Upstream staging 2024 03 22 (#467)

* Refactor threefry tests. The threefry tests for 2x32 and 4x32 were essentially identical, and same happened with tests for 2x64 and 4x54. So I implemented a similar approach as in the sobol tests: - Added common headers (test_rocrand_threefryNx32_20_prng.hpp and test_rocrand_threefryNx32_20_prng.hpp) in which the typed tests are registered. - Removed all the duplicated code from the old tests, and add the correspondent instantiation of the gerator API tess. - The engine API tests remain unmodified because that refactor is out of the scope of the current issue. * Refactor lfsr113, mrg, mt19937, mtgp32, philox, sobol and xorwow tests. Implemented the "typed test suite" approach: - Changed the old implementation to use typed tests. - Homogeneized code structure and variables names. - Engine API tests remain unmodified because that refactor is out of the scope of the current issue. * Added lfsr113 generator tests to match the other prng generators' - init_test - uniform_float_test - normal_double_test - log_normal_float/double_test - continuity tests * Added mrg generator tests to match the other prng generators' - init_test - normal_double_test - log_normal_float/double_test - different_seed_uint4_test * Added mt19937 generator tests to match the other prng generators' - init_test - uniform_double - normal_double_test - log_normal_float/double_test - different_seed_uint4_test * Added mgp32 generator tests to match the other prng generators' - init_test - uniform_double_test - normal_double_test - log_normal_float/double_test - different_seed_uint4_test - continuity tets * Added philox generator tests to match the other prng generators' - init_test - uniform_double_test - normal_float/double_test - log_normal_float/double_test - poisson_test - different_seed_uint4_test * Added sobol generator tests for the functionality left to be tested - init_test (generator.init()) - log_normal_float/double_test (generator.generate_log_normal()) * Added threefry generator tests to match the other prng generators' - init_test - uniform_double_test - normal_float/double_test - log_normal_float/double_test - poisson_test * Added xorwow generator tests to match the other prng generators' - uniform_double_test - normal_float/double_test - log_normal_float/double_test - poisson_test - different_seed_uint4_test * Move out common functionality to header file test_rocrand_prng * mrg mad_u64_u32_test not typed * Use ordering in templated generator tests * Separate engine-only tests from generator ones for mt19937 * Add different_seed_test to common generator test suite and keep lfsr113's particular implementation * Disable lfsr113 continuity tests * Add different_seed_test to common generator test suite and keep lfsr113's particular implementation * Don't run MT19937 with dynamic ordering * Remove unnecessary includes from internal hpp headers Deprecate src/rng/distribution/distributions.hpp Deprecated src/rng/device_engines.hpp Rename and unify the existing ROCRAND_DETAIL_.*_BM_NOT_IN_STATE to ROCRAND_DETAIL_BM_NOT_IN_STATE Directly include some headers that were previously only included transitively * Added functions for float2 and double2, with tests for float2 * Clang format * Added mtgp32 tests for double and double2 * Clang format * Division of work in test with constant to prevent deadlock in cuda * MTGP32 tests cleanup and added new MTGP32 functions to CHANGELOG * Re-enable continuity tests for lfsr113 * Re-tuned lfsr113 for gfx908 * Re-tuned lfsr113 for gfx1030 * Re-tuned lfsr113 for gfx906 * Add comment for mtgp32 continuity tests * Add offsets test to threefyNx64 generators * Fix test types for threefryNx64_20 * Fix incorrect output type in sobol64 test for poisson distribution and doxygen docs for (scrambled) sobol64 * clarify and add check for dynamic ordering on host * add api function to create block host generator * add mtgp32 host generator * Removed FQUALIFIERS and MT_FQUALIFIERS * Made rocrand, rocrand_mtgp32_block_copy and rocrand_mtgp32_set_params device function for mtgp32_engine * Removed FQUALIFIER added in merge * Made the function calls for mtpg32 in discrete, log_normal, normal and uniform only device * Add generator of precomputed state matrices for LFSR113 * Add common header with utility functions for matrix exponentiation * Implement skipahead for LFSR113 * Add engine API tests for LFSR113's discard()/discard(offset) functions * Implement offset for LFSR113 * Add generator API offsets test for LFSR113 * Update CHANGELOG * Fix python tests * Add comments to utils_matrix_exponentiation functions and merge equivalent implementations * Fix comments from LFSR113's precomputed matrices generation * Removed unneded methods and extra semicolon * Default initialize to prevent breaking change * Fix typo in LFSR113's precomputed state matrices generation * Use hip_runtime's uint4 * Remove arguments passed by value and global mutable variables * Enable host generator test for LFSR113 * Cmake files to build all non kernel tests with cpp * Add g++ wrapper package test to CI * Removed unused Windows code from cmake in cpp_wrapper * add dynamic ordering for gfx1102 * fix changelog * Move internal symbols to namespace rocrand_impl * Updated Doxyfile * Updated changelog --------- Co-authored-by: Beatriz Navidad Vilches <[email protected]> Co-authored-by: Matthias Knorr <[email protected]> Co-authored-by: Nick Breed <[email protected]> Co-authored-by: Nol Moonen <[email protected]> Co-authored-by: Lőrinc Serfőző <[email protected]> Co-authored-by: Nara Prasetya <[email protected]>
ROCm · Apr 12, 2024 · 02b2669 · 02b2669
1 parent 0f63134
commit 02b2669
Show file tree

Hide file tree

Showing 143 changed files with 10,062 additions and 7,192 deletions.
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -482,6 +482,24 @@ test:nvcc-package:
   needs:
     - build:nvcc-cmake-minimum
 
+test:cpp-wrapper:
+  stage: test
+  extends:
+    - .rocm:cmake-minimum
+    - .rules:test
+  needs:
+    - build:rocm-cmake-minimum
+  script:
+    - cd $CI_PROJECT_DIR/build
+    - $SUDO_CMD dpkg -i rocrand*.deb
+    - cmake
+      -S $CI_PROJECT_DIR/test/cpp_wrapper/
+      -B $CI_PROJECT_DIR/build_cpp_wrapper_test
+    - cmake --build $CI_PROJECT_DIR/build_cpp_wrapper_test
+    - cd $CI_PROJECT_DIR/build_cpp_wrapper_test
+    - ctest --output-on-failure
+    - $SUDO_CMD dpkg -r rocrand-benchmarks rocrand-clients rocrand-dev rocrand-tests rocrand
+
 .test:install:
   stage: test
   extends:

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -32,13 +32,24 @@ Documentation for rocRAND is available at
 * C++ wrapper:
   * `lfsr113_engine` now also supports being constructed with a seed of type `unsigned long long`, not only `uint4`.
   * added optional order parameter to constructor of `mt19937_engine`
+* Added the following functions for the `ROCRAND_RNG_PSEUDO_MTGP32` generator:
+  * `rocrand_normal2`
+  * `rocrand_normal_double2`
+  * `rocrand_log_normal2`
+  * `rocrand_log_normal_double2`
+* Added `rocrand_create_generator_host_blocking` which dispatches without stream semantics.
+* Added host-side generator for `ROCRAND_RNG_PSEUDO_MTGP32`.
+* Added offset and skipahead functionality to LFSR113 generator.
+* Added dynamic ordering for architecture `gfx1102`.
 
 ### Changes
 
 * Building rocRAND now requires a C++17 capable compiler, as the internal library sources now require it. However consuming rocRAND is still possible from C++11 as public headers don't make use of the new features.
 * Building rocRAND should be faster on machines with multiple CPU cores as the library has been
   split to multiple compilation units.
 * C++ wrapper: the `min()` and `max()` member functions of the generators and distributions are now `static constexpr`.
+* Rename and unify the existing ROCRAND_DETAIL_.*_BM_NOT_IN_STATE to ROCRAND_DETAIL_BM_NOT_IN_STATE
+* Static & dynamic library: moved all internal symbols to namespaces to avoid potential symbol name collisions when linking.
 
 ### Deprecations
 
@@ -47,12 +58,17 @@ Documentation for rocRAND is available at
   * `rocrand_device::threefry2x64_20_engine::threefry2x64_20_state`
   * `rocrand_device::threefry4x32_20_engine::threefry4x32_20_state`
   * `rocrand_device::threefry4x64_20_engine::threefry4x64_20_state`
+* Deprecated internal header: src/rng/distribution/distributions.hpp
+* Deprecated internal header: src/rng/device_engines.hpp
 
 ### Removals
 
 * Removed references to and workarounds for deprecated hcc.
 * Support for HIP-CPU
 
+### Known issues
+- SOBOL64 and SCRAMBLED_SOBOL64 generate poisson-distributed `unsigned long long int` numbers instead of `unsigned int`. This will be fixed in the next major release.
+
 ## (Unreleased) rocRAND-3.0.0 for ROCm 6.0.0
 
 ### Additions

diff --git a/benchmark/tuning/benchmark_tuning.cpp b/benchmark/tuning/benchmark_tuning.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -51,32 +51,28 @@ int main(int argc, char** argv)
     add_common_benchmark_rocrand_info();
 
     std::vector<benchmark::internal::Benchmark*> benchmarks;
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_lfsr113_template>(
-        benchmarks,
-        config);
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_mrg31k3p_template>(
-        benchmarks,
-        config);
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_mrg32k3a_template>(
-        benchmarks,
-        config);
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_mt19937_template>(
-        benchmarks,
-        config);
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_mtgp32_template>(
+    benchmark_tuning::add_all_benchmarks_for_generator<
+        benchmark_tuning::lfsr113_generator_template>(benchmarks, config);
+    benchmark_tuning::add_all_benchmarks_for_generator<
+        benchmark_tuning::mrg31k3p_generator_template>(benchmarks, config);
+    benchmark_tuning::add_all_benchmarks_for_generator<
+        benchmark_tuning::mrg32k3a_generator_template>(benchmarks, config);
+    benchmark_tuning::add_all_benchmarks_for_generator<
+        benchmark_tuning::mt19937_generator_template>(benchmarks, config);
+    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::mtgp32_generator_template>(
         benchmarks,
         config);
     benchmark_tuning::add_all_benchmarks_for_generator<
-        benchmark_tuning::rocrand_philox4x32_10_template>(benchmarks, config);
+        benchmark_tuning::philox4x32_10_generator_template>(benchmarks, config);
     benchmark_tuning::add_all_benchmarks_for_generator<
-        benchmark_tuning::rocrand_threefry2x32_20_template>(benchmarks, config);
+        benchmark_tuning::threefry2x32_20_generator_template>(benchmarks, config);
     benchmark_tuning::add_all_benchmarks_for_generator<
-        benchmark_tuning::rocrand_threefry2x64_20_template>(benchmarks, config);
+        benchmark_tuning::threefry2x64_20_generator_template>(benchmarks, config);
     benchmark_tuning::add_all_benchmarks_for_generator<
-        benchmark_tuning::rocrand_threefry4x32_20_template>(benchmarks, config);
+        benchmark_tuning::threefry4x32_20_generator_template>(benchmarks, config);
     benchmark_tuning::add_all_benchmarks_for_generator<
-        benchmark_tuning::rocrand_threefry4x64_20_template>(benchmarks, config);
-    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::rocrand_xorwow_template>(
+        benchmark_tuning::threefry4x64_20_generator_template>(benchmarks, config);
+    benchmark_tuning::add_all_benchmarks_for_generator<benchmark_tuning::xorwow_generator_template>(
         benchmarks,
         config);
 

diff --git a/benchmark/tuning/benchmark_tuning.hpp b/benchmark/tuning/benchmark_tuning.hpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -56,7 +56,7 @@ struct distribution_input
 template<template<class> class GeneratorTemplate>
 using distribution_input_t = typename distribution_input<GeneratorTemplate>::type;
 
-using rocrand_host::detail::generator_config;
+using rocrand_impl::host::generator_config;
 
 /// @brief Provides a way to opt out from benchmarking certain configs for certain generators and types
 /// @note See benchmarked_generators.hpp for specializations
@@ -140,37 +140,39 @@ class generator_benchmark_factory
         else if constexpr(std::is_integral_v<T>)
         {
             using uniform_distribution_t
-                = uniform_distribution<T, distribution_input_t<GeneratorTemplate>>;
+                = rocrand_impl::host::uniform_distribution<T,
+                                                           distribution_input_t<GeneratorTemplate>>;
             add_benchmarks_impl<T, uniform_distribution_t>();
 
             if constexpr(std::is_same_v<T, unsigned int>)
             {
                 // The poisson distribution is only supported for unsigned int.
-                using poisson_distribution_t
-                    = rocrand_poisson_distribution<ROCRAND_DISCRETE_METHOD_ALIAS>;
+                using poisson_distribution_t = rocrand_impl::host::poisson_distribution<
+                    rocrand_impl::host::DISCRETE_METHOD_ALIAS>;
                 add_benchmarks_impl<T, poisson_distribution_t>();
             }
         }
         else if constexpr(std::is_floating_point_v<T> || std::is_same_v<T, half>)
         {
             // float, double and half support these distributions only.
             using uniform_distribution_t
-                = uniform_distribution<T, distribution_input_t<GeneratorTemplate>>;
+                = rocrand_impl::host::uniform_distribution<T,
+                                                           distribution_input_t<GeneratorTemplate>>;
             add_benchmarks_impl<T, uniform_distribution_t>();
 
             constexpr rocrand_rng_type rng_type
-                = rocrand_host::detail::gen_template_type_v<GeneratorTemplate>;
+                = rocrand_impl::host::gen_template_type_v<GeneratorTemplate>;
 
-            using normal_distribution_t
-                = normal_distribution<T,
-                                      distribution_input_t<GeneratorTemplate>,
-                                      normal_distribution_max_input_width<rng_type, T>>;
+            using normal_distribution_t = rocrand_impl::host::normal_distribution<
+                T,
+                distribution_input_t<GeneratorTemplate>,
+                rocrand_impl::host::normal_distribution_max_input_width<rng_type, T>>;
             add_benchmarks_impl<T, normal_distribution_t>();
 
-            using log_normal_distribution_t
-                = log_normal_distribution<T,
-                                          distribution_input_t<GeneratorTemplate>,
-                                          log_normal_distribution_max_input_width<rng_type, T>>;
+            using log_normal_distribution_t = rocrand_impl::host::log_normal_distribution<
+                T,
+                distribution_input_t<GeneratorTemplate>,
+                rocrand_impl::host::log_normal_distribution_max_input_width<rng_type, T>>;
             add_benchmarks_impl<T, log_normal_distribution_t>();
         }
     }
@@ -183,7 +185,7 @@ class generator_benchmark_factory
     // The elements of the arrays can be controlled with CMake cache variables
     // BENCHMARK_TUNING_THREAD_OPTIONS and BENCHMARK_TUNING_BLOCK_OPTIONS
     static constexpr inline auto s_param_combinations
-        = cpp_utils::numeric_combinations(thread_options, block_options);
+        = rocrand_impl::cpp_utils::numeric_combinations(thread_options, block_options);
 
     template<class Distribution, class StaticConfigProvider>
     static std::string get_benchmark_name()
@@ -220,8 +222,7 @@ class generator_benchmark_factory
                  if constexpr(grid_size < min_benchmarked_grid_size)
                      return;
 
-                 using ConfigProvider
-                     = rocrand_host::detail::static_config_provider<threads, blocks>;
+                 using ConfigProvider = rocrand_impl::host::static_config_provider<threads, blocks>;
 
                  if constexpr(config_filter<GeneratorTemplate, T>::is_enabled(
                                   ConfigProvider::static_config))

diff --git a/benchmark/tuning/benchmark_tuning_lfsr113.cpp b/benchmark/tuning/benchmark_tuning_lfsr113.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_lfsr113_template>(
+template void add_all_benchmarks_for_generator<lfsr113_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_mrg31k3p.cpp b/benchmark/tuning/benchmark_tuning_mrg31k3p.cpp
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_mrg31k3p_template>(
+template void add_all_benchmarks_for_generator<mrg31k3p_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_mrg32k3a.cpp b/benchmark/tuning/benchmark_tuning_mrg32k3a.cpp
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_mrg32k3a_template>(
+template void add_all_benchmarks_for_generator<mrg32k3a_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_mt19937.cpp b/benchmark/tuning/benchmark_tuning_mt19937.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_mt19937_template>(
+template void add_all_benchmarks_for_generator<mt19937_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_mtgp32.cpp b/benchmark/tuning/benchmark_tuning_mtgp32.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_mtgp32_template>(
+template void add_all_benchmarks_for_generator<mtgp32_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_philox.cpp b/benchmark/tuning/benchmark_tuning_philox.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_philox4x32_10_template>(
+template void add_all_benchmarks_for_generator<philox4x32_10_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_threefry2x32_20.cpp b/benchmark/tuning/benchmark_tuning_threefry2x32_20.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_threefry2x32_20_template>(
+template void add_all_benchmarks_for_generator<threefry2x32_20_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_threefry2x64_20.cpp b/benchmark/tuning/benchmark_tuning_threefry2x64_20.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_threefry2x64_20_template>(
+template void add_all_benchmarks_for_generator<threefry2x64_20_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_threefry4x32_20.cpp b/benchmark/tuning/benchmark_tuning_threefry4x32_20.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_threefry4x32_20_template>(
+template void add_all_benchmarks_for_generator<threefry4x32_20_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_threefry4x64_20.cpp b/benchmark/tuning/benchmark_tuning_threefry4x64_20.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_threefry4x64_20_template>(
+template void add_all_benchmarks_for_generator<threefry4x64_20_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning
diff --git a/benchmark/tuning/benchmark_tuning_xorwow.cpp b/benchmark/tuning/benchmark_tuning_xorwow.cpp
@@ -1,4 +1,4 @@
-// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2023-2024 Advanced Micro Devices, Inc. All rights reserved.
 //
 // Permission is hereby granted, free of charge, to any person obtaining a copy
 // of this software and associated documentation files (the "Software"), to deal
@@ -24,7 +24,7 @@
 namespace benchmark_tuning
 {
 
-template void add_all_benchmarks_for_generator<rocrand_xorwow_template>(
+template void add_all_benchmarks_for_generator<xorwow_generator_template>(
     std::vector<benchmark::internal::Benchmark*>& benchmarks, const benchmark_config& config);
 
 } // namespace benchmark_tuning