCode PUBLIC_API CLBlastScasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event);
+StatusCode PUBLIC_API CLBlastDzasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event);
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
diff --git a/scripts/generator/generator.py b/scripts/generator/generator.py
index 8cd35f95..e9e9276b 100644
--- a/scripts/generator/generator.py
+++ b/scripts/generator/generator.py
@@ -67,6 +67,7 @@ routines = [
Routine(True, "1", "dotu", T, [C,Z], ["n"], [], ["x","y"], ["dot"], [], "n", "Dot product of two complex vectors"),
Routine(True, "1", "dotc", T, [C,Z], ["n"], [], ["x","y"], ["dot"], [], "n", "Dot product of two complex vectors, one conjugated"),
Routine(True, "1", "nrm2", T, [S,D,Sc,Dz],["n"], [], ["x"], ["nrm2"], [], "2*n", "Euclidian norm of a vector"),
+ Routine(False, "1", "asum", T, [S,D,Sc,Dz],["n"], [], ["x"], ["asum"], [], "n", "Absolute sum of values in a vector"),
],
[ # Level 2: matrix-vector
Routine(True, "2a", "gemv", T, [S,D,C,Z], ["m","n"], ["layout","a_transpose"], ["a","x"], ["y"], ["alpha","beta"], "", "General matrix-vector multiplication"),
diff --git a/scripts/generator/routine.py b/scripts/generator/routine.py
index fffa19f6..7ddd7a12 100644
--- a/scripts/generator/routine.py
+++ b/scripts/generator/routine.py
@@ -72,7 +72,7 @@ class Routine():
# List of scalar buffers
def ScalarBuffersFirst(self):
- return ["dot","nrm2"]
+ return ["dot","nrm2","asum"]
def ScalarBuffersSecond(self):
return ["sa","sb","sc","ss","sd1","sd2","sx1","sy1","sparam"]
diff --git a/src/clblast.cc b/src/clblast.cc
index 4f4b6078..4888faed 100644
--- a/src/clblast.cc
+++ b/src/clblast.cc
@@ -396,6 +396,31 @@ template StatusCode PUBLIC_API Nrm2<double2>(const size_t,
const cl_mem, const size_t, const size_t,
cl_command_queue*, cl_event*);
+// Absolute sum of values in a vector: SASUM/DASUM/ScASUM/DzASUM
+template <typename T>
+StatusCode Asum(const size_t,
+ cl_mem, const size_t,
+ const cl_mem, const size_t, const size_t,
+ cl_command_queue*, cl_event*) {
+ return StatusCode::kNotImplemented;
+}
+template StatusCode PUBLIC_API Asum<float>(const size_t,
+ cl_mem, const size_t,
+ const cl_mem, const size_t, const size_t,
+ cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Asum<double>(const size_t,
+ cl_mem, const size_t,
+ const cl_mem, const size_t, const size_t,
+ cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Asum<float2>(const size_t,
+ cl_mem, const size_t,
+ const cl_mem, const size_t, const size_t,
+ cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Asum<double2>(const size_t,
+ cl_mem, const size_t,
+ const cl_mem, const size_t, const size_t,
+ cl_command_queue*, cl_event*);
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
diff --git a/src/clblast_c.cc b/src/clblast_c.cc
index 23e97bd5..c36edbca 100644
--- a/src/clblast_c.cc
+++ b/src/clblast_c.cc
@@ -433,6 +433,48 @@ StatusCode CLBlastDznrm2(const size_t n,
return static_cast<StatusCode>(status);
}
+// ASUM
+StatusCode CLBlastSasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event) {
+ auto status = clblast::Asum<float>(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, x_inc,
+ queue, event);
+ return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastDasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event) {
+ auto status = clblast::Asum<double>(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, x_inc,
+ queue, event);
+ return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastScasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event) {
+ auto status = clblast::Asum<float2>(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, x_inc,
+ queue, event);
+ return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastDzasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_command_queue* queue, cl_event* event) {
+ auto status = clblast::Asum<double2>(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, x_inc,
+ queue, event);
+ return static_cast<StatusCode>(status);
+}
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
diff --git a/test/correctness/routines/level1/xasum.cc b/test/correctness/routines/level1/xasum.cc
new file mode 100644
index 00000000..5ec20596
--- /dev/null
+++ b/test/correctness/routines/level1/xasum.cc
@@ -0,0 +1,28 @@
+
+// =================================================================================================
+// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This
+// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max-
+// width of 100 characters per line.
+//
+// Author(s):
+// Cedric Nugteren <www.cedricnugteren.nl>
+//
+// =================================================================================================
+
+#include "correctness/testblas.h"
+#include "routines/level1/xasum.h"
+
+// Shortcuts to the clblast namespace
+using float2 = clblast::float2;
+using double2 = clblast::double2;
+
+// Main function (not within the clblast namespace)
+int main(int argc, char *argv[]) {
+ clblast::RunTests<clblast::TestXasum<float>, float, float>(argc, argv, false, "SASUM");
+ clblast::RunTests<clblast::TestXasum<double>, double, double>(argc, argv, true, "DASUM");
+ clblast::RunTests<clblast::TestXasum<float2>, float2, float2>(argc, argv, true, "ScASUM");
+ clblast::RunTests<clblast::TestXasum<double2>, double2, double2>(argc, argv, true, "DzASUM");
+ return 0;
+}
+
+// =================================================================================================
diff --git a/test/performance/routines/level1/xasum.cc b/test/performance/routines/level1/xasum.cc
new file mode 100644
index 00000000..2680966e
--- /dev/null
+++ b/test/performance/routines/level1/xasum.cc
@@ -0,0 +1,35 @@
+
+// =================================================================================================
+// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This
+// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max-
+// width of 100 characters per line.
+//
+// Author(s):
+// Cedric Nugteren <www.cedricnugteren.nl>
+//
+// =================================================================================================
+
+#include "performance/client.h"
+#include "routines/level1/xasum.h"
+
+// Shortcuts to the clblast namespace
+using float2 = clblast::float2;
+using double2 = clblast::double2;
+
+// Main function (not within the clblast namespace)
+int main(int argc, char *argv[]) {
+ switch(clblast::GetPrecision(argc, argv, clblast::Precision::kSingle)) {
+ case clblast::Precision::kHalf: throw std::runtime_error("Unsupported precision mode");
+ case clblast::Precision::kSingle:
+ clblast::RunClient<clblast::TestXasum<float>, float, float>(argc, argv); break;
+ case clblast::Precision::kDouble:
+ clblast::RunClient<clblast::TestXasum<double>, double, double>(argc, argv); break;
+ case clblast::Precision::kComplexSingle:
+ clblast::RunClient<clblast::TestXasum<float2>, float2, float2>(argc, argv); break;
+ case clblast::Precision::kComplexDouble:
+ clblast::RunClient<clblast::TestXasum<double2>, double2, double2>(argc, argv); break;
+ }
+ return 0;
+}
+
+// =================================================================================================
diff --git a/test/wrapper_cblas.h b/test/wrapper_cblas.h
index dec272b0..af0eec9b 100644
--- a/test/wrapper_cblas.h
+++ b/test/wrapper_cblas.h
@@ -345,6 +345,32 @@ void cblasXnrm2(const size_t n,
reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
}
+// Forwards the Netlib BLAS calls for SASUM/DASUM/ScASUM/DzASUM
+void cblasXasum(const size_t n,
+ std::vector<float>& asum_buffer, const size_t asum_offset,
+ const std::vector<float>& x_buffer, const size_t x_offset, const size_t x_inc) {
+ asum_buffer[asum_offset] = cblas_sasum(n,
+ &x_buffer[x_offset], static_cast<int>(x_inc));
+}
+void cblasXasum(const size_t n,
+ std::vector<double>& asum_buffer, const size_t asum_offset,
+ const std::vector<double>& x_buffer, const size_t x_offset, const size_t x_inc) {
+ asum_buffer[asum_offset] = cblas_dasum(n,
+ &x_buffer[x_offset], static_cast<int>(x_inc));
+}
+void cblasXasum(const size_t n,
+ std::vector<float2>& asum_buffer, const size_t asum_offset,
+ const std::vector<float2>& x_buffer, const size_t x_offset, const size_t x_inc) {
+ asum_buffer[asum_offset] = cblas_scasum(n,
+ reinterpret_cast<const float*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
+}
+void cblasXasum(const size_t n,
+ std::vector<double2>& asum_buffer, const size_t asum_offset,
+ const std::vector<double2>& x_buffer, const size_t x_offset, const size_t x_inc) {
+ asum_buffer[asum_offset] = cblas_dzasum(n,
+ reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
+}
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
diff --git a/test/wrapper_clblas.h b/test/wrapper_clblas.h
index 89b708b8..09b3310b 100644
--- a/test/wrapper_clblas.h
+++ b/test/wrapper_clblas.h
@@ -558,6 +558,74 @@ clblasStatus clblasXnrm2<double2>(const size_t n,
num_queues, queues, num_wait_events, wait_events, events);
}
+// Forwards the clBLAS calls for SASUM/DASUM/ScASUM/DzASUM
+template <typename T>
+clblasStatus clblasXasum(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_uint num_queues, cl_command_queue *queues,
+ cl_uint num_wait_events, const cl_event *wait_events, cl_event *events);
+template <>
+clblasStatus clblasXasum<float>(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_uint num_queues, cl_command_queue *queues,
+ cl_uint num_wait_events, const cl_event *wait_events, cl_event *events) {
+ auto queue = Queue(queues[0]);
+ auto context = queue.GetContext();
+ auto scratch_buffer = Buffer<float>(context, n);
+ return clblasSasum(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, static_cast<int>(x_inc),
+ scratch_buffer(),
+ num_queues, queues, num_wait_events, wait_events, events);
+}
+template <>
+clblasStatus clblasXasum<double>(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_uint num_queues, cl_command_queue *queues,
+ cl_uint num_wait_events, const cl_event *wait_events, cl_event *events) {
+ auto queue = Queue(queues[0]);
+ auto context = queue.GetContext();
+ auto scratch_buffer = Buffer<double>(context, n);
+ return clblasDasum(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, static_cast<int>(x_inc),
+ scratch_buffer(),
+ num_queues, queues, num_wait_events, wait_events, events);
+}
+template <>
+clblasStatus clblasXasum<float2>(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_uint num_queues, cl_command_queue *queues,
+ cl_uint num_wait_events, const cl_event *wait_events, cl_event *events) {
+ auto queue = Queue(queues[0]);
+ auto context = queue.GetContext();
+ auto scratch_buffer = Buffer<float2>(context, n);
+ return clblasScasum(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, static_cast<int>(x_inc),
+ scratch_buffer(),
+ num_queues, queues, num_wait_events, wait_events, events);
+}
+template <>
+clblasStatus clblasXasum<double2>(const size_t n,
+ cl_mem asum_buffer, const size_t asum_offset,
+ const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+ cl_uint num_queues, cl_command_queue *queues,
+ cl_uint num_wait_events, const cl_event *wait_events, cl_event *events) {
+ auto queue = Queue(queues[0]);
+ auto context = queue.GetContext();
+ auto scratch_buffer = Buffer<double2>(context, n);
+ return clblasDzasum(n,
+ asum_buffer, asum_offset,
+ x_buffer, x_offset, static_cast<int>(x_inc),
+ scratch_buffer(),
+ num_queues, queues, num_wait_events, wait_events, events);
+}
+
// =================================================================================================
// BLAS level-2 (matrix-vector) routines
// =================================================================================================
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-30 23:06:40 +0000