TheRock/rocm-libraries/projects/rocblas/clients/include/blas1/testing_scal.hpp

/* ************************************************************************
 * Copyright (C) 2018-2025 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
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell cop-
 * ies of the Software, and to permit persons to whom the Software is furnished
 * to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IM-
 * PLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
 * FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
 * COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
 * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNE-
 * CTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 *
 * ************************************************************************ */

#pragma once

#include "benchmark.hpp"
#include "testing_common.hpp"

#include "blas1/rocblas_scal.hpp" // internal API

template <typename T, typename U = T>
void testing_scal_bad_arg(const Arguments& arg)
{
    auto rocblas_scal_fn
        = arg.api & c_API_FORTRAN ? rocblas_scal<T, U, true> : rocblas_scal<T, U, false>;
    auto rocblas_scal_fn_64
        = arg.api & c_API_FORTRAN ? rocblas_scal_64<T, U, true> : rocblas_scal_64<T, U, false>;

    int64_t N     = 100;
    int64_t incx  = 1;
    U       alpha = (U)0.6;

    rocblas_local_handle handle{arg};

    // Allocate device memory
    DEVICE_MEMCHECK(device_vector<T>, dx, (N, incx));

    DAPI_EXPECT(rocblas_status_invalid_handle, rocblas_scal_fn, (nullptr, N, &alpha, dx, incx));
    DAPI_EXPECT(rocblas_status_invalid_pointer, rocblas_scal_fn, (handle, N, nullptr, dx, incx));
    DAPI_EXPECT(
        rocblas_status_invalid_pointer, rocblas_scal_fn, (handle, N, &alpha, nullptr, incx));
}

template <typename T, typename U = T>
void testing_scal(const Arguments& arg)
{
    auto rocblas_scal_fn
        = arg.api & c_API_FORTRAN ? rocblas_scal<T, U, true> : rocblas_scal<T, U, false>;
    auto rocblas_scal_fn_64
        = arg.api & c_API_FORTRAN ? rocblas_scal_64<T, U, true> : rocblas_scal_64<T, U, false>;

    int64_t N       = arg.N;
    int64_t incx    = arg.incx;
    U       h_alpha = arg.get_alpha<U>();
    bool    HMM     = arg.HMM;

    rocblas_local_handle handle{arg};

    // argument sanity check before allocating invalid memory
    if(N <= 0 || incx <= 0)
    {
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_host));
        DAPI_CHECK(rocblas_scal_fn, (handle, N, nullptr, nullptr, incx));
        return;
    }

    // Naming: `h` is in CPU (host) memory(eg hx), `d` is in GPU (device) memory (eg dx).
    // Allocate host memory
    HOST_MEMCHECK(host_vector<T>, hx, (N, incx));
    HOST_MEMCHECK(host_vector<T>, hx_gold, (N, incx));
    HOST_MEMCHECK(host_vector<U>, halpha, (1));
    halpha[0] = h_alpha;

    // Allocate device memory
    DEVICE_MEMCHECK(device_vector<U>, d_alpha, (1));

    // Initial Data on CPU
    rocblas_init_vector(hx, arg, rocblas_client_alpha_sets_nan, true);

    // copy vector is easy in STL; hx_gold = hx: hx_gold which will be output of CPU BLAS
    hx_gold = hx;

    double gpu_time_used, cpu_time_used;
    double rocblas_error_host   = 0.0;
    double rocblas_error_device = 0.0;
    if(arg.unit_check || arg.norm_check)
    {
        // Allocate device memory
        DEVICE_MEMCHECK(device_vector<T>, dx, (N, incx));

        // copy data from CPU to device
        CHECK_HIP_ERROR(dx.transfer_from(hx));

        if(arg.pointer_mode_host)
        {
            // GPU BLAS, rocblas_pointer_mode_host
            CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_host));

            handle.pre_test(arg);
            if(arg.api != INTERNAL)
            {
                DAPI_CHECK(rocblas_scal_fn, (handle, N, &h_alpha, dx, incx));
            }
            else
            {
                rocblas_stride offset_x = arg.lda;
                CHECK_ROCBLAS_ERROR((rocblas_internal_scal_template(
                    handle, N, &h_alpha, 0, dx + offset_x, -offset_x, incx, arg.stride_x, 1)));
            }
            handle.post_test(arg);

            // Transfer output from device to CPU
            CHECK_HIP_ERROR(hx.transfer_from(dx));
        }

        if(arg.pointer_mode_device)
        {
            // GPU BLAS, rocblas_pointer_mode_device
            CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_device));

            // copy data from CPU to device for rocblas_pointer_mode_device tests
            CHECK_HIP_ERROR(dx.transfer_from(hx_gold));
            CHECK_HIP_ERROR(d_alpha.transfer_from(halpha));

            handle.pre_test(arg);
            DAPI_CHECK(rocblas_scal_fn, (handle, N, d_alpha, dx, incx));
            handle.post_test(arg);

            if(arg.repeatability_check)
            {
                HOST_MEMCHECK(host_vector<T>, hx_copy, (N, incx));
                CHECK_HIP_ERROR(hx.transfer_from(dx));

                // multi-GPU support
                int device_id, device_count;
                CHECK_HIP_ERROR(limit_device_count(device_count, (int)arg.devices));

                for(int dev_id = 0; dev_id < device_count; dev_id++)
                {
                    CHECK_HIP_ERROR(hipGetDevice(&device_id));
                    if(device_id != dev_id)
                        CHECK_HIP_ERROR(hipSetDevice(dev_id));

                    //New rocblas handle for new device
                    rocblas_local_handle handle_copy{arg};

                    // Allocate device memory
                    DEVICE_MEMCHECK(device_vector<T>, dx_copy, (N, incx));
                    DEVICE_MEMCHECK(device_vector<U>, d_alpha_copy, (1));

                    CHECK_HIP_ERROR(d_alpha_copy.transfer_from(halpha));

                    CHECK_ROCBLAS_ERROR(
                        rocblas_set_pointer_mode(handle_copy, rocblas_pointer_mode_device));

                    for(int runs = 0; runs < arg.iters; runs++)
                    {
                        CHECK_HIP_ERROR(dx_copy.transfer_from(hx_gold));
                        DAPI_CHECK(rocblas_scal_fn, (handle_copy, N, d_alpha_copy, dx_copy, incx));
                        CHECK_HIP_ERROR(hx_copy.transfer_from(dx_copy));
                        unit_check_general<T>(1, N, incx, hx, hx_copy);
                    }
                }
                return;
            }
        }

        // CPU BLAS
        cpu_time_used = get_time_us_no_sync();
        ref_scal(N, h_alpha, (T*)hx_gold, incx);
        cpu_time_used = get_time_us_no_sync() - cpu_time_used;

        if(arg.pointer_mode_host)
        {
            if(arg.unit_check)
            {
                unit_check_general<T>(1, N, incx, hx_gold, hx);
            }

            if(arg.norm_check)
            {
                rocblas_error_host = norm_check_general<T>('F', 1, N, incx, hx_gold, hx);
            }
        }

        if(arg.pointer_mode_device)
        {
            // Transfer output from device to CPU
            CHECK_HIP_ERROR(hx.transfer_from(dx));

            if(arg.unit_check)
            {
                unit_check_general<T>(1, N, incx, hx_gold, hx);
            }

            if(arg.norm_check)
            {
                rocblas_error_device = norm_check_general<T>('F', 1, N, incx, hx_gold, hx);
            }
        }
    } // end of if unit/norm check

    if(arg.timing)
    {
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_host));
        size_t aligned_stride_x = align_stride<T>(size_t(N) * (incx >= 0 ? incx : -incx));

        size_t x_size        = incx >= 0 ? N * incx * sizeof(T) : N * (-incx) * sizeof(T);
        size_t x_cached_size = x_size;

        size_t flush_batch_count = calculate_flush_batch_count(
            arg.flush_batch_count, arg.flush_memory_size, x_cached_size);

        // allocate device rotating buffer arrays
        DEVICE_MEMCHECK(device_strided_batch_vector<T>,
                        dx_rot_buff,
                        (N, incx, aligned_stride_x, flush_batch_count, HMM));

        CHECK_HIP_ERROR(dx_rot_buff.broadcast_one_vector_from(hx));

        hipStream_t stream;
        CHECK_ROCBLAS_ERROR(rocblas_get_stream(handle, &stream));

        auto lambda_to_benchmark = [&](int flush_index) {
            DAPI_DISPATCH(rocblas_scal_fn, (handle, N, &h_alpha, dx_rot_buff[flush_index], incx));
        };

        Benchmark<decltype(lambda_to_benchmark)> benchmark_scal(
            lambda_to_benchmark, stream, arg, flush_batch_count);

        benchmark_scal.run_timer();

        ArgumentModel<e_N, e_alpha, e_incx>{}.log_args<T>(rocblas_cout,
                                                          arg,
                                                          benchmark_scal.get_hot_time(),
                                                          scal_gflop_count<T, U>(N),
                                                          scal_gbyte_count<T>(N),
                                                          cpu_time_used,
                                                          rocblas_error_host,
                                                          rocblas_error_device);
    }
}