CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
773 lines
47 KiB
YAML
773 lines
47 KiB
YAML
---
|
|
include: rocblas_common.yaml
|
|
include: known_bugs.yaml
|
|
|
|
Definitions:
|
|
|
|
# using stride_x,y,d to test out offsets of rocblas_internal_gemm_interface()
|
|
- &gemm_internal_matrix_size
|
|
- { M: 511, N: 511, K: 511, lda: 511, ldb: 511, ldc: 511, ldd: 511, stride_a: 0, stride_b: 261121, stride_c: 261121, stride_x: 4294967296, stride_y: 4294967297, stride_d: 4294967298 }
|
|
|
|
- &small_matrix_size_range
|
|
- { M: 0, N: 8, K: 8, lda: 8, ldb: 8, ldc: 8, ldd: 8, stride_a: 64, stride_b: 64, stride_c: 64, stride_d: 64 } # m==0
|
|
- { M: 8, N: 0, K: 8, lda: 8, ldb: 8, ldc: 8, ldd: 8, stride_a: 64, stride_b: 64, stride_c: 64, stride_d: 64 } # n==0
|
|
- { M: 8, N: 8, K: 0, lda: 8, ldb: 8, ldc: 8, ldd: 8, stride_a: 64, stride_b: 64, stride_c: 64, stride_d: 64 } # k==0
|
|
- { M: -1, N: -1, K: -1, lda: -1, ldb: 1, ldc: 1, ldd: 1, stride_a: 1, stride_b: 1, stride_c: 1, stride_d: 1 }
|
|
- { M: 8, N: 9, K: 10, lda: 8, ldb: 10, ldc: 8, ldd: 8, stride_a: 80, stride_b: 90, stride_c: 82, stride_d: 82 }
|
|
- { M: 4, N: 3, K: 4, lda: 4, ldb: 4, ldc: 4, ldd: 4, stride_a: 16, stride_b: 12, stride_c: 12, stride_d: 12 }
|
|
- { M: 3, N: 3, K: 3, lda: 3, ldb: 3, ldc: 3, ldd: 3, stride_a: 0, stride_b: 9, stride_c: 9, stride_d: 9 }
|
|
- { M: 3, N: 3, K: 3, lda: 3, ldb: 3, ldc: 3, ldd: 3, stride_a: 9, stride_b: 0, stride_c: 9, stride_d: 9 }
|
|
- { M: 3, N: 3, K: 3, lda: 3, ldb: 3, ldc: 3, ldd: 3, stride_a: 9, stride_b: 9, stride_c: 9, stride_d: 9 }
|
|
- { M: 4, N: 3, K: 4, lda: 4, ldb: 4, ldc: 4, ldd: 4, stride_a: 16, stride_b: 16, stride_c: 16, stride_d: 16 }
|
|
- { M: 4, N: 4, K: 4, lda: 4, ldb: 4, ldc: 5, ldd: 5, stride_a: 16, stride_b: 16, stride_c: 20, stride_d: 20 }
|
|
- { M: 4, N: 4, K: 4, lda: 4, ldb: 4, ldc: 5, ldd: 5, stride_a: 17, stride_b: 17, stride_c: 20, stride_d: 20 }
|
|
- { M: 4, N: 4, K: 4, lda: 4, ldb: 4, ldc: 5, ldd: 5, stride_a: 17, stride_b: 16, stride_c: 20, stride_d: 20 }
|
|
- { M: 4, N: 4, K: 4, lda: 4, ldb: 4, ldc: 5, ldd: 5, stride_a: 16, stride_b: 17, stride_c: 20, stride_d: 20 }
|
|
- { M: 8, N: 8, K: 8, lda: 8, ldb: 8, ldc: 8, ldd: 8, stride_a: 64, stride_b: 64, stride_c: 64, stride_d: 64 }
|
|
- { M: 8, N: 9, K: 10, lda: 8, ldb: 10, ldc: 8, ldd: 8, stride_a: 100, stride_b: 100, stride_c: 100, stride_d: 100 }
|
|
- { M: 15, N: 15, K: 15, lda: 15, ldb: 15, ldc: 15, ldd: 15, stride_a: 225, stride_b: 225, stride_c: 225, stride_d: 225 }
|
|
- { M: 16, N: 16, K: 16, lda: 16, ldb: 16, ldc: 16, ldd: 16, stride_a: 256, stride_b: 256, stride_c: 256, stride_d: 256 }
|
|
- { M: 17, N: 17, K: 17, lda: 17, ldb: 17, ldc: 17, ldd: 17, stride_a: 289, stride_b: 289, stride_c: 289, stride_d: 289 }
|
|
- { M: 31, N: 33, K: 35, lda: 101, ldb: 102, ldc: 103, ldd: 103, stride_a: 3605, stride_b: 3605, stride_c: 3605, stride_d: 3605 }
|
|
- { M: 59, N: 61, K: 63, lda: 129, ldb: 131, ldc: 137, ldd: 137, stride_a: 8631, stride_b: 8631, stride_c: 8631, stride_d: 8631 }
|
|
- { M: 15, N: 15, K: 15, lda: 15, ldb: 15, ldc: 15, ldd: 15, stride_a: 225, stride_b: 225, stride_c: 225, stride_d: 225 }
|
|
- { M: 16, N: 16, K: 16, lda: 16, ldb: 16, ldc: 16, ldd: 16, stride_a: 256, stride_b: 256, stride_c: 256, stride_d: 256 }
|
|
- { M: 17, N: 17, K: 17, lda: 17, ldb: 17, ldc: 17, ldd: 17, stride_a: 289, stride_b: 289, stride_c: 289, stride_d: 289 }
|
|
- { M: 63, N: 63, K: 63, lda: 63, ldb: 63, ldc: 63, ldd: 63, stride_a: 3969, stride_b: 3969, stride_c: 3969, stride_d: 3969 }
|
|
- { M: 64, N: 64, K: 64, lda: 64, ldb: 64, ldc: 64, ldd: 64, stride_a: 4096, stride_b: 4096, stride_c: 4096, stride_d: 4096 }
|
|
- { M: 65, N: 65, K: 65, lda: 65, ldb: 65, ldc: 65, ldd: 65, stride_a: 4225, stride_b: 4225, stride_c: 4225, stride_d: 4225 }
|
|
- { M: 127, N: 127, K: 127, lda: 127, ldb: 127, ldc: 127, ldd: 127, stride_a: 16129, stride_b: 16129, stride_c: 16129, stride_d: 16129 }
|
|
- { M: 128, N: 128, K: 128, lda: 128, ldb: 128, ldc: 128, ldd: 128, stride_a: 16384, stride_b: 16384, stride_c: 16384, stride_d: 16384 }
|
|
- { M: 129, N: 129, K: 129, lda: 129, ldb: 129, ldc: 129, ldd: 129, stride_a: 16641, stride_b: 16641, stride_c: 16641, stride_d: 16641 }
|
|
|
|
- &small_matrix_size_stride_a_range
|
|
- { M: 3, N: 3, K: 3, lda: 3, ldb: 3, ldc: 3, ldd: 3, stride_a: 9, stride_b: 9, stride_c: 9, stride_d: 9 }
|
|
- { M: 15, N: 15, K: 15, lda: 15, ldb: 15, ldc: 15, ldd: 15, stride_a: 225, stride_b: 0, stride_c: 225, stride_d: 225 }
|
|
- { M: 16, N: 16, K: 16, lda: 16, ldb: 16, ldc: 16, ldd: 16, stride_a: 0, stride_b: 256, stride_c: 256, stride_d: 256 }
|
|
- { M: 17, N: 17, K: 17, lda: 17, ldb: 17, ldc: 17, ldd: 17, stride_a: 289, stride_b: 0, stride_c: 289, stride_d: 289 }
|
|
- { M: 63, N: 63, K: 63, lda: 63, ldb: 63, ldc: 63, ldd: 63, stride_a: 0, stride_b: 3969, stride_c: 3969, stride_d: 3969 }
|
|
- { M: 64, N: 64, K: 64, lda: 64, ldb: 64, ldc: 64, ldd: 64, stride_a: 4096, stride_b: 0, stride_c: 4096, stride_d: 4096 }
|
|
- { M: 65, N: 65, K: 65, lda: 65, ldb: 65, ldc: 65, ldd: 65, stride_a: 0, stride_b: 4225, stride_c: 4225, stride_d: 4225 }
|
|
- { M: 127, N: 127, K: 127, lda: 127, ldb: 127, ldc: 127, ldd: 127, stride_a: 16129, stride_b: 0, stride_c: 16129, stride_d: 16129 }
|
|
- { M: 128, N: 128, K: 128, lda: 128, ldb: 128, ldc: 128, ldd: 128, stride_a: 0, stride_b: 16384, stride_c: 16384, stride_d: 16384 }
|
|
- { M: 129, N: 129, K: 129, lda: 129, ldb: 129, ldc: 129, ldd: 129, stride_a: 16641, stride_b: 0, stride_c: 16641, stride_d: 16641 }
|
|
|
|
- &medium_matrix_size_range
|
|
- { M: 129, N: 130, K: 131, lda: 132, ldb: 133, ldc: 134, ldd: 134, stride_a: 17554, stride_b: 17554, stride_c: 17554, stride_d: 17554 }
|
|
- { M: 255, N: 255, K: 255, lda: 255, ldb: 255, ldc: 255, ldd: 255, stride_a: 65025, stride_b: 65025, stride_c: 65025, stride_d: 65025 }
|
|
- { M: 256, N: 256, K: 256, lda: 256, ldb: 256, ldc: 256, ldd: 256, stride_a: 65536, stride_b: 65536, stride_c: 65536, stride_d: 65536 }
|
|
- { M: 257, N: 257, K: 257, lda: 257, ldb: 257, ldc: 257, ldd: 257, stride_a: 66049, stride_b: 66049, stride_c: 66049, stride_d: 66049 }
|
|
- { M: 501, N: 502, K: 103, lda: 504, ldb: 605, ldc: 506, ldd: 506, stride_a: 340010, stride_b: 340010, stride_c: 340010, stride_d: 340010 }
|
|
|
|
- &medium_matrix_size_stride_a_range
|
|
- { M: 255, N: 255, K: 255, lda: 255, ldb: 255, ldc: 255, ldd: 255, stride_a: 65025, stride_b: 0, stride_c: 65025, stride_d: 65025 }
|
|
- { M: 256, N: 256, K: 256, lda: 256, ldb: 256, ldc: 256, ldd: 256, stride_a: 0, stride_b: 65536, stride_c: 65536, stride_d: 65536 }
|
|
- { M: 257, N: 257, K: 257, lda: 257, ldb: 257, ldc: 257, ldd: 257, stride_a: 66049, stride_b: 0, stride_c: 66049, stride_d: 66049 }
|
|
|
|
- &large_matrix_size_range
|
|
- { M: 511, N: 511, K: 511, lda: 511, ldb: 511, ldc: 511, ldd: 511, stride_a: 261121, stride_b: 261121, stride_c: 261121, stride_d: 261121 }
|
|
- { M: 512, N: 512, K: 512, lda: 512, ldb: 512, ldc: 512, ldd: 512, stride_a: 262144, stride_b: 262144, stride_c: 262144, stride_d: 262144 }
|
|
- { M: 513, N: 513, K: 513, lda: 513, ldb: 513, ldc: 513, ldd: 513, stride_a: 263169, stride_b: 263169, stride_c: 263169, stride_d: 263169 }
|
|
- { M: 513, N: 514, K: 515, lda: 516, ldb: 517, ldc: 518, ldd: 518, stride_a: 266771, stride_b: 266772, stride_c: 266773, stride_d: 266773 }
|
|
|
|
- &large_matrix_size_stride_a_range
|
|
- { M: 511, N: 511, K: 511, lda: 511, ldb: 511, ldc: 511, ldd: 511, stride_a: 0, stride_b: 261121, stride_c: 261121, stride_d: 261121 }
|
|
- { M: 512, N: 512, K: 512, lda: 512, ldb: 512, ldc: 512, ldd: 512, stride_a: 262144, stride_b: 0, stride_c: 262144, stride_d: 262144 }
|
|
- { M: 513, N: 513, K: 513, lda: 513, ldb: 513, ldc: 513, ldd: 513, stride_a: 0, stride_b: 263169, stride_c: 263169, stride_d: 263169 }
|
|
- { M: 513, N: 514, K: 515, lda: 516, ldb: 517, ldc: 518, ldd: 518, stride_a: 266771, stride_b: 0, stride_c: 266773, stride_d: 266773 }
|
|
|
|
- &deepbench_vec
|
|
- { M: 12544, N: 64, K: 64, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 802816, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 12544, N: 64, K: 64, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 802816, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 784, N: 64, K: 192, lda: 784, ldb: 192, ldc: 784, ldd: 784, stride_a: 150528, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 12544, N: 64, K: 64, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 802816, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 12544, N: 64, K: 64, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 802816, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 196, N: 1024, K: 256, lda: 196, ldb: 1024, ldc: 196, ldd: 196, stride_a: 50176, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 196, N: 1024, K: 256, lda: 196, ldb: 1024, ldc: 196, ldd: 196, stride_a: 50176, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 196, N: 256, K: 1024, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 200704, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 196, N: 256, K: 1024, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 200704, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 196, N: 256, K: 256, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 50176, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 196, N: 256, K: 256, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 50176, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 196, N: 512, K: 192, lda: 196, ldb: 512, ldc: 196, ldd: 196, stride_a: 37632, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 49, N: 2048, K: 512, lda: 49, ldb: 2048, ldc: 49, ldd: 49, stride_a: 25088, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 49, N: 2048, K: 512, lda: 49, ldb: 2048, ldc: 49, ldd: 49, stride_a: 25088, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 49, N: 512, K: 2048, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 100352, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 49, N: 512, K: 2048, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 100352, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 49, N: 512, K: 512, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 25088, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 49, N: 512, K: 512, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 25088, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 49, N: 832, K: 256, lda: 49, ldb: 832, ldc: 49, ldd: 49, stride_a: 12544, stride_b: 0, stride_c: 40768, stride_d: 40768 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 192, K: 64, lda: 784, ldb: 192, ldc: 784, ldd: 784, stride_a: 50176, stride_b: 0, stride_c: 150528, stride_d: 150528 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 3025, N: 256, K: 64, lda: 3025, ldb: 64, ldc: 3025, ldd: 3025, stride_a: 193600, stride_b: 0, stride_c: 774400, stride_d: 774400 }
|
|
|
|
- &conv_resnet50_fwd
|
|
- { M: 3025, N: 64, K: 256, lda: 3025, ldb: 256, ldc: 3025, ldd: 3025, stride_a: 774400, stride_b: 0, stride_c: 193600, stride_d: 193600 }
|
|
- { M: 3025, N: 64, K: 64, lda: 3025, ldb: 64, ldc: 3025, ldd: 3025, stride_a: 193600, stride_b: 0, stride_c: 193600, stride_d: 193600 }
|
|
- { M: 3025, N: 256, K: 64, lda: 3025, ldb: 64, ldc: 3025, ldd: 3025, stride_a: 193600, stride_b: 0, stride_c: 774400, stride_d: 774400 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 3136, N: 64, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
|
|
- &conv_resnet50_bwddata
|
|
- { M: 196, N: 1024, K: 256, lda: 196, ldb: 1024, ldc: 196, ldd: 196, stride_a: 50176, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 196, N: 256, K: 1024, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 200704, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 3025, N: 256, K: 64, lda: 3025, ldb: 256, ldc: 3025, ldd: 3025, stride_a: 193600, stride_b: 0, stride_c: 774400, stride_d: 774400 }
|
|
- { M: 3025, N: 64, K: 256, lda: 3025, ldb: 64, ldc: 3025, ldd: 3025, stride_a: 774400, stride_b: 0, stride_c: 193600, stride_d: 193600 }
|
|
- { M: 3025, N: 64, K: 64, lda: 3025, ldb: 64, ldc: 3025, ldd: 3025, stride_a: 193600, stride_b: 0, stride_c: 193600, stride_d: 193600 }
|
|
- { M: 3136, N: 256, K: 64, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 64, K: 256, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 3136, N: 64, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 49, N: 2048, K: 512, lda: 49, ldb: 2048, ldc: 49, ldd: 49, stride_a: 25088, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 49, N: 512, K: 2048, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 100352, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 784, N: 128, K: 512, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 512, K: 128, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
|
|
- &conv_inception4_fwd
|
|
- { M: 1225, N: 192, K: 384, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 470400, stride_b: 0, stride_c: 235200, stride_d: 235200 }
|
|
- { M: 1225, N: 64, K: 384, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 470400, stride_b: 0, stride_c: 78400, stride_d: 78400 }
|
|
- { M: 1225, N: 96, K: 384, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 470400, stride_b: 0, stride_c: 117600, stride_d: 117600 }
|
|
- { M: 289, N: 128, K: 1024, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 295936, stride_b: 0, stride_c: 36992, stride_d: 36992 }
|
|
- { M: 289, N: 192, K: 1024, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 295936, stride_b: 0, stride_c: 55488, stride_d: 55488 }
|
|
- { M: 289, N: 256, K: 1024, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 295936, stride_b: 0, stride_c: 73984, stride_d: 73984 }
|
|
- { M: 289, N: 384, K: 1024, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 295936, stride_b: 0, stride_c: 110976, stride_d: 110976 }
|
|
- { M: 5329, N: 64, K: 160, lda: 5329, ldb: 160, ldc: 5329, ldd: 5329, stride_a: 852640, stride_b: 0, stride_c: 341056, stride_d: 341056 }
|
|
|
|
- &conv_inception4_bwddata
|
|
- { M: 1225, N: 384, K: 192, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 235200, stride_b: 0, stride_c: 470400, stride_d: 470400 }
|
|
- { M: 1225, N: 384, K: 64, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 78400, stride_b: 0, stride_c: 470400, stride_d: 470400 }
|
|
- { M: 1225, N: 384, K: 96, lda: 1225, ldb: 384, ldc: 1225, ldd: 1225, stride_a: 117600, stride_b: 0, stride_c: 470400, stride_d: 470400 }
|
|
- { M: 289, N: 1024, K: 128, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 36992, stride_b: 0, stride_c: 295936, stride_d: 295936 }
|
|
- { M: 289, N: 1024, K: 192, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 55488, stride_b: 0, stride_c: 295936, stride_d: 295936 }
|
|
- { M: 289, N: 1024, K: 256, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 73984, stride_b: 0, stride_c: 295936, stride_d: 295936 }
|
|
- { M: 289, N: 1024, K: 384, lda: 289, ldb: 1024, ldc: 289, ldd: 289, stride_a: 110976, stride_b: 0, stride_c: 295936, stride_d: 295936 }
|
|
- { M: 5329, N: 160, K: 64, lda: 5329, ldb: 160, ldc: 5329, ldd: 5329, stride_a: 341056, stride_b: 0, stride_c: 852640, stride_d: 852640 }
|
|
- { M: 64, N: 1536, K: 256, lda: 64, ldb: 1536, ldc: 64, ldd: 64, stride_a: 16384, stride_b: 0, stride_c: 98304, stride_d: 98304 }
|
|
- { M: 64, N: 1536, K: 384, lda: 64, ldb: 1536, ldc: 64, ldd: 64, stride_a: 24576, stride_b: 0, stride_c: 98304, stride_d: 98304 }
|
|
|
|
- &conv_ctest_bwddata
|
|
- { M: 121, N: 2048, K: 1, lda: 121, ldb: 2048, ldc: 121, ldd: 121, stride_a: 121, stride_b: 0, stride_c: 247808, stride_d: 247808 }
|
|
- { M: 12544, N: 64, K: 1, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 12544, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 144, N: 1024, K: 1, lda: 144, ldb: 1024, ldc: 144, ldd: 144, stride_a: 144, stride_b: 0, stride_c: 147456, stride_d: 147456 }
|
|
- { M: 144, N: 256, K: 1, lda: 144, ldb: 256, ldc: 144, ldd: 144, stride_a: 144, stride_b: 0, stride_c: 36864, stride_d: 36864 }
|
|
- { M: 144, N: 512, K: 1, lda: 144, ldb: 512, ldc: 144, ldd: 144, stride_a: 144, stride_b: 0, stride_c: 73728, stride_d: 73728 }
|
|
- { M: 169, N: 256, K: 1, lda: 169, ldb: 256, ldc: 169, ldd: 169, stride_a: 169, stride_b: 0, stride_c: 43264, stride_d: 43264 }
|
|
- { M: 16, N: 512, K: 1, lda: 16, ldb: 512, ldc: 16, ldd: 16, stride_a: 16, stride_b: 0, stride_c: 8192, stride_d: 8192 }
|
|
- { M: 16, N: 528, K: 1, lda: 16, ldb: 528, ldc: 16, ldd: 16, stride_a: 16, stride_b: 0, stride_c: 8448, stride_d: 8448 }
|
|
- { M: 16, N: 576, K: 1, lda: 16, ldb: 576, ldc: 16, ldd: 16, stride_a: 16, stride_b: 0, stride_c: 9216, stride_d: 9216 }
|
|
- { M: 16, N: 608, K: 1, lda: 16, ldb: 608, ldc: 16, ldd: 16, stride_a: 16, stride_b: 0, stride_c: 9728, stride_d: 9728 }
|
|
- { M: 196, N: 128, K: 1, lda: 196, ldb: 128, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 196, N: 192, K: 1, lda: 196, ldb: 192, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 37632, stride_d: 37632 }
|
|
- { M: 196, N: 256, K: 1, lda: 196, ldb: 256, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 196, N: 480, K: 1, lda: 196, ldb: 480, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 94080, stride_d: 94080 }
|
|
- { M: 196, N: 512, K: 1, lda: 196, ldb: 512, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 196, N: 528, K: 1, lda: 196, ldb: 528, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 103488, stride_d: 103488 }
|
|
- { M: 196, N: 576, K: 1, lda: 196, ldb: 576, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 112896, stride_d: 112896 }
|
|
- { M: 196, N: 608, K: 1, lda: 196, ldb: 608, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 119168, stride_d: 119168 }
|
|
- { M: 196, N: 64, K: 1, lda: 196, ldb: 64, ldc: 196, ldd: 196, stride_a: 196, stride_b: 0, stride_c: 12544, stride_d: 12544 }
|
|
- { M: 3136, N: 128, K: 1, lda: 3136, ldb: 128, ldc: 3136, ldd: 3136, stride_a: 3136, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 3136, N: 256, K: 1, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 3136, stride_b: 0, stride_c: 802816, stride_d: 802816 }
|
|
- { M: 3136, N: 64, K: 1, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 3136, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 32768, N: 480, K: 1, lda: 32768, ldb: 480, ldc: 32768, ldd: 32768, stride_a: 32768, stride_b: 0, stride_c: 15728640, stride_d: 15728640 }
|
|
- { M: 49, N: 1024, K: 1, lda: 49, ldb: 1024, ldc: 49, ldd: 49, stride_a: 49, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 49, N: 1056, K: 1, lda: 49, ldb: 1056, ldc: 49, ldd: 49, stride_a: 49, stride_b: 0, stride_c: 51744, stride_d: 51744 }
|
|
- { M: 49, N: 192, K: 1, lda: 49, ldb: 192, ldc: 49, ldd: 49, stride_a: 49, stride_b: 0, stride_c: 9408, stride_d: 9408 }
|
|
- { M: 49, N: 512, K: 1, lda: 49, ldb: 512, ldc: 49, ldd: 49, stride_a: 49, stride_b: 0, stride_c: 25088, stride_d: 25088 }
|
|
- { M: 49, N: 832, K: 1, lda: 49, ldb: 832, ldc: 49, ldd: 49, stride_a: 49, stride_b: 0, stride_c: 40768, stride_d: 40768 }
|
|
- { M: 729, N: 64, K: 1, lda: 729, ldb: 64, ldc: 729, ldd: 729, stride_a: 729, stride_b: 0, stride_c: 46656, stride_d: 46656 }
|
|
- { M: 784, N: 128, K: 1, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 100352, stride_d: 100352 }
|
|
- { M: 784, N: 192, K: 1, lda: 784, ldb: 192, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 150528, stride_d: 150528 }
|
|
- { M: 784, N: 256, K: 1, lda: 784, ldb: 256, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 200704, stride_d: 200704 }
|
|
- { M: 784, N: 320, K: 1, lda: 784, ldb: 320, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 250880, stride_d: 250880 }
|
|
- { M: 784, N: 512, K: 1, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 401408, stride_d: 401408 }
|
|
- { M: 784, N: 64, K: 1, lda: 784, ldb: 64, ldc: 784, ldd: 784, stride_a: 784, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
- { M: 8192, N: 480, K: 1, lda: 8192, ldb: 480, ldc: 8192, ldd: 8192, stride_a: 8192, stride_b: 0, stride_c: 3932160, stride_d: 3932160 }
|
|
- { M: 8192, N: 512, K: 1, lda: 8192, ldb: 512, ldc: 8192, ldd: 8192, stride_a: 8192, stride_b: 0, stride_c: 4194304, stride_d: 4194304 }
|
|
- { M: 8192, N: 528, K: 1, lda: 8192, ldb: 528, ldc: 8192, ldd: 8192, stride_a: 8192, stride_b: 0, stride_c: 4325376, stride_d: 4325376 }
|
|
- { M: 8192, N: 832, K: 1, lda: 8192, ldb: 832, ldc: 8192, ldd: 8192, stride_a: 8192, stride_b: 0, stride_c: 6815744, stride_d: 6815744 }
|
|
|
|
- &conv_ctest_fwd
|
|
- { M: 12544, N: 1, K: 64, lda: 12544, ldb: 64, ldc: 12544, ldd: 12544, stride_a: 802816, stride_b: 0, stride_c: 12544, stride_d: 12544 }
|
|
- { M: 3136, N: 1, K: 128, lda: 3136, ldb: 128, ldc: 3136, ldd: 3136, stride_a: 401408, stride_b: 0, stride_c: 3136, stride_d: 3136 }
|
|
- { M: 3136, N: 1, K: 256, lda: 3136, ldb: 256, ldc: 3136, ldd: 3136, stride_a: 802816, stride_b: 0, stride_c: 3136, stride_d: 3136 }
|
|
- { M: 3136, N: 1, K: 64, lda: 3136, ldb: 64, ldc: 3136, ldd: 3136, stride_a: 200704, stride_b: 0, stride_c: 3136, stride_d: 3136 }
|
|
- { M: 32768, N: 1, K: 480, lda: 32768, ldb: 480, ldc: 32768, ldd: 32768, stride_a: 15728640, stride_b: 0, stride_c: 32768, stride_d: 32768 }
|
|
- { M: 729, N: 1, K: 64, lda: 729, ldb: 64, ldc: 729, ldd: 729, stride_a: 46656, stride_b: 0, stride_c: 729, stride_d: 729 }
|
|
- { M: 784, N: 1, K: 128, lda: 784, ldb: 128, ldc: 784, ldd: 784, stride_a: 100352, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 784, N: 1, K: 192, lda: 784, ldb: 192, ldc: 784, ldd: 784, stride_a: 150528, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 784, N: 1, K: 256, lda: 784, ldb: 256, ldc: 784, ldd: 784, stride_a: 200704, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 784, N: 1, K: 320, lda: 784, ldb: 320, ldc: 784, ldd: 784, stride_a: 250880, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 784, N: 1, K: 512, lda: 784, ldb: 512, ldc: 784, ldd: 784, stride_a: 401408, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 784, N: 1, K: 64, lda: 784, ldb: 64, ldc: 784, ldd: 784, stride_a: 50176, stride_b: 0, stride_c: 784, stride_d: 784 }
|
|
- { M: 8192, N: 1, K: 480, lda: 8192, ldb: 480, ldc: 8192, ldd: 8192, stride_a: 3932160, stride_b: 0, stride_c: 8192, stride_d: 8192 }
|
|
- { M: 8192, N: 1, K: 512, lda: 8192, ldb: 512, ldc: 8192, ldd: 8192, stride_a: 4194304, stride_b: 0, stride_c: 8192, stride_d: 8192 }
|
|
- { M: 8192, N: 1, K: 528, lda: 8192, ldb: 528, ldc: 8192, ldd: 8192, stride_a: 4325376, stride_b: 0, stride_c: 8192, stride_d: 8192 }
|
|
- { M: 8192, N: 1, K: 832, lda: 8192, ldb: 832, ldc: 8192, ldd: 8192, stride_a: 6815744, stride_b: 0, stride_c: 8192, stride_d: 8192 }
|
|
|
|
- &transA_transB_range
|
|
- { transA: N, transB: N }
|
|
- { transA: N, transB: T }
|
|
- { transA: C, transB: N }
|
|
- { transA: T, transB: C }
|
|
|
|
- &alpha_beta_range
|
|
- { alpha: 1.0, beta: 0.0 }
|
|
- { alpha: -2.0, beta: -3.0 }
|
|
- { alpha: 0.0, beta: 1.0 }
|
|
|
|
- &complex_alpha_beta_range
|
|
- { alpha: 1, beta: 3, alphai: 3, betai: 1 }
|
|
- { alpha: -1, beta: -3, alphai: 3, betai: 1 }
|
|
|
|
- &alpha_beta_range_small
|
|
- { alpha: 2, alphai: 2, beta: -1.0, betai: 2.0 }
|
|
|
|
Tests:
|
|
- name: gemm_strided_batched_bad_arg
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched_bad_arg: *half_single_double_complex_real_precisions
|
|
transA: N
|
|
transB: N
|
|
api: [ FORTRAN, FORTRAN_64 ]
|
|
|
|
- name: gemm_strided_batched_xx_bad_arg
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched_ex_bad_arg: *real_precisions
|
|
- gemm_strided_batched_ex_bad_arg: *complex_precisions
|
|
transA: N
|
|
transB: N
|
|
api: [ C, FORTRAN ]
|
|
|
|
# Tests confirm no NaN propagation when alpha = 0, 2 and beta = 0. Value .NaN is converted into zero
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, precision: *single_precision, batch_count: 1,
|
|
function: gemm_strided_batched, transA: N, transB: N, M: 256, N: 128, K: 64, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, precision: *double_precision, batch_count: 2,
|
|
function: gemm_strided_batched, transA: N, transB: T, M: 128, N: 64, K: 256, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, precision: *single_precision_complex, batch_count: 3,
|
|
function: gemm_strided_batched, transA: T, transB: N, M: 64, N: 256, K: 128, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, precision: *double_precision_complex, batch_count: 4,
|
|
function: gemm_strided_batched, transA: T, transB: T, M: 128, N: 64, K: 256, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, function: gemm_strided_batched_ex, batch_count: 5,
|
|
a_type: f16_r, b_type: f16_r, c_type: f16_r, d_type: f16_r, compute_type: f32_r,
|
|
transA: N, transB: N, M: 255, N: 127, K: 63, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, function: gemm_strided_batched_ex, batch_count: 6,
|
|
a_type: bf16_r, b_type: bf16_r, c_type: bf16_r, d_type: bf16_r, compute_type: f32_r,
|
|
transA: N, transB: T, M: 62, N: 126, K: 254, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, function: gemm_strided_batched_ex, batch_count: 7,
|
|
a_type: f32_r, b_type: f32_r, c_type: f32_r, d_type: f32_r, compute_type: f32_r,
|
|
transA: T, transB: N, M: 253, N: 61, K: 125, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
- {name: alpha_beta_zero_NaN, category: pre_checkin, function: gemm_strided_batched_ex, batch_count: 8,
|
|
a_type: f64_r, b_type: f64_r, c_type: f64_r, d_type: f64_r, compute_type: f64_r,
|
|
transA: T, transB: T, M: 256, N: 128, K: 64, alpha: [ .NaN, 2 ], beta: [ .NaN, 2 ] }
|
|
|
|
# Split *real_precisions into *int8 and *nonint8_real_precisions. TODO: merge nonint8 and int8 test groups
|
|
|
|
- name: gemm_strided_batched_fortran
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched: *half_single_double_precisions
|
|
- gemm_strided_batched: *single_double_precisions_complex
|
|
- gemm_strided_batched_ex: *nonint8_real_precisions
|
|
- gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *small_matrix_size_range
|
|
pointer_mode_host: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range_small
|
|
batch_count: [ -1, 0, 3 ]
|
|
api: FORTRAN
|
|
|
|
- name: gemm_strided_batched_int8_fortran
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched_ex: *int8_precision
|
|
matrix_size: *small_matrix_size_range
|
|
pointer_mode_device: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range_small
|
|
batch_count: [ -1, 0, 3 ]
|
|
api: FORTRAN
|
|
|
|
- name: gemm_strided_batched_internal
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched: *single_precision
|
|
matrix_size: *gemm_internal_matrix_size
|
|
pointer_mode_host: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range_small
|
|
batch_count: [ 3 ]
|
|
api: INTERNAL
|
|
|
|
- name: gemm_strided_batched_small
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched: *half_single_double_precisions
|
|
gemm_strided_batched_ex: *nonint8_real_precisions
|
|
matrix_size: *small_matrix_size_range
|
|
pointer_mode_device: false
|
|
alpha_beta: *alpha_beta_range
|
|
transA_transB: *transA_transB_range
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_small_int8
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched_ex: *int8_precision
|
|
matrix_size: *small_matrix_size_range
|
|
pointer_mode_host: false
|
|
alpha_beta: *alpha_beta_range
|
|
transA_transB: *transA_transB_range
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_small_complex
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *small_matrix_size_range
|
|
pointer_mode_device: false
|
|
alpha_beta: *complex_alpha_beta_range
|
|
transA_transB: *transA_transB_range
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_small_stride_zero
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched: *half_single_double_precisions
|
|
gemm_strided_batched_ex: *nonint8_real_precisions
|
|
matrix_size: *small_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ 1, 3 ]
|
|
|
|
- name: gemm_strided_batched_small_stride_zero_int8
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched_ex: *int8_precision
|
|
matrix_size: *small_matrix_size_stride_a_range
|
|
pointer_mode_device: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ 1, 3 ]
|
|
|
|
- name: gemm_strided_batched_small_stride_zero_complex
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *small_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ 3 ]
|
|
|
|
- name: gemm_strided_batched_medium
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *medium_matrix_size_range
|
|
pointer_mode_device: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_medium_alt
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched_ex: *hpa_half_precision
|
|
matrix_size: *medium_matrix_size_range
|
|
pointer_mode_host: false
|
|
transA_transB: *transA_transB_range
|
|
alpha: 1
|
|
beta: 0
|
|
flags: [4, 20]
|
|
initialization: [ denorm, denorm2 ]
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_medium_complex
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *medium_matrix_size_range
|
|
pointer_mode_device: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *complex_alpha_beta_range
|
|
batch_count: [ 3 ]
|
|
|
|
- name: gemm_strided_batched_medium_stride_zero
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *medium_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: 3
|
|
|
|
- name: gemm_strided_batched_medium_stride_zero_alt
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched_ex: *hpa_half_precision
|
|
matrix_size: *medium_matrix_size_stride_a_range
|
|
pointer_mode_device: false
|
|
alpha: 1
|
|
beta: 0
|
|
flags: [4, 20]
|
|
initialization: [ denorm, denorm2 ]
|
|
transA: N
|
|
transB: N
|
|
batch_count: 3
|
|
|
|
- name: gemm_strided_batched_medium_stride_zero_complex
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *medium_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: 3
|
|
|
|
- name: gemm_strided_batched_large
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *half_single_double_precisions
|
|
gemm_strided_batched_ex: *nonint8_real_precisions
|
|
matrix_size: *large_matrix_size_range
|
|
pointer_mode_device: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range
|
|
batch_count: [ 3 ]
|
|
|
|
- name: gemm_strided_batched_large_int8
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched_ex: *int8_precision
|
|
matrix_size: *large_matrix_size_range
|
|
pointer_mode_host: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range
|
|
batch_count: [ 3 ]
|
|
|
|
- name: gemm_strided_batched_large_complex
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *large_matrix_size_range
|
|
pointer_mode_device: false
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range
|
|
batch_count: [ 3 ]
|
|
|
|
- name: gemm_strided_batched_large_stride_zero
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *half_single_double_precisions
|
|
gemm_strided_batched_ex: *nonint8_real_precisions
|
|
matrix_size: *large_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_large_stride_zero_int8
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched_ex: *int8_precision
|
|
matrix_size: *large_matrix_size_stride_a_range
|
|
pointer_mode_device: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_large_stride_zero_complex
|
|
category: pre_checkin
|
|
function:
|
|
gemm_strided_batched: *single_double_precisions_complex
|
|
gemm_strided_batched_ex: *single_double_precisions_complex
|
|
matrix_size: *large_matrix_size_stride_a_range
|
|
pointer_mode_host: false
|
|
alpha: 2.0
|
|
alphai: 1.0
|
|
beta: 3.0
|
|
betai: -1.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: [ -1, 0, 3 ]
|
|
|
|
- name: gemm_strided_batched_deepbench
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *deepbench_vec
|
|
pointer_mode_device: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: T
|
|
batch_count: [ 2 ]
|
|
|
|
- name: gemm_strided_batched_deepbench_alt
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched_ex: *hpa_half_precision
|
|
matrix_size: *deepbench_vec
|
|
pointer_mode_host: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
flags: [4, 20]
|
|
initialization: [ denorm, denorm2 ]
|
|
transA: N
|
|
transB: T
|
|
batch_count: [ 2 ]
|
|
|
|
- name: gemm_strided_batched_conv_resnet50_fwd
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_resnet50_fwd
|
|
pointer_mode_device: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: 4
|
|
|
|
- name: gemm_strided_batched_conv_resnet50_bwddata
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_resnet50_bwddata
|
|
pointer_mode_host: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: T
|
|
batch_count: 4
|
|
|
|
- name: gemm_strided_batched_conv_inception4_fwd
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_inception4_fwd
|
|
pointer_mode_device: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: 4
|
|
|
|
- name: gemm_strided_batched_conv_inception4_bwddata
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_inception4_bwddata
|
|
pointer_mode_host: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: T
|
|
batch_count: 4
|
|
|
|
- name: gemm_strided_batched_conv_ctest_bwddata
|
|
category: nightly
|
|
function:
|
|
gemm_strided_batched: *half_single_precisions
|
|
gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_ctest_bwddata
|
|
pointer_mode_host: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: T
|
|
batch_count: 1
|
|
|
|
- name: gemm_strided_batched_conv_ctest_fwd
|
|
category: nightly
|
|
function:
|
|
- gemm_strided_batched: *half_single_precisions
|
|
- gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size: *conv_ctest_fwd
|
|
pointer_mode_device: false
|
|
alpha: 1.0
|
|
beta: 0.0
|
|
transA: N
|
|
transB: N
|
|
batch_count: 1
|
|
|
|
- name: gemm_strided_batched_ex_hpa_fp16
|
|
category: quick
|
|
function:
|
|
gemm_strided_batched_ex: *hpa_half_precision
|
|
M: 128
|
|
N: 128
|
|
K: 1000
|
|
pointer_mode_host: false
|
|
transA: N
|
|
transB: N
|
|
alpha: 0.001
|
|
beta: 1
|
|
batch_count: [ 1, 10, 12 ]
|
|
lda: 128
|
|
ldb: 1000
|
|
ldc: 128
|
|
ldd: 128
|
|
stride_a: 128000
|
|
stride_b: 128000
|
|
stride_c: 16384
|
|
stride_d: 16384
|
|
|
|
- name: gemm_strided_batched_xx_zerok
|
|
category: quick
|
|
function:
|
|
- gemm_strided_batched: *half_single_double_precisions_complex_real
|
|
- gemm_strided_batched_ex: *complex_precisions
|
|
- gemm_strided_batched_ex: *real_precisions
|
|
transA_transB: *transA_transB_range
|
|
alpha_beta: *alpha_beta_range
|
|
K: 0
|
|
matrix_size:
|
|
- { M: 1, N: 2 }
|
|
- { M: 3, N: 5 }
|
|
- { M: 512, N: 100 }
|
|
- { M: 63, N: 512 }
|
|
pointer_mode_host: false
|
|
batch_count: 1
|
|
|
|
- name: gemm_ex_large_batch
|
|
category: stress
|
|
function:
|
|
- gemm_strided_batched: *single_precision
|
|
- gemm_strided_batched_ex: *nonint8_real_precisions
|
|
- gemm_strided_batched_ex: *int8_precision
|
|
matrix_size:
|
|
- { M: 3, N: 3, K: 3, lda: 3, ldb: 3, ldc: 3, ldd: 3, batch_count: *c_grid_yz_require_passes }
|
|
pointer_mode_device: false
|
|
transA: [ N, T ]
|
|
transB: [ N, T ]
|
|
alpha: 1
|
|
beta: 1
|
|
|
|
- name: gemm_strided_batched_graph_test
|
|
category: pre_checkin
|
|
function:
|
|
- gemm_strided_batched: *half_single_precisions
|
|
- gemm_strided_batched_ex: *hpa_half_single_precisions
|
|
matrix_size:
|
|
- { M: 33, N: 100, K: 32 }
|
|
pointer_mode_device: false
|
|
transA: N
|
|
transB: N
|
|
alpha_beta: *alpha_beta_range
|
|
batch_count: [ 3 ]
|
|
graph_test: true
|
|
|
|
- name: gemm_strided_batched_ILP64
|
|
category: stress
|
|
function:
|
|
- gemm_strided_batched: *half_precision
|
|
- gemm_strided_batched_ex: *nonint8_real_precisions
|
|
- gemm_strided_batched_ex: *int8_precision
|
|
arguments:
|
|
- { M: 4, N: 4, K: 4, lda: 4, ldb: 4, ldc: 4, ldd: 4 }
|
|
pointer_mode_device: false
|
|
transA: [ N ]
|
|
transB: [ N ]
|
|
batch_count: *c_grid_yz_require_passes
|
|
alpha: 1
|
|
beta: 1
|
|
api: [ C_64 ]
|
|
os_flags: LINUX
|
|
|
|
- name: gemm_strided_batched_ex_ILP64
|
|
category: pre_checkin
|
|
function:
|
|
- gemm_strided_batched_ex: *nonint8_real_precisions
|
|
- gemm_strided_batched_ex: *int8_precision
|
|
matrix_size:
|
|
- { M: 64, N: 64, K: 64, lda: 64, ldb: 64, ldc: 64, ldd: 64, batch_count: 511 }
|
|
pointer_mode_device: false
|
|
transA: [ N ]
|
|
transB: [ T ]
|
|
alpha: 1
|
|
beta: 1
|
|
api: [ C_64 ]
|
|
os_flags: LINUX
|
|
|
|
- name: gemm_strided_batched_repeatability_check
|
|
category: stress
|
|
function:
|
|
- gemm_strided_batched: *half_single_double_precisions_complex_real
|
|
- gemm_strided_batched_ex: *half_single_double_precisions_complex_real
|
|
matrix_size:
|
|
- { M: 784, N: 64, K: 192, lda: 784, ldb: 192, ldc: 784, ldd: 784, stride_a: 150528, stride_b: 0, stride_c: 50176, stride_d: 50176 }
|
|
initialization: hpl
|
|
transA: [ N, T ]
|
|
transB: [ N, T ]
|
|
alpha_beta: *alpha_beta_range
|
|
atomics_mode: 0
|
|
batch_count: [ 2 ]
|
|
iters: 5
|
|
pointer_mode_host: false
|
|
devices: 4
|
|
repeatability_check: true
|
|
...
|