--- include: rocblas_common.yaml include: known_bugs.yaml Definitions: - &size_t_n_incx_incy - { N: 2048, lda: 2097200, incx: 1, incy: 1 } - { N: 2048, lda: 2048, incx: 2097200, incy: 1 } - { N: 2048, lda: 2048, incx: 1, incy: 2097200 } - { N: 2048, lda: 2048, incx: -2097200, incy: 1 } - { N: 2048, lda: 2048, incx: 1, incy: -2097200 } - &tiny_matrix - { N: 16, lda: 16 } - &small_matrix_size_range - { N: 4, lda: 3 } # invalid lda - { N: 1, lda: 1 } - { N: 10, lda: 10 } - { N: 16, lda: 100 } - { N: 33, lda: 35 } - { N: 65, lda: 65 } - &special_case_range # Quick return - { N: 0, lda: 1, incx: 1, incy: 1, batch_count: 1 } - { N: 1, lda: 1, incx: 1, incy: 1, batch_count: 0 } # invalid_arg checks - { N: -1, lda: 1, incx: 1, incy: 1, batch_count: 0 } - { N: 0, lda: 0, incx: 1, incy: 1, batch_count: 0 } - { N: 2, lda: 1, incx: 1, incy: 1, batch_count: 0 } - { N: 0, lda: 1, incx: 0, incy: 1, batch_count: 0 } - { N: 0, lda: 1, incx: 1, incy: 0, batch_count: 0 } - { N: 0, lda: 1, incx: 1, incy: 1, batch_count: -1 } - &medium_matrix_size_range - { N: 1000, lda: 1000 } - { N: 2011, lda: 2500 } - &large_matrix_size_range - { N: 2048, lda: 2048 } - { N: 4011, lda: 4012 } - &size_t_index - { N: 46341, lda: 46341, stride_a: 2147488281, stride_x: 46341 } - &incx_incy_range - { incx: -1, incy: 2 } - { incx: 3, incy: -1 } - { incx: 2, incy: 2 } - &incx_incy_range_small - { incx: 1, incy: 1 } - &alpha_range - { alpha: -1.0, alphai: 0.5 } - { alpha: 0.0, alphai: -0.5 } - { alpha: 0.0, alphai: 0.0 } - &alpha_range_small - { alpha: 4.0, alphai: 2.0 } Tests: - name: her2_bad_arg category: quick function: - her2_bad_arg: *single_double_precisions_complex - her2_batched_bad_arg: *single_double_precisions_complex - her2_strided_batched_bad_arg: *single_double_precisions_complex api: [ C, FORTRAN, FORTRAN_64 ] - name: her2_arg_check category: quick function: - her2 - her2_batched - her2_strided_batched precision: *single_double_precisions_complex uplo: L matrix_size: *special_case_range - name: her2_size_t_M category: stress function: - her2 precision: *single_precision_complex #- her2: *single_double_precisions_complex # shared code for size_t with non-batched #- her2_batched: *single_precision_complex #- her2_strided_batched: *single_precision_complex uplo: [U, L] matrix_size: *size_t_index pointer_mode_device: false incx: 1 incy: 1 gpu_arch: ['90a','942','950'] os_flags: LINUX - name: her2_fortran category: quick function: her2 precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small api: FORTRAN - name: her2_NaN category: quick function: her2 precision: *single_double_precisions_complex uplo: [ U ] matrix_size: *tiny_matrix incx_incy: *incx_incy_range_small alpha: [ .NaN ] - name: her2_small category: quick function: her2 precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_range - name: her2_medium category: pre_checkin function: her2 precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range pointer_mode_host: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small - name: her2_large category: nightly function: her2 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_range_small - name: her2_size_t category: stress function: her2 precision: *single_precision_complex uplo: [ U ] matrix_size: *size_t_n_incx_incy pointer_mode_device: false alpha_beta: *alpha_range_small gpu_arch: ['90a','942','950'] os_flags: LINUX # batched - name: her2_batched_fortran category: quick function: her2_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small batch_count: [ -1, 0, 3 ] api: FORTRAN - name: her2_batched_NaN category: quick function: her2_batched precision: *double_precision_complex uplo: [ U ] matrix_size: *tiny_matrix incx_incy: *incx_incy_range_small alpha: [ .NaN ] batch_count: [ 2 ] - name: her2_batched_small category: quick function: her2_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_range batch_count: [ 3 ] - name: her2_batched_medium category: pre_checkin function: her2_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range pointer_mode_host: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small batch_count: [ 3 ] - name: her2_batched_large category: nightly function: her2_batched precision: *single_double_precisions_complex uplo: [ U ] matrix_size: *large_matrix_size_range pointer_mode_device: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small batch_count: [ 3 ] - name: her2_batched_grid_z_16bit category: stress arguments: - { N: 3, lda: 3, incx: 1, incy: 1, batch_count: *c_grid_yz_require_passes} #generic kernel alpha: [ 2.0 ] uplo: [ U, L ] pointer_mode_host: false function: - her2_batched: *single_precision_complex - her2_strided_batched: *single_precision_complex stride_scale: 1 # strided batched - name: her2_strided_batched_fortran category: quick function: her2_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_range_small batch_count: [ -1, 0, 3 ] stride_scale: [ 2 ] api: FORTRAN - name: her2_strided_batched_NaN category: quick function: her2_strided_batched precision: *single_precision_complex uplo: [ L ] matrix_size: *tiny_matrix incx_incy: *incx_incy_range_small alpha: [ .NaN ] batch_count: [ 2 ] stride_scale: [ 2 ] - name: her2_strided_batched_small category: quick function: her2_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *small_matrix_size_range incx_incy: *incx_incy_range alpha_beta: *alpha_range stride_scale: [ 2.5 ] batch_count: [ 3 ] - name: her2_strided_batched_medium category: pre_checkin function: her2_strided_batched precision: *single_double_precisions_complex uplo: [ U, L ] matrix_size: *medium_matrix_size_range pointer_mode_device: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small stride_scale: [ 1.0 ] batch_count: [ 3 ] - name: her2_strided_large category: nightly function: her2_strided_batched precision: *single_double_precisions_complex uplo: [ L ] matrix_size: *large_matrix_size_range pointer_mode_host: false incx_incy: *incx_incy_range_small alpha_beta: *alpha_range_small stride_scale: [ 1.0 ] batch_count: [ 1, 3 ] - name: her2_ILP64 category: stress arguments: - { N: 2, lda: 2147483649, incx: 1, incy: 1, batch_count: 1 } - { N: 2, lda: 2, incx: 2147483649, incy: 1, batch_count: 1 } - { N: 2, lda: 2, stride_a: 2, incx: 1, incy: 1, batch_count: *c_grid_yz_require_passes } uplo: [ U, L ] stride_scale: [ 1 ] alpha_beta: *alpha_range_small pointer_mode_host: false api: [ C_64 ] os_flags: [ LINUX ] function: - her2: *single_precision_complex - her2_batched: *single_precision_complex - her2_strided_batched: *single_precision_complex - name: her2_graph_test category: pre_checkin function: - her2 - her2_batched - her2_strided_batched precision: *single_double_precisions_complex matrix_size: - { N: 33, lda: 35 } uplo: [ U ] incx_incy: *incx_incy_range_small alpha: [ 2.0 ] stride_scale: [ 1 ] batch_count: [ 5 ] graph_test: true - name: her2_repeatability_check category: stress function: - her2 - her2_batched - her2_strided_batched precision: *single_double_precisions_complex matrix_size: - { N: 33, lda: 35 } uplo: [ U ] incx_incy: *incx_incy_range_small alpha: [ 2.0 ] stride_scale: [ 1 ] batch_count: [ 2 ] pointer_mode_host: false initialization: hpl atomics_mode: 0 iters: 5 devices: 4 repeatability_check: true ...