TheRock/rocm-libraries/projects/rocblas/clients/include/blas1/testing_axpy_batched.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 "bytes.hpp"
#include "cblas_interface.hpp"
#include "client_utility.hpp"
#include "flops.hpp"
#include "norm.hpp"
#include "rocblas.hpp"
#include "rocblas_init.hpp"
#include "rocblas_math.hpp"
#include "rocblas_random.hpp"
#include "rocblas_test.hpp"
#include "rocblas_vector.hpp"
#include "unit.hpp"

/* ============================================================================================ */
template <typename T>
void testing_axpy_batched_bad_arg(const Arguments& arg)
{
    auto rocblas_axpy_batched_fn
        = arg.api & c_API_FORTRAN ? rocblas_axpy_batched<T, true> : rocblas_axpy_batched<T, false>;
    auto rocblas_axpy_batched_fn_64 = arg.api & c_API_FORTRAN ? rocblas_axpy_batched_64<T, true>
                                                              : rocblas_axpy_batched_64<T, false>;

    for(auto pointer_mode : {rocblas_pointer_mode_host, rocblas_pointer_mode_device})
    {
        rocblas_local_handle handle{arg};
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, pointer_mode));

        int64_t N = 100, incx = 1, incy = 1, batch_count = 2;

        DEVICE_MEMCHECK(device_vector<T>, alpha_d, (1));
        DEVICE_MEMCHECK(device_vector<T>, zero_d, (1));

        const T alpha_h(1), zero_h(0);

        const T* alpha = &alpha_h;
        const T* zero  = &zero_h;

        if(pointer_mode == rocblas_pointer_mode_device)
        {
            CHECK_HIP_ERROR(hipMemcpy(alpha_d, alpha, sizeof(*alpha), hipMemcpyHostToDevice));
            alpha = alpha_d;
            CHECK_HIP_ERROR(hipMemcpy(zero_d, zero, sizeof(*zero), hipMemcpyHostToDevice));
            zero = zero_d;
        }

        // Allocate device memory
        DEVICE_MEMCHECK(device_batch_vector<T>, dx, (N, incx, batch_count));
        DEVICE_MEMCHECK(device_batch_vector<T>, dy, (N, incy, batch_count));

        DAPI_EXPECT(
            rocblas_status_invalid_handle,
            rocblas_axpy_batched_fn,
            (nullptr, N, alpha, dx.ptr_on_device(), incx, dy.ptr_on_device(), incy, batch_count));

        DAPI_EXPECT(
            rocblas_status_invalid_pointer,
            rocblas_axpy_batched_fn,
            (handle, N, nullptr, dx.ptr_on_device(), incx, dy.ptr_on_device(), incy, batch_count));

        if(pointer_mode == rocblas_pointer_mode_host)
        {
            DAPI_EXPECT(rocblas_status_invalid_pointer,
                        rocblas_axpy_batched_fn,
                        (handle, N, alpha, nullptr, incx, dy.ptr_on_device(), incy, batch_count));

            DAPI_EXPECT(rocblas_status_invalid_pointer,
                        rocblas_axpy_batched_fn,
                        (handle, N, alpha, dx.ptr_on_device(), incx, nullptr, incy, batch_count));
        }

        // When N==0, alpha, X and Y can be nullptr without error
        DAPI_EXPECT(rocblas_status_success,
                    rocblas_axpy_batched_fn,
                    (handle, 0, nullptr, nullptr, incx, nullptr, incy, batch_count));

        // When alpha==0, X and Y can be nullptr without error
        DAPI_EXPECT(rocblas_status_success,
                    rocblas_axpy_batched_fn,
                    (handle, N, zero, nullptr, incx, nullptr, incy, batch_count));

        // When batch_count==0, alpha, X and Y can be nullptr without error
        DAPI_EXPECT(rocblas_status_success,
                    rocblas_axpy_batched_fn,
                    (handle, N, nullptr, nullptr, incx, nullptr, incy, 0));
    }
}

template <typename T>
void testing_axpy_batched(const Arguments& arg)
{
    auto rocblas_axpy_batched_fn
        = arg.api & c_API_FORTRAN ? rocblas_axpy_batched<T, true> : rocblas_axpy_batched<T, false>;
    auto rocblas_axpy_batched_fn_64 = arg.api & c_API_FORTRAN ? rocblas_axpy_batched_64<T, true>
                                                              : rocblas_axpy_batched_64<T, false>;

    rocblas_local_handle handle{arg};
    int64_t              N = arg.N, incx = arg.incx, incy = arg.incy, batch_count = arg.batch_count;

    T h_alpha = arg.get_alpha<T>();

    bool    ab_striding  = arg.alpha_beta_stride;
    int64_t alpha_stride = ab_striding ? arg.stride_c : 0;

    // argument sanity check before allocating invalid memory
    if(N <= 0 || batch_count <= 0)
    {
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_host));
        DAPI_EXPECT(rocblas_status_success,
                    rocblas_axpy_batched_fn,
                    (handle, N, nullptr, nullptr, incx, nullptr, incy, batch_count));
        return;
    }

    // Naming: `h` is in CPU (host) memory(eg hx), `d` is in GPU (device) memory (eg dx).
    // Allocate host memory
    HOST_MEMCHECK(host_batch_vector<T>, hx, (N, incx, batch_count));
    HOST_MEMCHECK(host_batch_vector<T>, hy_1, (N, incy, batch_count));
    HOST_MEMCHECK(host_batch_vector<T>, hy_2, (N, incy, batch_count));
    HOST_MEMCHECK(host_batch_vector<T>, hy_gold, (N, incy, batch_count));
    HOST_MEMCHECK(host_vector<T>, halpha, (batch_count, alpha_stride));

    // Allocate device memory
    DEVICE_MEMCHECK(device_batch_vector<T>, dx, (N, incx, batch_count));
    DEVICE_MEMCHECK(device_batch_vector<T>, dy, (N, incy, batch_count));
    DEVICE_MEMCHECK(device_vector<T>, dalpha, (batch_count, alpha_stride));

    // Assign host alpha (per-batch when alpha striding is enabled).
    rocblas_init_vector_alternating_sign(halpha, h_alpha);

    // Initialize data on host memory
    rocblas_init_vector(hx, arg, rocblas_client_alpha_sets_nan, true);
    rocblas_init_vector(hy_1, arg, rocblas_client_alpha_sets_nan, false, true);

    hy_gold.copy_from(hy_1);
    hy_2.copy_from(hy_1);

    double cpu_time_used;
    double rocblas_error_1 = 0.0;
    double rocblas_error_2 = 0.0;

    if(arg.unit_check || arg.norm_check)
    {
        // Transfer host to device
        CHECK_HIP_ERROR(dx.transfer_from(hx));

        // Per-batch alpha (alpha striding) is only supported in device pointer mode.
        if(!ab_striding)
        {
            // Call routine with pointer mode on host.
            CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_host));

            // Call routine.
            CHECK_HIP_ERROR(dy.transfer_from(hy_1));

            handle.pre_test(arg);
            DAPI_CHECK(rocblas_axpy_batched_fn,
                       (handle,
                        N,
                        halpha,
                        dx.ptr_on_device(),
                        incx,
                        dy.ptr_on_device(),
                        incy,
                        batch_count));
            handle.post_test(arg);

            // Transfer from device to host.
            CHECK_HIP_ERROR(hy_1.transfer_from(dy));
        }

        // Pointer mode device.
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_device));

        // Call routine.
        CHECK_HIP_ERROR(dalpha.transfer_from(halpha));
        CHECK_HIP_ERROR(dy.transfer_from(hy_2));
        handle.pre_test(arg);
        DAPI_CHECK(
            rocblas_axpy_batched_fn,
            (handle, N, dalpha, dx.ptr_on_device(), incx, dy.ptr_on_device(), incy, batch_count));
        handle.post_test(arg);

        // Transfer from device to host.
        CHECK_HIP_ERROR(hy_2.transfer_from(dy));

        if(arg.repeatability_check)
        {
            HOST_MEMCHECK(host_batch_vector<T>, hy_copy, (N, incy, batch_count));

            // 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 in new device
                DEVICE_MEMCHECK(device_batch_vector<T>, dx_copy, (N, incx, batch_count));
                DEVICE_MEMCHECK(device_batch_vector<T>, dy_copy, (N, incy, batch_count));
                DEVICE_MEMCHECK(device_vector<T>, dalpha_copy, (batch_count, alpha_stride));

                CHECK_HIP_ERROR(dx_copy.transfer_from(hx));
                CHECK_HIP_ERROR(dalpha_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(dy_copy.transfer_from(hy_gold));
                    DAPI_CHECK(rocblas_axpy_batched_fn,
                               (handle_copy,
                                N,
                                dalpha_copy,
                                dx_copy.ptr_on_device(),
                                incx,
                                dy_copy.ptr_on_device(),
                                incy,
                                batch_count));
                    CHECK_HIP_ERROR(hy_copy.transfer_from(dy_copy));
                    unit_check_general<T>(1, N, incy, hy_2, hy_copy, batch_count);
                }
            }
            return;
        }

        // CPU BLAS
        {
            cpu_time_used = get_time_us_no_sync();

            // Compute the host solution (per-batch alpha when alpha striding is enabled).
            for(int64_t batch_index = 0; batch_index < batch_count; ++batch_index)
            {
                ref_axpy<T>(N,
                            halpha[batch_index * alpha_stride],
                            hx[batch_index],
                            incx,
                            hy_gold[batch_index],
                            incy);
            }
            cpu_time_used = get_time_us_no_sync() - cpu_time_used;
        }

        //
        // Compare with with hsolution.
        //
        if(arg.unit_check)
        {
            if(!ab_striding)
                unit_check_general<T>(1, N, incy, hy_gold, hy_1, batch_count);

            unit_check_general<T>(1, N, incy, hy_gold, hy_2, batch_count);
        }

        if(arg.norm_check)
        {
            if(!ab_striding)
                rocblas_error_1
                    = norm_check_general<T>('I', 1, N, incy, hy_gold, hy_1, batch_count);
            rocblas_error_2 = norm_check_general<T>('I', 1, N, incy, hy_gold, hy_2, batch_count);
        }
    }

    if(arg.timing)
    {
        double gpu_time_used;
        int    number_cold_calls = arg.cold_iters;
        int    total_calls       = number_cold_calls + arg.iters;

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

        // Transfer from host to device.
        CHECK_HIP_ERROR(dy.transfer_from(hy_gold));

        const T* alpha = &h_alpha;
        if(ab_striding)
        {
            CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_device));
            CHECK_HIP_ERROR(dalpha.transfer_from(halpha));
            alpha = dalpha;
            handle.pre_test(arg);
        }

        for(int iter = 0; iter < total_calls; iter++)
        {
            if(iter == number_cold_calls)
                gpu_time_used = get_time_us_sync(stream);

            DAPI_DISPATCH(rocblas_axpy_batched_fn,
                          (handle,
                           N,
                           alpha,
                           dx.ptr_on_device(),
                           incx,
                           dy.ptr_on_device(),
                           incy,
                           batch_count));
        }
        gpu_time_used = get_time_us_sync(stream) - gpu_time_used;

        if(ab_striding)
            handle.post_test(arg);

        ArgumentModel<e_N, e_alpha, e_incx, e_incy, e_batch_count>{}.log_args<T>(
            rocblas_cout,
            arg,
            gpu_time_used,
            axpy_gflop_count<T>(N),
            axpy_gbyte_count<T>(N),
            cpu_time_used,
            rocblas_error_1,
            rocblas_error_2);
    }
}