---
include: rocblas_common.yaml
include: known_bugs.yaml

Definitions:
  - &small_matrix_size_range
    - { N:    -1, lda:    1, stride_a:       1 }
    - { N:     1, lda:    0, stride_a:       1 }
    - { N:     0, lda:    1, stride_a:       1 }
    - { N:     1, lda:    -1, stride_a:       1 }
    - { N:    10, lda:   10, stride_a:       100 }
    - { N:    1, lda:   -1, stride_a:       1 }
    - { N:    10, lda:    2, stride_a:      20 }
    - { N:   100, lda:  200, stride_a:    40000 }

  - &special_case_range
    - { N: 0, lda: 1, incx: 1, batch_count: 1, stride_a: 1 }
    - { N: 1, lda: 1, incx: 1, batch_count: 0, stride_a: 1 }

    - { N: -1, lda: 1, incx: 1, batch_count:  0, stride_a: 1 }
    - { N:  0, lda: 0, incx: 1, batch_count:  0, stride_a: 1 }
    - { N:  2, lda: 1, incx: 1, batch_count:  0, stride_a: 1 }
    - { N:  0, lda: 1, incx: 0, batch_count:  0, stride_a: 1 }
    - { N:  0, lda: 1, incx: 1, batch_count: -1, stride_a: 1 }

  - &medium_matrix_size_range
    - { N:   300, lda:  400, stride_a:   160000 }
    - { N:   600, lda:  601, stride_a:   400000 }

  - &large_matrix_size_range
    - { N:  1000, lda: 1000, stride_a:  1000000 }
    - { N:  2000, lda: 2000, stride_a:  4000000 }
    - { N:  4011, lda: 4011, stride_a: 16088200 }
    - { N:  8000, lda: 8000, stride_a: 64000000 }

  - &size_t_index
    - { N:  46341, lda: 46341, stride_a:  2147488281, stride_x: 46341, incx: 1 }

  - &size_t_incx
    - { N: 3, lda: 3, incx: *c_pos_x2_overflow_int32 }
    - { N: 3, lda: 3, incx: *c_neg_x2_overflow_int32 }

  - &incx_range
    - [2, -2, 1, -1, 3, 10]

  - &incx_range_small
    - [ 2 ]


Tests:
- name: trmv_bad_arg
  category: quick
  function: trmv_bad_arg
  precision: *single_double_precisions
  uplo: L
  transA: N
  diag: N
  api: [ C, FORTRAN, FORTRAN_64 ]

- name: trmv_arg_check
  category: quick
  function:
  - trmv: *single_double_precisions
  - trmv_batched: *single_double_precisions
  - trmv_strided_batched: *single_double_precisions
  uplo: L
  transA: N
  diag: N
  matrix_size: *special_case_range

- name: trmv_size_t_lda
  category: stress
  function: trmv
  precision: *single_precision
  uplo: [U, L]
  transA: [N, T]
  diag: [N, U]
  matrix_size: *size_t_index
  os_flags: LINUX
  gpu_arch: ['90a','942','950']

- name: trmv_size_t_inc
  category: stress
  function: trmv
  precision: *single_precision
  uplo: [U, L]
  transA: [N, T]
  diag: [N, U]
  matrix_size: *size_t_incx
  os_flags: LINUX
  gpu_arch: ['90a','942','950']

- name: trmv_fortran
  category: quick
  function: trmv
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range_small
  api: FORTRAN

- name: trmv_small
  category: quick
  function: trmv
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range

- name: trmv_medium
  category: pre_checkin
  function: trmv
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *medium_matrix_size_range
  incx: *incx_range

- name: trmv_large
  category: nightly
  function: trmv
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [  N, T, C ]
  diag: [N, U]
  matrix_size: *large_matrix_size_range
  incx: *incx_range_small

- name: trmv_ILP64
  category: stress
  arguments:
    - { N: 2, lda: 2, incx: 2147483649, batch_count: 1 }
    - { N: 2, lda: 2, incx: -2147483649, batch_count: 1 }
    - { N: 2, lda: 2147483649, incx: 1, batch_count: 1 }
    - { N: 2, stride_a: 4, stride_x: 2, incx: 1, batch_count: *c_grid_yz_require_passes }
  uplo: [ L ]
  transA: [  N, T ]
  diag: [ N ]
  api: [ C_64 ]
  os_flags: [ LINUX ]
  function:
    - trmv: *single_precision
    - trmv_batched: *single_precision
    - trmv_strided_batched: *single_precision

- name: trmv_batched_bad_arg
  category: quick
  function: trmv_batched_bad_arg
  precision: *single_double_precisions
  uplo: [L, U]
  transA: [  N, T, C ]
  diag: [N, U]
  api: [ C, FORTRAN, FORTRAN_64 ]

- name: trmv_batched_fortran
  category: quick
  function: trmv_batched
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range_small
  batch_count: [ -1, 0, 3 ]
  api: FORTRAN

- name: trmv_batched_small
  category: quick
  function: trmv_batched
  precision: *single_double_precisions
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range
  batch_count: [ -1, 0, 1, 3 ]

- name: trmv_batched_medium
  category: pre_checkin
  function: trmv_batched
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *medium_matrix_size_range
  incx: *incx_range
  batch_count: [ 3 ]

- name: trmv_batched_large
  category: nightly
  function: trmv_batched
  precision: *single_double_precisions
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *large_matrix_size_range
  incx: *incx_range_small
  batch_count: [ 3 ]

- name: trmv_batched_grid_z_16bit
  category: stress
  arguments:
    - { N:   3, lda:    3, incx: 1, batch_count: *c_grid_yz_require_passes} #generic kernel
  uplo: [ U, L ]
  transA: [N, T, C]
  diag: [N, U]
  pointer_mode_host: false
  function:
    - trmv_batched: *single_precision
    - trmv_strided_batched: *single_precision
  stride_scale: 1

- name: trmv_strided_batched_bad_arg
  category: quick
  function: trmv_strided_batched_bad_arg
  precision: *single_double_precisions
  transA: N
  uplo: [L]
  transA: [  T ]
  diag: [N]
  stride_scale: [ 1 ]
  api: [ C, FORTRAN, FORTRAN_64 ]

- name: trmv_strided_batched_fortran
  category: quick
  function: trmv_strided_batched
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range_small
  batch_count: [ -1, 0, 3 ]
  stride_scale: [ 2 ]
  api: FORTRAN

- name: trmv_strided_batched_small
  category: quick
  function: trmv_strided_batched
  precision: *single_double_precisions
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *small_matrix_size_range
  incx: *incx_range
  batch_count: [ -1, 0, 1, 3 ]
  stride_scale: [ 2 ]

- name: trmv_strided_batched_medium
  category: pre_checkin
  function: trmv_strided_batched
  precision: *single_double_precisions_complex_real
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *medium_matrix_size_range
  incx: *incx_range
  batch_count: [ 3 ]
  stride_scale: [ 1.2 ]

- name: trmv_strided_batched_large
  category: nightly
  function: trmv_strided_batched
  precision: *single_double_precisions
  uplo: [L, U]
  transA: [N, T, C]
  diag: [N, U]
  matrix_size: *large_matrix_size_range
  incx: *incx_range_small
  batch_count: [ 3 ]
  stride_scale: [ 1.2 ]

- name: trmv_graph_test
  category: pre_checkin
  function:
        - trmv
        - trmv_batched
        - trmv_strided_batched
  precision: *single_double_precisions_complex_real
  uplo: [ U ]
  transA: [N ]
  diag: [ N ]
  matrix_size:
    - { N:    10, lda:    2, stride_a:      20 }
  incx: *incx_range
  batch_count: [ 3 ]
  stride_scale: [ 1.2 ]
  graph_test: true

- name: trmv_repeatability_check
  category: stress
  function:
        - trmv
        - trmv_batched
        - trmv_strided_batched
  precision: *single_double_precisions_complex_real
  uplo: [ U ]
  transA: [N ]
  diag: [ N ]
  matrix_size:
    - { N:    10, lda:    2, stride_a:      20 }
  incx: *incx_range
  batch_count: [ 2 ]
  stride_scale: [ 1 ]
  initialization: hpl
  pointer_mode_host: false
  atomics_mode: 0
  iters: 5
  devices: 4
  repeatability_check: true
...