--- include: rocblas_common.yaml include: known_bugs.yaml Definitions: - &size_t_index - { N: 65536, incx: 1, incy: 1 } # overflow packed index as int (n*(n+1)/2) - &size_t_incx_incy - { N: 4097, incx: 1048600, incy: 1 } - { N: 4097, incx: -1048600, incy: 1 } - { N: 4097, incx: 1, incy: 1048600 } - { N: 4097, incx: 1, incy: -1048600 } - &small_matrix_size_range - { N: 1 } - { N: 4 } - { N: 9 } - { N: 16 } - { N: 63 } - { N: 65 } - { N: 128 } - { N: 200 } - &special_case_range # Quick return - { N: 0, incx: 1, incy: 1, batch_count: 1 } - { N: 1, incx: 1, incy: 1, batch_count: 0 } # invalid_arg checks - { N: -1, incx: 1, incy: 1, batch_count: 0 } - { N: 0, incx: 0, incy: 1, batch_count: 0 } - { N: 0, incx: 1, incy: 0, batch_count: 0 } - { N: 0, incx: 1, incy: 1, batch_count: -1 } - &medium_matrix_size_range - { N: 400 } - { N: 500 } - &large_matrix_size_range - { N: 1000 } - { N: 2000 } - { N: 4011 } - &incx_incy_range - { incx: 1, incy: 1 } - { incx: -1, incy: 3 } - { incx: 3, incy: -1 } - { incx: 10, incy: 100 } - &incx_incy_range_small - { incx: 2, incy: -2 } - &alpha_beta_range - { alpha: 2.0, beta: 0.0, alphai: 1.5, betai: 0.5 } - { alpha: -1.0, beta: -1.0, alphai: 0.5, betai: 1.5 } - { alpha: 2.0, beta: 1.0, alphai: -1.5, betai: 0.5 } - { alpha: 0.0, beta: 1.0, alphai: -0.5, betai: 0 } - &alpha_beta_range_small - { alpha: 2.0, beta: 1.5, alphai: 0.5, betai: -1.5 } Tests: - name: hpmv_bad_arg category: quick function: - hpmv_bad_arg - hpmv_batched_bad_arg - hpmv_strided_batched_bad_arg precision: *single_double_precisions_complex uplo: N api: [ C, FORTRAN, FORTRAN_64 ] - name: hpmv_arg_check category: quick function: - hpmv - hpmv_batched - hpmv_strided_batched precision: *single_double_precisions_complex uplo: L matrix_size: *special_case_range # non-batched - name: hpmv_NaN category: pre_checkin function: hpmv precision: *single_double_precisions_complex uplo: [ U, L ] N: [ 129 ] incx_incy: *incx_incy_range_small alpha: [ 2.0, .NaN ] # NaN is converted to 0.0 in test code beta: [ 1.0, 0.5, .NaN ] - name: hpmv_fortran category: quick function: hpmv precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small api: FORTRAN - name: hpmv_small category: quick function: hpmv precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range - name: hpmv_medium category: pre_checkin function: hpmv precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range - name: hpmv_large category: nightly function: hpmv precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *large_matrix_size_range pointer_mode_device: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small # batched - name: hpmv_batched_NaN category: pre_checkin function: hpmv_batched precision: *single_double_precisions_complex uplo: [ L ] N: [ 129 ] incx_incy: *incx_incy_range_small alpha: [ 2.0, .NaN ] # NaN is converted to 0.0 in test code beta: [ 0.5, 1.0, .NaN ] batch_count: [ 3 ] - name: hpmv_batched_fortran category: quick function: hpmv_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small batch_count: [ -1, 0, 3 ] api: FORTRAN - name: hpmv_batched_small category: quick function: hpmv_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range batch_count: [ 3 ] - name: hpmv_batched_medium category: pre_checkin function: hpmv_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range batch_count: [ 3 ] - name: hpmv_batched_large category: nightly function: hpmv_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *large_matrix_size_range pointer_mode_device: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small batch_count: [ 3 ] - name: hpmv_batched_grid_z_16bit category: stress arguments: - { N: 3, incx: 1, incy: 1, batch_count: *c_grid_yz_require_passes} #generic kernel uplo: [ U, L ] alpha_beta: *alpha_beta_range_small pointer_mode_host: false function: - hpmv_batched: *single_precision_complex - hpmv_strided_batched: *single_precision_complex stride_scale: 1 # hpmv_strided_batched - name: hpmv_strided_batched_NaN category: pre_checkin function: hpmv_strided_batched precision: *single_double_precisions_complex uplo: [ U ] N: [ 129 ] incx_incy: *incx_incy_range_small alpha: [ 2.0, .NaN ] # NaN is converted to 0.0 in test code alphai: [ 0.5 ] beta: [ 0.5, 1.0, .NaN ] batch_count: [ 2 ] stride_scale: [ 1.5 ] - name: hpmv_strided_batched_fortran category: quick function: hpmv_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small batch_count: [ -1, 0, 3 ] stride_scale: [ 2 ] api: FORTRAN - name: hpmv_strided_batched_small category: quick function: hpmv_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range batch_count: [ 3 ] stride_scale: [ 2 ] - name: hpmv_strided_batched_medium category: pre_checkin function: hpmv_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_beta_range batch_count: [ 3 ] stride_scale: [ 1 ] - name: hpmv_strided_batched_large category: nightly function: hpmv_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *large_matrix_size_range pointer_mode_host: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_beta_range_small batch_count: [ 3 ] stride_scale: [ 1 ] - name: hpmv_graph_test category: pre_checkin function: - hpmv - hpmv_batched - hpmv_strided_batched precision: *single_double_precisions_complex matrix_size: - { N : 33} uplo: [ U ] incx: [ 1 ] alpha: [ 2.0 ] stride_scale: [ 1 ] batch_count: [ 5 ] graph_test: true - name: hpmv_size_t category: stress function: hpmv precision: *single_precision_complex matrix_size: *size_t_index pointer_mode_host: false uplo: [ U, L ] alpha_beta: *alpha_beta_range_small os_flags: LINUX gpu_arch: ['90a','942','950'] - name: hpmv_size_t_incx_incy category: stress function: hpmv precision: *single_precision_complex matrix_size: *size_t_incx_incy pointer_mode_device: false uplo: [ U, L ] alpha_beta: *alpha_beta_range_small os_flags: LINUX gpu_arch: ['90a','942','950'] - name: hpmv_ILP64 category: stress arguments: - { N: 2048, incx: 2, incy: -1, batch_count: 2 } - { N: 2, incx: 2147483649, incy: 1, batch_count: 1 } - { N: 2, incx: 1, incy: 2147483649, batch_count: 1 } - { N: 2, incx: 1, incy: 1, batch_count: *c_grid_yz_require_passes } uplo: [ U, L ] stride_scale: [ 1 ] alpha_beta: *alpha_beta_range_small pointer_mode_host: false api: [ C_64 ] os_flags: [ LINUX ] function: - hpmv: *single_precision_complex - hpmv_batched: *single_precision_complex - hpmv_strided_batched: *single_precision_complex - name: hpmv_repeatability_check category: stress function: - hpmv - hpmv_batched - hpmv_strided_batched precision: *single_double_precisions_complex matrix_size: - { N : 33} uplo: [ U ] incx: [ 1 ] alpha: [ 2.0 ] stride_scale: [ 1 ] batch_count: [ 2 ] initialization: hpl pointer_mode_host: false atomics_mode: 0 iters: 5 devices: 4 repeatability_check: true ...