AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Removing outdated API documentation (#2888)

Browse Source
This commit is contained in:
Saad Rahim
2022-09-26 20:24:07 -06:00
committed by GitHub
parent 3a9866730d
commit 1f6cb120fe
5 changed files with 0 additions and 2174 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/markdown/CUBLAS_API_supported_by_HIP.md
-538
View File
@@ -1,538 +0,0 @@
# CUBLAS API supported by HIP
## **1. CUBLAS Data types**
| **type** | **CUDA** |**CUDA version\***| **HIP** |**HIP value** (if differs) |
|-------------:|---------------------------------------------------------------|:----------------:|------------------------------------------------------------|---------------------------|
| define |`CUBLAS_VER_MAJOR` | 10.1 Update 2 | |
| define |`CUBLAS_VER_MINOR` | 10.1 Update 2 | |
| define |`CUBLAS_VER_PATCH` | 10.1 Update 2 | |
| define |`CUBLAS_VER_BUILD` | 10.1 Update 2 | |
| define |`CUBLAS_VERSION` | 10.1 Update 2 | |
| enum |***`cublasStatus`*** | |***`hipblasStatus_t`*** |
| enum |***`cublasStatus_t`*** | |***`hipblasStatus_t`*** |
| 0 |*`CUBLAS_STATUS_SUCCESS`* | |*`HIPBLAS_STATUS_SUCCESS`* |
| 1 |*`CUBLAS_STATUS_NOT_INITIALIZED`* | |*`HIPBLAS_STATUS_NOT_INITIALIZED`* |
| 3 |*`CUBLAS_STATUS_ALLOC_FAILED`* | |*`HIPBLAS_STATUS_ALLOC_FAILED`* | 2 |
| 7 |*`CUBLAS_STATUS_INVALID_VALUE`* | |*`HIPBLAS_STATUS_INVALID_VALUE`* | 3 |
| 8 |*`CUBLAS_STATUS_ARCH_MISMATCH`* | |*`HIPBLAS_STATUS_ARCH_MISMATCH`* | |
| 11 |*`CUBLAS_STATUS_MAPPING_ERROR`* | |*`HIPBLAS_STATUS_MAPPING_ERROR`* | 4 |
| 13 |*`CUBLAS_STATUS_EXECUTION_FAILED`* | |*`HIPBLAS_STATUS_EXECUTION_FAILED`* | 5 |
| 14 |*`CUBLAS_STATUS_INTERNAL_ERROR`* | |*`HIPBLAS_STATUS_INTERNAL_ERROR`* | 6 |
| 15 |*`CUBLAS_STATUS_NOT_SUPPORTED`* | |*`HIPBLAS_STATUS_NOT_SUPPORTED`* | 7 |
| 16 |*`CUBLAS_STATUS_LICENSE_ERROR`* | | |
| enum |***`cublasOperation_t`*** | |***`hipblasOperation_t`*** |
| 0 |*`CUBLAS_OP_N`* | |*`HIPBLAS_OP_N`* | 111 |
| 1 |*`CUBLAS_OP_T`* | |*`HIPBLAS_OP_T`* | 112 |
| 2 |*`CUBLAS_OP_C`* | |*`HIPBLAS_OP_C`* | 113 |
| 2 |*`CUBLAS_OP_HERMITAN`* | 10.1 |*`HIPBLAS_OP_C`* | 113 |
| 3 |*`CUBLAS_OP_CONJG`* | 10.1 | |
| enum |***`cublasFillMode_t`*** | |***`hipblasFillMode_t`*** |
| 0 |*`CUBLAS_FILL_MODE_LOWER`* | |*`HIPBLAS_FILL_MODE_LOWER`* | 121 |
| 1 |*`CUBLAS_FILL_MODE_UPPER`* | |*`HIPBLAS_FILL_MODE_UPPER`* | 122 |
| 2 |*`CUBLAS_FILL_MODE_FULL`* | 10.1 |*`HIPBLAS_FILL_MODE_FULL`* | 123 |
| enum |***`cublasDiagType_t`*** | |***`hipblasDiagType_t`*** |
| 0 |*`CUBLAS_DIAG_NON_UNIT`* | |*`HIPBLAS_DIAG_NON_UNIT`* | 131 |
| 1 |*`CUBLAS_DIAG_UNIT`* | |*`HIPBLAS_DIAG_UNIT`* | 132 |
| enum |***`cublasSideMode_t`*** | |***`hipblasSideMode_t`*** |
| 0 |*`CUBLAS_SIDE_LEFT`* | |*`HIPBLAS_SIDE_LEFT`* | 141 |
| 1 |*`CUBLAS_SIDE_RIGHT`* | |*`HIPBLAS_SIDE_RIGHT`* | 142 |
| enum |***`cublasPointerMode_t`*** | |***`hipblasPointerMode_t`*** |
| 0 |*`CUBLAS_POINTER_MODE_HOST`* | |*`HIPBLAS_POINTER_MODE_HOST`* |
| 1 |*`CUBLAS_POINTER_MODE_DEVICE`* | |*`HIPBLAS_POINTER_MODE_DEVICE`* |
| enum |***`cublasAtomicsMode_t`*** | | |
| 0 |*`CUBLAS_ATOMICS_NOT_ALLOWED`* | | |
| 1 |*`CUBLAS_ATOMICS_ALLOWED`* | | |
| enum |***`cublasGemmAlgo_t`*** | 8.0 |***`hipblasGemmAlgo_t`*** |
| -1 |*`CUBLAS_GEMM_DFALT`* | 8.0 |*`HIPBLAS_GEMM_DEFAULT`* | 160 |
| -1 |*`CUBLAS_GEMM_DEFAULT`* | 8.0 |*`HIPBLAS_GEMM_DEFAULT`* | 160 |
| 0 |*`CUBLAS_GEMM_ALGO0`* | 8.0 | |
| 1 |*`CUBLAS_GEMM_ALGO1`* | 8.0 | |
| 2 |*`CUBLAS_GEMM_ALGO2`* | 8.0 | |
| 3 |*`CUBLAS_GEMM_ALGO3`* | 8.0 | |
| 4 |*`CUBLAS_GEMM_ALGO4`* | 8.0 | |
| 5 |*`CUBLAS_GEMM_ALGO5`* | 8.0 | |
| 6 |*`CUBLAS_GEMM_ALGO6`* | 8.0 | |
| 7 |*`CUBLAS_GEMM_ALGO7`* | 8.0 | |
| 8 |*`CUBLAS_GEMM_ALGO8`* | 9.0 | |
| 9 |*`CUBLAS_GEMM_ALGO9`* | 9.0 | |
| 10 |*`CUBLAS_GEMM_ALGO10`* | 9.0 | |
| 11 |*`CUBLAS_GEMM_ALGO11`* | 9.0 | |
| 12 |*`CUBLAS_GEMM_ALGO12`* | 9.0 | |
| 13 |*`CUBLAS_GEMM_ALGO13`* | 9.0 | |
| 14 |*`CUBLAS_GEMM_ALGO14`* | 9.0 | |
| 15 |*`CUBLAS_GEMM_ALGO15`* | 9.0 | |
| 16 |*`CUBLAS_GEMM_ALGO16`* | 9.0 | |
| 17 |*`CUBLAS_GEMM_ALGO17`* | 9.0 | |
| 18 |*`CUBLAS_GEMM_ALGO18`* | 9.2 | |
| 19 |*`CUBLAS_GEMM_ALGO19`* | 9.2 | |
| 20 |*`CUBLAS_GEMM_ALGO20`* | 9.2 | |
| 21 |*`CUBLAS_GEMM_ALGO21`* | 9.2 | |
| 22 |*`CUBLAS_GEMM_ALGO22`* | 9.2 | |
| 23 |*`CUBLAS_GEMM_ALGO23`* | 9.2 | |
| 99 |*`CUBLAS_GEMM_DEFAULT_TENSOR_OP`* | 9.0 | |
| 99 |*`CUBLAS_GEMM_DFALT_TENSOR_OP`* | 9.0 | |
| 100 |*`CUBLAS_GEMM_ALGO0_TENSOR_OP`* | 9.0 | |
| 101 |*`CUBLAS_GEMM_ALGO1_TENSOR_OP`* | 9.0 | |
| 102 |*`CUBLAS_GEMM_ALGO2_TENSOR_OP`* | 9.0 | |
| 103 |*`CUBLAS_GEMM_ALGO3_TENSOR_OP`* | 9.0 | |
| 104 |*`CUBLAS_GEMM_ALGO4_TENSOR_OP`* | 9.0 | |
| 105 |*`CUBLAS_GEMM_ALGO5_TENSOR_OP`* | 9.2 | |
| 106 |*`CUBLAS_GEMM_ALGO6_TENSOR_OP`* | 9.2 | |
| 107 |*`CUBLAS_GEMM_ALGO7_TENSOR_OP`* | 9.2 | |
| 108 |*`CUBLAS_GEMM_ALGO8_TENSOR_OP`* | 9.2 | |
| 109 |*`CUBLAS_GEMM_ALGO9_TENSOR_OP`* | 9.2 | |
| 110 |*`CUBLAS_GEMM_ALGO10_TENSOR_OP`* | 9.2 | |
| 111 |*`CUBLAS_GEMM_ALGO11_TENSOR_OP`* | 9.2 | |
| 112 |*`CUBLAS_GEMM_ALGO12_TENSOR_OP`* | 9.2 | |
| 113 |*`CUBLAS_GEMM_ALGO13_TENSOR_OP`* | 9.2 | |
| 114 |*`CUBLAS_GEMM_ALGO14_TENSOR_OP`* | 9.2 | |
| 115 |*`CUBLAS_GEMM_ALGO15_TENSOR_OP`* | 9.2 | |
| enum |***`cublasMath_t`*** | 9.0 | |
| 0 |*`CUBLAS_DEFAULT_MATH`* | 9.0 | |
| 1 |*`CUBLAS_TENSOR_OP_MATH`* | 9.0 | |
| enum* |`cublasDataType_t` | 7.5 | |
| struct |`cublasContext` | | |
| struct* |`cublasHandle_t` | |`hipblasHandle_t` |
## **2. CUBLAS API functions**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cublasCreate` |`hipblasCreate` |
|`cublasCreate_v2` |`hipblasCreate` |
|`cublasDestroy` |`hipblasDestroy` |
|`cublasDestroy_v2` |`hipblasDestroy` |
|`cublasGetVersion` | |
|`cublasGetVersion_v2` | |
|`cublasGetProperty` | | 8.0 |
|`cublasGetCudartVersion` | | 10.1 |
|`cublasGetStream` |`hipblasGetStream` |
|`cublasGetStream_v2` |`hipblasGetStream` |
|`cublasSetStream` |`hipblasSetStream` |
|`cublasSetStream_v2` |`hipblasSetStream` |
|`cublasGetPointerMode` |`hipblasGetPointerMode` |
|`cublasGetPointerMode_v2` |`hipblasGetPointerMode` |
|`cublasSetPointerMode` |`hipblasSetPointerMode` |
|`cublasSetPointerMode_v2` |`hipblasSetPointerMode` |
|`cublasGetAtomicsMode` | |
|`cublasSetAtomicsMode` | |
|`cublasGetMathMode` | | 9.0 |
|`cublasSetMathMode` | | 9.0 |
|`cublasLogCallback` | | 9.2 |
|`cublasLoggerConfigure` | | 9.2 |
|`cublasSetLoggerCallback` | | 9.2 |
|`cublasGetLoggerCallback` | | 9.2 |
|`cublasSetVector` |`hipblasSetVector` |
|`cublasGetVector` |`hipblasGetVector` |
|`cublasSetMatrix` |`hipblasSetMatrix` |
|`cublasGetMatrix` |`hipblasGetMatrix` |
|`cublasSetVectorAsync` | |
|`cublasGetVectorAsync` | |
|`cublasSetMatrixAsync` | |
|`cublasGetMatrixAsync` | |
|`cublasXerbla` | |
|`cublasNrm2Ex` | | 8.0 |
|`cublasSnrm2` |`hipblasSnrm2` |
|`cublasSnrm2_v2` |`hipblasSnrm2` |
|`cublasDnrm2` |`hipblasDnrm2` |
|`cublasDnrm2_v2` |`hipblasDnrm2` |
|`cublasScnrm2` |`hipblasScnrm2` |
|`cublasScnrm2_v2` |`hipblasScnrm2` |
|`cublasDznrm2` |`hipblasDznrm2` |
|`cublasDznrm2_v2` |`hipblasDznrm2` |
|`cublasDotEx` | | 8.0 |
|`cublasDotcEx` | | 8.0 |
|`cublasSdot` |`hipblasSdot` |
|`cublasSdot_v2` |`hipblasSdot` |
|`cublasDdot` |`hipblasDdot` |
|`cublasDdot_v2` |`hipblasDdot` |
|`cublasCdotu` |`hipblasCdotu` |
|`cublasCdotu_v2` |`hipblasCdotu` |
|`cublasCdotc` |`hipblasCdotc` |
|`cublasCdotc_v2` |`hipblasCdotc` |
|`cublasZdotu` |`hipblasZdotu` |
|`cublasZdotu_v2` |`hipblasZdotu` |
|`cublasZdotc` |`hipblasZdotc` |
|`cublasZdotc_v2` |`hipblasZdotc` |
|`cublasScalEx` | | 8.0 |
|`cublasSscal` |`hipblasSscal` |
|`cublasSscal_v2` |`hipblasSscal` |
|`cublasDscal` |`hipblasDscal` |
|`cublasDscal_v2` |`hipblasDscal` |
|`cublasCscal` |`hipblasCscal` |
|`cublasCscal_v2` |`hipblasCscal` |
|`cublasCsscal` |`hipblasCsscal` |
|`cublasCsscal_v2` |`hipblasCsscal` |
|`cublasZscal` |`hipblasZscal` |
|`cublasZscal_v2` |`hipblasZscal` |
|`cublasZdscal` |`hipblasZdscal` |
|`cublasZdscal_v2` |`hipblasZdscal` |
|`cublasAxpyEx` | | 8.0 |
|`cublasSaxpy` |`hipblasSaxpy` |
|`cublasSaxpy_v2` |`hipblasSaxpy` |
|`cublasDaxpy` |`hipblasDaxpy` |
|`cublasDaxpy_v2` |`hipblasDaxpy` |
|`cublasCaxpy` |`hipblasCaxpy` |
|`cublasCaxpy_v2` |`hipblasCaxpy` |
|`cublasZaxpy` |`hipblasZaxpy` |
|`cublasZaxpy_v2` |`hipblasZaxpy` |
|`cublasScopy` |`hipblasScopy` |
|`cublasScopy_v2` |`hipblasScopy` |
|`cublasDcopy` |`hipblasDcopy` |
|`cublasDcopy_v2` |`hipblasDcopy` |
|`cublasCcopy` |`hipblasCcopy` |
|`cublasCopyEx` | | 10.1 |
|`cublasCcopy_v2` |`hipblasCcopy` |
|`cublasZcopy` |`hipblasZcopy` |
|`cublasZcopy_v2` |`hipblasZcopy` |
|`cublasSswap` |`hipblasSswap` |
|`cublasSswap_v2` |`hipblasSswap` |
|`cublasDswap` |`hipblasDswap` |
|`cublasDswap_v2` |`hipblasDswap` |
|`cublasCswap` |`hipblasCswap` |
|`cublasCswap_v2` |`hipblasCswap` |
|`cublasZswap` |`hipblasZswap` |
|`cublasZswap_v2` |`hipblasZswap` |
|`cublasIamaxEx` | | 10.1 |
|`cublasIsamax` |`hipblasIsamax` |
|`cublasIsamax_v2` |`hipblasIsamax` |
|`cublasIdamax` |`hipblasIdamax` |
|`cublasIdamax_v2` |`hipblasIdamax` |
|`cublasIcamax` |`hipblasIcamax` |
|`cublasIcamax_v2` |`hipblasIcamax` |
|`cublasIzamax` |`hipblasIzamax` |
|`cublasIzamax_v2` |`hipblasIzamax` |
|`cublasIaminEx` | | 10.1 |
|`cublasIsamin` |`hipblasIsamin` |
|`cublasIsamin_v2` |`hipblasIsamin` |
|`cublasIdamin` |`hipblasIdamin` |
|`cublasIdamin_v2` |`hipblasIdamin` |
|`cublasIcamin` |`hipblasIcamin` |
|`cublasIcamin_v2` |`hipblasIcamin` |
|`cublasIzamin` |`hipblasIzamin` |
|`cublasIzamin_v2` |`hipblasIzamin` |
|`cublasAsumEx` | | 10.1 |
|`cublasSasum` |`hipblasSasum` |
|`cublasSasum_v2` |`hipblasSasum` |
|`cublasDasum` |`hipblasDasum` |
|`cublasDasum_v2` |`hipblasDasum` |
|`cublasScasum` |`hipblasScasum` |
|`cublasScasum_v2` |`hipblasScasum` |
|`cublasDzasum` |`hipblasDzasum` |
|`cublasDzasum_v2` |`hipblasDzasum` |
|`cublasRotEx` | | 10.1 |
|`cublasSrot` |`hipblasSrot` |
|`cublasSrot_v2` |`hipblasSrot` |
|`cublasDrot` |`hipblasDrot` |
|`cublasDrot_v2` |`hipblasDrot` |
|`cublasCrot` |`hipblasCrot` |
|`cublasCrot_v2` |`hipblasCrot` |
|`cublasCsrot` |`hipblasCsrot` |
|`cublasCsrot_v2` |`hipblasCsrot` |
|`cublasZrot` |`hipblasZrot` |
|`cublasZrot_v2` |`hipblasZrot` |
|`cublasRotgEx` | | 10.1 |
|`cublasZdrot` |`hipblasZdrot` |
|`cublasZdrot_v2` |`hipblasZdrot` |
|`cublasSrotg` |`hipblasSrotg` |
|`cublasSrotg_v2` |`hipblasSrotg` |
|`cublasDrotg` |`hipblasDrotg` |
|`cublasDrotg_v2` |`hipblasDrotg` |
|`cublasCrotg` |`hipblasCrotg` |
|`cublasCrotg_v2` |`hipblasCrotg` |
|`cublasZrotg` |`hipblasZrotg` |
|`cublasZrotg_v2` |`hipblasZrotg` |
|`cublasRotmEx` | | 10.1 |
|`cublasSrotm` |`hipblasSrotm` |
|`cublasSrotm_v2` |`hipblasSrotm` |
|`cublasDrotm` |`hipblasDrotm` |
|`cublasDrotm_v2` |`hipblasDrotm` |
|`cublasRotmgEx` | | 10.1 |
|`cublasSrotmg` |`hipblasSrotmg` |
|`cublasSrotmg_v2` |`hipblasSrotmg` |
|`cublasDrotmg` |`hipblasDrotmg` |
|`cublasDrotmg_v2` |`hipblasDrotmg` |
|`cublasSgemv` |`hipblasSgemv` |
|`cublasSgemv_v2` |`hipblasSgemv` |
|`cublasSgemvBatched` |`hipblasSgemvBatched` |
|`cublasDgemv` |`hipblasDgemv` |
|`cublasDgemv_v2` |`hipblasDgemv` |
|`cublasCgemv` |`hipblasCgemv` |
|`cublasCgemv_v2` |`hipblasCgemv` |
|`cublasZgemv` |`hipblasZgemv` |
|`cublasZgemv_v2` |`hipblasZgemv` |
|`cublasSgbmv` | |
|`cublasSgbmv_v2` | |
|`cublasDgbmv` | |
|`cublasDgbmv_v2` | |
|`cublasCgbmv` | |
|`cublasCgbmv_v2` | |
|`cublasZgbmv` | |
|`cublasZgbmv_v2` | |
|`cublasStrmv` | |
|`cublasStrmv_v2` | |
|`cublasDtrmv` | |
|`cublasDtrmv_v2` | |
|`cublasCtrmv` | |
|`cublasCtrmv_v2` | |
|`cublasZtrmv` | |
|`cublasZtrmv_v2` | |
|`cublasStbmv` | |
|`cublasStbmv_v2` | |
|`cublasDtbmv` | |
|`cublasDtbmv_v2` | |
|`cublasCtbmv` | |
|`cublasCtbmv_v2` | |
|`cublasZtbmv` | |
|`cublasZtbmv_v2` | |
|`cublasStpmv` | |
|`cublasStpmv_v2` | |
|`cublasDtpmv` | |
|`cublasDtpmv_v2` | |
|`cublasCtpmv` | |
|`cublasCtpmv_v2` | |
|`cublasZtpmv` | |
|`cublasZtpmv_v2` | |
|`cublasStrsv` |`hipblasStrsv` |
|`cublasStrsv_v2` |`hipblasStrsv` |
|`cublasDtrsv` |`hipblasDtrsv` |
|`cublasDtrsv_v2` |`hipblasDtrsv` |
|`cublasCtrsv` | |
|`cublasCtrsv_v2` | |
|`cublasZtrsv` | |
|`cublasZtrsv_v2` | |
|`cublasStpsv` | |
|`cublasStpsv_v2` | |
|`cublasDtpsv` | |
|`cublasDtpsv_v2` | |
|`cublasCtpsv` | |
|`cublasCtpsv_v2` | |
|`cublasZtpsv` | |
|`cublasZtpsv_v2` | |
|`cublasStbsv` | |
|`cublasStbsv_v2` | |
|`cublasDtbsv` | |
|`cublasDtbsv_v2` | |
|`cublasCtbsv` | |
|`cublasCtbsv_v2` | |
|`cublasZtbsv` | |
|`cublasZtbsv_v2` | |
|`cublasSsymv` | |
|`cublasSsymv_v2` | |
|`cublasDsymv` | |
|`cublasDsymv_v2` | |
|`cublasCsymv` | |
|`cublasCsymv_v2` | |
|`cublasZsymv` | |
|`cublasZsymv_v2` | |
|`cublasChemv` | |
|`cublasChemv_v2` | |
|`cublasZhemv` | |
|`cublasZhemv_v2` | |
|`cublasSsbmv` | |
|`cublasSsbmv_v2` | |
|`cublasDsbmv` | |
|`cublasDsbmv_v2` | |
|`cublasChbmv` | |
|`cublasChbmv_v2` | |
|`cublasZhbmv` | |
|`cublasZhbmv_v2` | |
|`cublasSspmv` | |
|`cublasSspmv_v2` | |
|`cublasDspmv` | |
|`cublasDspmv_v2` | |
|`cublasChpmv` | |
|`cublasChpmv_v2` | |
|`cublasZhpmv` | |
|`cublasZhpmv_v2` | |
|`cublasSger` |`hipblasSger` |
|`cublasSger_v2` |`hipblasSger` |
|`cublasDger` |`hipblasDger` |
|`cublasDger_v2` |`hipblasDger` |
|`cublasCgeru` | |
|`cublasCgeru_v2` | |
|`cublasCgerc` | |
|`cublasCgerc_v2` | |
|`cublasZgeru` | |
|`cublasZgeru_v2` | |
|`cublasZgerc` | |
|`cublasZgerc_v2` | |
|`cublasSsyr` |`hipblasSsyr` |
|`cublasSsyr_v2` |`hipblasSsyr` |
|`cublasDsyr` |`hipblasDsyr` |
|`cublasDsyr_v2` |`hipblasDsyr` |
|`cublasCsyr` | |
|`cublasCsyr_v2` | |
|`cublasZsyr` | |
|`cublasZsyr_v2` | |
|`cublasCher` | |
|`cublasCher_v2` | |
|`cublasZher` | |
|`cublasZher_v2` | |
|`cublasSspr` | |
|`cublasSspr_v2` | |
|`cublasDspr` | |
|`cublasDspr_v2` | |
|`cublasChpr` | |
|`cublasChpr_v2` | |
|`cublasZhpr` | |
|`cublasZhpr_v2` | |
|`cublasSsyr2` | |
|`cublasSsyr2_v2` | |
|`cublasDsyr2` | |
|`cublasDsyr2_v2` | |
|`cublasCsyr2` | |
|`cublasCsyr2_v2` | |
|`cublasZsyr2` | |
|`cublasZsyr2_v2` | |
|`cublasCher2` | |
|`cublasCher2_v2` | |
|`cublasZher2` | |
|`cublasZher2_v2` | |
|`cublasSspr2` | |
|`cublasSspr2_v2` | |
|`cublasDspr2` | |
|`cublasDspr2_v2` | |
|`cublasChpr2` | |
|`cublasChpr2_v2` | |
|`cublasZhpr2` | |
|`cublasZhpr2_v2` | |
|`cublasSgemm` |`hipblasSgemm` |
|`cublasSgemm_v2` |`hipblasSgemm` |
|`cublasDgemm` |`hipblasDgemm` |
|`cublasDgemm_v2` |`hipblasDgemm` |
|`cublasCgemm` | |
|`cublasCgemm_v2` | |
|`cublasCgemm3m` | | 8.0 |
|`cublasCgemm3mEx` | | 8.0 |
|`cublasZgemm` |`hipblasZgemm` |
|`cublasZgemm_v2` |`hipblasZgemm` |
|`cublasZgemm3m` | | 8.0 |
|`cublasHgemm` |`hipblasHgemm` | 7.5 |
|`cublasSgemmEx` | | 7.5 |
|`cublasGemmEx` |`hipblasGemmEx` | 8.0 |
|`cublasCgemmEx` | | 8.0 |
|`cublasUint8gemmBias` | | 8.0 |
|`cublasSsyrk` | |
|`cublasSsyrk_v2` | |
|`cublasDsyrk` | |
|`cublasDsyrk_v2` | |
|`cublasCsyrk` | |
|`cublasCsyrk_v2` | |
|`cublasZsyrk` | |
|`cublasZsyrk_v2` | |
|`cublasCsyrkEx` | | 8.0 |
|`cublasCsyrk3mEx` | | 8.0 |
|`cublasCherk` | |
|`cublasCherk_v2` | |
|`cublasZherk` | |
|`cublasZherk_v2` | |
|`cublasCherkEx` | | 8.0 |
|`cublasCherk3mEx` | | 8.0 |
|`cublasSsyr2k` | |
|`cublasSsyr2k_v2` | |
|`cublasDsyr2k` | |
|`cublasDsyr2k_v2` | |
|`cublasCsyr2k` | |
|`cublasCsyr2k_v2` | |
|`cublasZsyr2k` | |
|`cublasZsyr2k_v2` | |
|`cublasCher2k` | |
|`cublasCher2k_v2` | |
|`cublasZher2k` | |
|`cublasZher2k_v2` | |
|`cublasSsyrkx` | |
|`cublasDsyrkx` | |
|`cublasCsyrkx` | |
|`cublasZsyrkx` | |
|`cublasCherkx` | |
|`cublasZherkx` | |
|`cublasSsymm` | |
|`cublasSsymm_v2` | |
|`cublasDsymm` | |
|`cublasDsymm_v2` | |
|`cublasCsymm` | |
|`cublasCsymm_v2` | |
|`cublasZsymm` | |
|`cublasZsymm_v2` | |
|`cublasChemm` | |
|`cublasChemm_v2` | |
|`cublasZhemm` | |
|`cublasZhemm_v2` | |
|`cublasStrsm` |`hipblasStrsm` |
|`cublasStrsm_v2` |`hipblasStrsm` |
|`cublasDtrsm` |`hipblasDtrsm` |
|`cublasDtrsm_v2` |`hipblasDtrsm` |
|`cublasCtrsm` | |
|`cublasCtrsm_v2` | |
|`cublasZtrsm` | |
|`cublasZtrsm_v2` | |
|`cublasStrmm` | |
|`cublasStrmm_v2` | |
|`cublasDtrmm` | |
|`cublasDtrmm_v2` | |
|`cublasCtrmm` | |
|`cublasCtrmm_v2` | |
|`cublasZtrmm` | |
|`cublasZtrmm_v2` | |
|`cublasHgemmBatched` |`hipblasHgemmBatched` | 9.0 |
|`cublasSgemmBatched` |`hipblasSgemmBatched` |
|`cublasDgemmBatched` |`hipblasDgemmBatched` |
|`cublasCgemmBatched` |`hipblasCgemmBatched` |
|`cublasCgemm3mBatched` | | 8.0 |
|`cublasZgemmBatched` |`hipblasZgemmBatched` |
|`cublasGemmBatchedEx` | | 9.1 |
|`cublasGemmStridedBatchedEx` | | 9.1 |
|`cublasSgemmStridedBatched` |`hipblasSgemmStridedBatched` | 8.0 |
|`cublasDgemmStridedBatched` |`hipblasDgemmStridedBatched` | 8.0 |
|`cublasCgemmStridedBatched` |`hipblasCgemmStridedBatched` | 8.0 |
|`cublasCgemm3mStridedBatched` | | 8.0 |
|`cublasZgemmStridedBatched` |`hipblasZgemmStridedBatched` | 8.0 |
|`cublasHgemmStridedBatched` |`hipblasHgemmStridedBatched` | 8.0 |
|`cublasSgeam` |`hipblasSgeam` |
|`cublasDgeam` |`hipblasDgeam` |
|`cublasCgeam` | |
|`cublasZgeam` | |
|`cublasSgetrfBatched` | |
|`cublasDgetrfBatched` | |
|`cublasCgetrfBatched` | |
|`cublasZgetrfBatched` | |
|`cublasSgetriBatched` | |
|`cublasDgetriBatched` | |
|`cublasCgetriBatched` | |
|`cublasZgetriBatched` | |
|`cublasSgetrsBatched` | |
|`cublasDgetrsBatched` | |
|`cublasCgetrsBatched` | |
|`cublasZgetrsBatched` | |
|`cublasStrsmBatched` | |
|`cublasDtrsmBatched` | |
|`cublasCtrsmBatched` | |
|`cublasZtrsmBatched` | |
|`cublasSmatinvBatched` | |
|`cublasDmatinvBatched` | |
|`cublasCmatinvBatched` | |
|`cublasZmatinvBatched` | |
|`cublasSgeqrfBatched` | |
|`cublasDgeqrfBatched` | |
|`cublasCgeqrfBatched` | |
|`cublasZgeqrfBatched` | |
|`cublasSgelsBatched` | |
|`cublasDgelsBatched` | |
|`cublasCgelsBatched` | |
|`cublasZgelsBatched` | |
|`cublasSdgmm` | |
|`cublasDdgmm` | |
|`cublasCdgmm` | |
|`cublasZdgmm` | |
|`cublasStpttr` | |
|`cublasDtpttr` | |
|`cublasCtpttr` | |
|`cublasZtpttr` | |
|`cublasStrttp` | |
|`cublasDtrttp` | |
|`cublasCtrttp` | |
|`cublasZtrttp` | |
\* CUDA version, in which API has appeared and (optional) last version before abandoning it; no value in case of earlier versions < 7.5.
docs/markdown/CUDNN_API_supported_by_HIP.md
-586
View File
@@ -1,586 +0,0 @@
# CUDNN API supported by HIP
## **1. CUDNN Data types**
| **type** | **CUDA** |**CUDA version\***| **HIP** |**HIP value** (if differs) |
|-------------:|---------------------------------------------------------------|:----------------:|------------------------------------------------------------|---------------------------|
| define |`CUDNN_VERSION` | |`HIPDNN_VERSION` |
| struct |`cudnnContext` | | |
| struct* |`cudnnHandle_t` | |`hipdnnHandle_t` |
| enum |***`cudnnStatus_t`*** | |***`hipdnnStatus_t`*** |
| 0 |*`CUDNN_STATUS_SUCCESS`* | |*`HIPDNN_STATUS_SUCCESS`* |
| 1 |*`CUDNN_STATUS_NOT_INITIALIZED`* | |*`HIPDNN_STATUS_NOT_INITIALIZED`* |
| 2 |*`CUDNN_STATUS_ALLOC_FAILED`* | |*`HIPDNN_STATUS_ALLOC_FAILED`* |
| 3 |*`CUDNN_STATUS_BAD_PARAM`* | |*`HIPDNN_STATUS_BAD_PARAM`* |
| 4 |*`CUDNN_STATUS_INTERNAL_ERROR`* | |*`HIPDNN_STATUS_INTERNAL_ERROR`* |
| 5 |*`CUDNN_STATUS_INVALID_VALUE`* | |*`HIPDNN_STATUS_INVALID_VALUE`* |
| 6 |*`CUDNN_STATUS_ARCH_MISMATCH`* | |*`HIPDNN_STATUS_ARCH_MISMATCH`* |
| 7 |*`CUDNN_STATUS_MAPPING_ERROR`* | |*`HIPDNN_STATUS_MAPPING_ERROR`* |
| 8 |*`CUDNN_STATUS_EXECUTION_FAILED`* | |*`HIPDNN_STATUS_EXECUTION_FAILED`* |
| 9 |*`CUDNN_STATUS_NOT_SUPPORTED`* | |*`HIPDNN_STATUS_NOT_SUPPORTED`* |
| 10 |*`CUDNN_STATUS_LICENSE_ERROR`* | |*`HIPDNN_STATUS_LICENSE_ERROR`* |
| 11 |*`CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING`* | 7.5 |*`HIPDNN_STATUS_RUNTIME_PREREQUISITE_MISSING`* |
| 12 |*`CUDNN_STATUS_RUNTIME_IN_PROGRESS`* | 8.0 | |
| 13 |*`CUDNN_STATUS_RUNTIME_FP_OVERFLOW`* | 8.0 | |
| struct |`cudnnRuntimeTag_t` | 8.0 | |
| enum |***`cudnnErrQueryMode_t`*** | 8.0 | |
| 0 |*`CUDNN_ERRQUERY_RAWCODE`* | 8.0 | |
| 1 |*`CUDNN_ERRQUERY_NONBLOCKING`* | 8.0 | |
| 2 |*`CUDNN_ERRQUERY_BLOCKING`* | 8.0 | |
| struct |`cudnnTensorStruct` | | |
| struct* |`cudnnTensorDescriptor_t` | |`hipdnnTensorDescriptor_t` |
| struct |`cudnnConvolutionStruct` | | |
| struct* |`cudnnConvolutionDescriptor_t` | |`hipdnnConvolutionDescriptor_t` |
| struct |`cudnnPoolingStruct` | | |
| struct* |`cudnnPoolingDescriptor_t` | |`hipdnnPoolingDescriptor_t` |
| struct |`cudnnFilterStruct` | | |
| struct* |`cudnnFilterDescriptor_t` | |`hipdnnFilterDescriptor_t` |
| struct |`cudnnLRNStruct` | | |
| struct* |`cudnnLRNDescriptor_t` | |`hipdnnLRNDescriptor_t` |
| struct |`cudnnActivationStruct` | | |
| struct* |`cudnnActivationDescriptor_t` | |`hipdnnActivationDescriptor_t` |
| struct |`cudnnSpatialTransformerStruct` | 7.5 | |
| struct* |`cudnnSpatialTransformerDescriptor_t` | 7.5 | |
| struct |`cudnnOpTensorStruct` | 7.5 | |
| struct* |`cudnnOpTensorDescriptor_t` | 7.5 |`hipdnnOpTensorDescriptor_t` |
| struct |`cudnnReduceTensorStruct` | 7.5 | |
| struct* |`cudnnReduceTensorDescriptor_t` | 7.5 |`hipdnnReduceTensorDescriptor_t` |
| struct |`cudnnCTCLossStruct` | 8.0 | |
| struct* |`cudnnCTCLossDescriptor_t` | 8.0 | |
| struct |`cudnnTensorTransformStruct` | 9.0 | |
| struct* |`cudnnTensorTransformDescriptor_t` | 9.0 | |
| enum |***`cudnnDataType_t`*** | |***`hipdnnDataType_t`*** |
| 0 |*`CUDNN_DATA_FLOAT`* | |*`HIPDNN_DATA_FLOAT`* |
| 1 |*`CUDNN_DATA_DOUBLE`* | |*`HIPDNN_DATA_DOUBLE`* |
| 2 |*`CUDNN_DATA_HALF`* | |*`HIPDNN_DATA_HALF`* |
| 3 |*`CUDNN_DATA_INT8`* | 7.5 |*`HIPDNN_DATA_INT8`* |
| 4 |*`CUDNN_DATA_INT32`* | 7.5 |*`HIPDNN_DATA_INT32`* |
| 5 |*`CUDNN_DATA_INT8x4`* | 7.5 |*`HIPDNN_DATA_INT8x4`* |
| 6 |*`CUDNN_DATA_UINT8`* | 8.0 | |
| 7 |*`CUDNN_DATA_UINT8x4`* | 8.0 | |
| 8 |*`CUDNN_DATA_INT8x32`* | 9.0 | |
| enum |***`cudnnMathType_t`*** | 8.0 |***`hipdnnMathType_t`*** |
| 0 |*`CUDNN_DEFAULT_MATH`* | 8.0 |*`HIPDNN_DEFAULT_MATH`* |
| 1 |*`CUDNN_TENSOR_OP_MATH`* | 8.0 |*`HIPDNN_TENSOR_OP_MATH`* |
| enum |***`cudnnNanPropagation_t`*** | |***`hipdnnNanPropagation_t`*** |
| 0 |*`CUDNN_NOT_PROPAGATE_NAN`* | |*`HIPDNN_NOT_PROPAGATE_NAN`* |
| 1 |*`CUDNN_PROPAGATE_NAN`* | |*`HIPDNN_PROPAGATE_NAN`* |
| enum |***`cudnnDeterminism_t`*** | 7.5 | |
| 0 |*`CUDNN_NON_DETERMINISTIC`* | 7.5 | |
| 1 |*`CUDNN_DETERMINISTIC`* | 7.5 | |
| define |`CUDNN_DIM_MAX` | | |
| enum |***`cudnnTensorFormat_t`*** | |***`hipdnnTensorFormat_t`*** |
| 0 |*`CUDNN_TENSOR_NCHW`* | |*`HIPDNN_TENSOR_NCHW`* |
| 1 |*`CUDNN_TENSOR_NHWC`* | |*`HIPDNN_TENSOR_NHWC`* |
| 2 |*`CUDNN_TENSOR_NCHW_VECT_C`* | 7.5 |*`HIPDNN_TENSOR_NCHW_VECT_C`* |
| enum |***`cudnnFoldingDirection_t`*** | 9.0 | |
| 0 |*`CUDNN_TRANSFORM_FOLD`* | 9.0 | |
| 1 |*`CUDNN_TRANSFORM_UNFOLD`* | 9.0 | |
| enum |***`cudnnOpTensorOp_t`*** | 7.5 |***`hipdnnOpTensorOp_t`*** |
| 0 |*`CUDNN_OP_TENSOR_ADD`* | 7.5 |*`HIPDNN_OP_TENSOR_ADD`* |
| 1 |*`CUDNN_OP_TENSOR_MUL`* | 7.5 |*`HIPDNN_OP_TENSOR_MUL`* |
| 2 |*`CUDNN_OP_TENSOR_MIN`* | 7.5 |*`HIPDNN_OP_TENSOR_MIN`* |
| 3 |*`CUDNN_OP_TENSOR_MAX`* | 7.5 |*`HIPDNN_OP_TENSOR_MAX`* |
| 4 |*`CUDNN_OP_TENSOR_SQRT`* | 7.5 |*`HIPDNN_OP_TENSOR_SQRT`* |
| 5 |*`CUDNN_OP_TENSOR_NOT`* | 8.0 |*`HIPDNN_OP_TENSOR_NOT`* |
| enum |***`cudnnReduceTensorOp_t`*** | 7.5 |***`hipdnnReduceTensorOp_t`*** |
| 0 |*`CUDNN_REDUCE_TENSOR_ADD`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_ADD`* |
| 1 |*`CUDNN_REDUCE_TENSOR_MUL`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_MUL`* |
| 2 |*`CUDNN_REDUCE_TENSOR_MIN`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_MIN`* |
| 3 |*`CUDNN_REDUCE_TENSOR_MAX`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_MAX`* |
| 4 |*`CUDNN_REDUCE_TENSOR_AMAX`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_AMAX`* |
| 5 |*`CUDNN_REDUCE_TENSOR_AVG`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_AVG`* |
| 6 |*`CUDNN_REDUCE_TENSOR_NORM1`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_NORM1`* |
| 7 |*`CUDNN_REDUCE_TENSOR_NORM2`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_NORM2`* |
| 8 |*`CUDNN_REDUCE_TENSOR_MUL_NO_ZEROS`* | 8.0 |*`HIPDNN_REDUCE_TENSOR_MUL_NO_ZEROS`* |
| enum |***`cudnnReduceTensorIndices_t`*** | 7.5 |***`hipdnnReduceTensorIndices_t`*** |
| 0 |*`CUDNN_REDUCE_TENSOR_NO_INDICES`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_NO_INDICES`* |
| 1 |*`CUDNN_REDUCE_TENSOR_FLATTENED_INDICES`* | 7.5 |*`HIPDNN_REDUCE_TENSOR_FLATTENED_INDICES`* |
| enum |***`cudnnIndicesType_t`*** | 7.5 |***`hipdnnIndicesType_t`*** |
| 0 |*`CUDNN_32BIT_INDICES`* | 7.5 |*`HIPDNN_32BIT_INDICES`* |
| 1 |*`CUDNN_64BIT_INDICES`* | 7.5 |*`HIPDNN_64BIT_INDICES`* |
| 2 |*`CUDNN_16BIT_INDICES`* | 7.5 |*`HIPDNN_16BIT_INDICES`* |
| 3 |*`CUDNN_8BIT_INDICES`* | 7.5 |*`HIPDNN_8BIT_INDICES`* |
| enum |***`cudnnConvolutionMode_t`*** | |***`hipdnnConvolutionMode_t`*** |
| 0 |*`CUDNN_CONVOLUTION`* | |*`HIPDNN_CONVOLUTION`* |
| 1 |*`CUDNN_CROSS_CORRELATION`* | |*`HIPDNN_CROSS_CORRELATION`* |
| enum |***`cudnnConvolutionFwdPreference_t`*** | |***`hipdnnConvolutionFwdPreference_t`*** |
| 0 |*`CUDNN_CONVOLUTION_FWD_NO_WORKSPACE`* | |*`HIPDNN_CONVOLUTION_FWD_NO_WORKSPACE`* |
| 1 |*`CUDNN_CONVOLUTION_FWD_PREFER_FASTEST`* | |*`HIPDNN_CONVOLUTION_FWD_PREFER_FASTEST`* |
| 2 |*`CUDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT`* | |*`HIPDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT`* |
| enum |***`cudnnConvolutionFwdAlgo_t`*** | |***`hipdnnConvolutionFwdAlgo_t`*** |
| 0 |*`CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM`* |
| 1 |*`CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_PRECOMP_GEMM`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_PRECOMP_GEMM`* |
| 2 |*`CUDNN_CONVOLUTION_FWD_ALGO_GEMM`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_GEMM`* |
| 3 |*`CUDNN_CONVOLUTION_FWD_ALGO_DIRECT`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_DIRECT`* |
| 4 |*`CUDNN_CONVOLUTION_FWD_ALGO_FFT`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_FFT`* |
| 5 |*`CUDNN_CONVOLUTION_FWD_ALGO_FFT_TILING`* | |*`HIPDNN_CONVOLUTION_FWD_ALGO_FFT_TILING`* |
| 6 |*`CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD`* | 7.5 |*`HIPDNN_CONVOLUTION_FWD_ALGO_WINOGRAD`* |
| 7 |*`CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD_NONFUSED`* | 7.5 |*`HIPDNN_CONVOLUTION_FWD_ALGO_WINOGRAD_NONFUSED`* |
| 8 |*`CUDNN_CONVOLUTION_FWD_ALGO_COUNT`* | 7.5 |*`HIPDNN_CONVOLUTION_FWD_ALGO_COUNT`* |
| struct |`cudnnConvolutionFwdAlgoPerf_t` | |`hipdnnConvolutionFwdAlgoPerf_t` |
| enum |***`cudnnConvolutionBwdFilterPreference_t`*** | |***`hipdnnConvolutionBwdFilterPreference_t`*** |
| 0 |*`CUDNN_CONVOLUTION_BWD_FILTER_NO_WORKSPACE`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_NO_WORKSPACE`* |
| 1 |*`CUDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST`* |
| 2 |*`CUDNN_CONVOLUTION_BWD_FILTER_SPECIFY_WORKSPACE_LIMIT`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_SPECIFY_WORKSPACE_LIMIT`* |
| enum |***`cudnnConvolutionBwdFilterAlgo_t`*** | |***`hipdnnConvolutionBwdFilterAlgo_t`*** |
| 0 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_0`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_0`* |
| 1 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_1`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_1`* |
| 2 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT`* |
| 3 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_3`* | |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_3`* |
| 4 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD`* | 7.5 |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD`* |
| 5 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD_NONFUSED`* | 7.5 |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD_NONFUSED`* |
| 6 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT_TILING`* | 7.5 |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT_TILING`* |
| 7 |*`CUDNN_CONVOLUTION_BWD_FILTER_ALGO_COUNT`* | 7.5 |*`HIPDNN_CONVOLUTION_BWD_FILTER_ALGO_COUNT`* |
| struct |`cudnnConvolutionBwdDataAlgoPerf_t` | |`hipdnnConvolutionBwdDataAlgoPerf_t` |
| enum |***`cudnnSoftmaxAlgorithm_t`*** | |***`hipdnnSoftmaxAlgorithm_t`*** |
| 0 |*`CUDNN_SOFTMAX_FAST`* | |*`HIPDNN_SOFTMAX_FAST`* |
| 1 |*`CUDNN_SOFTMAX_ACCURATE`* | |*`HIPDNN_SOFTMAX_ACCURATE`* |
| 2 |*`CUDNN_SOFTMAX_LOG`* | |*`HIPDNN_SOFTMAX_LOG`* |
| enum |***`cudnnSoftmaxMode_t`*** | |***`hipdnnSoftmaxMode_t`*** |
| 0 |*`CUDNN_SOFTMAX_MODE_INSTANCE`* | |*`HIPDNN_SOFTMAX_MODE_INSTANCE`* |
| 1 |*`CUDNN_SOFTMAX_MODE_CHANNEL`* | |*`HIPDNN_SOFTMAX_MODE_CHANNEL`* |
| enum |***`cudnnPoolingMode_t`*** | |***`hipdnnPoolingMode_t`*** |
| 0 |*`CUDNN_POOLING_MAX`* | |*`HIPDNN_POOLING_MAX`* |
| 1 |*`CUDNN_POOLING_AVERAGE_COUNT_INCLUDE_PADDING`* | |*`HIPDNN_POOLING_AVERAGE_COUNT_INCLUDE_PADDING`* |
| 2 |*`CUDNN_POOLING_AVERAGE_COUNT_EXCLUDE_PADDING`* | |*`HIPDNN_POOLING_AVERAGE_COUNT_EXCLUDE_PADDING`* |
| 3 |*`CUDNN_POOLING_MAX_DETERMINISTIC`* | 7.5 |*`HIPDNN_POOLING_MAX_DETERMINISTIC`* |
| enum |***`cudnnActivationMode_t`*** | |***`hipdnnActivationMode_t`*** |
| 0 |*`CUDNN_ACTIVATION_SIGMOID`* | |*`HIPDNN_ACTIVATION_SIGMOID`* |
| 1 |*`CUDNN_ACTIVATION_RELU`* | |*`HIPDNN_ACTIVATION_RELU`* |
| 2 |*`CUDNN_ACTIVATION_TANH`* | |*`HIPDNN_ACTIVATION_TANH`* |
| 3 |*`CUDNN_ACTIVATION_CLIPPED_RELU`* | |*`HIPDNN_ACTIVATION_CLIPPED_RELU`* |
| 4 |*`CUDNN_ACTIVATION_ELU`* | 7.5 |*`HIPDNN_ACTIVATION_ELU`* |
| 5 |*`CUDNN_ACTIVATION_IDENTITY`* | 8.0 |*`HIPDNN_ACTIVATION_PATHTRU`* |
| define |`CUDNN_LRN_MIN_N` | | |
| define |`CUDNN_LRN_MAX_N` | | |
| define |`CUDNN_LRN_MIN_K` | | |
| define |`CUDNN_LRN_MIN_BETA` | | |
| enum |***`cudnnLRNMode_t`*** | |***`hipdnnLRNMode_t`*** |
| 0 |*`CUDNN_LRN_CROSS_CHANNEL_DIM1`* | |*`HIPDNN_LRN_CROSS_CHANNEL`* |
| enum |***`cudnnDivNormMode_t`*** | | |
| 0 |*`CUDNN_DIVNORM_PRECOMPUTED_MEANS`* | | |
| enum |***`cudnnBatchNormMode_t`*** | |***`hipdnnBatchNormMode_t`*** |
| 0 |*`CUDNN_BATCHNORM_PER_ACTIVATION`* | |*`HIPDNN_BATCHNORM_PER_ACTIVATION`* |
| 1 |*`CUDNN_BATCHNORM_SPATIAL`* | |*`HIPDNN_BATCHNORM_SPATIAL`* |
| 2 |*`CUDNN_BATCHNORM_SPATIAL_PERSISTENT`* | 8.0 |*`HIPDNN_BATCHNORM_SPATIAL_PERSISTENT`* |
| define |`CUDNN_BN_MIN_EPSILON` | |`HIPDNN_BN_MIN_EPSILON` |
| enum |***`cudnnSamplerType_t`*** | 7.5 | |
| 0 |*`CUDNN_SAMPLER_BILINEAR`* | 7.5 | |
| struct |`cudnnDropoutStruct` | 7.5 | |
| struct* |`cudnnDropoutDescriptor_t` | 7.5 |`hipdnnDropoutDescriptor_t` |
| enum |***`cudnnRNNMode_t`*** | 7.5 |***`hipdnnRNNMode_t`*** |
| 0 |*`CUDNN_RNN_RELU`* | 7.5 |*`HIPDNN_RNN_RELU`* |
| 1 |*`CUDNN_RNN_TANH`* | 7.5 |*`HIPDNN_RNN_TANH`* |
| 2 |*`CUDNN_LSTM`* | 7.5 |*`HIPDNN_LSTM`* |
| 3 |*`CUDNN_GRU`* | 7.5 |*`HIPDNN_GRU`* |
| enum |***`cudnnRNNBiasMode_t`*** | 9.0 |***`hipdnnRNNBiasMode_t`*** |
| 0 |*`CUDNN_RNN_NO_BIAS`* | 9.0 |*`HIPDNN_RNN_NO_BIAS`* |
| 1 |*`CUDNN_RNN_SINGLE_INP_BIAS`* | 9.0 |*`HIPDNN_RNN_WITH_BIAS`* |
| 2 |*`CUDNN_RNN_DOUBLE_BIAS`* | 9.0 |*`HIPDNN_RNN_WITH_BIAS`* | 1 |
| 3 |*`CUDNN_RNN_SINGLE_REC_BIAS`* | 9.0 |*`HIPDNN_RNN_WITH_BIAS`* | 1 |
| enum |***`cudnnDirectionMode_t`*** | 7.5 |***`hipdnnDirectionMode_t`*** |
| 0 |*`CUDNN_UNIDIRECTIONAL`* | 7.5 |*`HIPDNN_UNIDIRECTIONAL`* |
| 1 |*`CUDNN_BIDIRECTIONAL`* | 7.5 |*`HIPDNN_BIDIRECTIONAL`* |
| enum |***`cudnnRNNAlgo_t`*** | 7.5 |***`hipdnnRNNAlgo_t`*** |
| 0 |*`CUDNN_RNN_ALGO_STANDARD`* | 7.5 |*`HIPDNN_RNN_ALGO_STANDARD`* |
| 1 |*`CUDNN_RNN_ALGO_PERSIST_STATIC`* | 7.5 |*`HIPDNN_RNN_ALGO_PERSIST_STATIC`* |
| 2 |*`CUDNN_RNN_ALGO_PERSIST_DYNAMIC`* | 7.5 |*`HIPDNN_RNN_ALGO_PERSIST_DYNAMIC`* |
| 3 |*`CUDNN_RNN_ALGO_COUNT`* | 8.0 | |
| struct |`cudnnAlgorithmStruct` | 8.0 | |
| struct* |`cudnnAlgorithmDescriptor_t` | 8.0 | |
| struct |`cudnnAlgorithmPerformanceStruct` | 8.0 | |
| struct* |`cudnnAlgorithmPerformance_t` | 8.0 | |
| struct |`cudnnRNNStruct` | 7.5 | |
| struct* |`cudnnRNNDescriptor_t` | 7.5 |`hipdnnRNNDescriptor_t` |
| struct |`cudnnPersistentRNNPlan` | 7.5 | |
| struct* |`cudnnPersistentRNNPlan_t` | 7.5 |`hipdnnPersistentRNNPlan_t` |
| enum |***`cudnnCTCLossAlgo_t`*** | 8.0 | |
| 0 |*`CUDNN_CTC_LOSS_ALGO_DETERMINISTIC`* | 8.0 | |
| 1 |*`CUDNN_CTC_LOSS_ALGO_NON_DETERMINISTIC`* | 8.0 | |
| struct |`cudnnAlgorithm_t` | 8.0 | |
| enum |***`cudnnSeverity_t`*** | 8.0 | |
| 0 |*`CUDNN_SEV_FATAL`* | 8.0 | |
| 1 |*`CUDNN_SEV_ERROR`* | 8.0 | |
| 2 |*`CUDNN_SEV_WARNING`* | 8.0 | |
| 3 |*`CUDNN_SEV_INFO`* | 8.0 | |
| define |`CUDNN_SEV_ERROR_EN` | 8.0 | |
| define |`CUDNN_SEV_WARNING_EN` | 8.0 | |
| define |`CUDNN_SEV_INFO_EN` | 8.0 | |
| struct |`cudnnDebug_t` | 8.0 | |
| struct |`cudnnCallback_t` | 8.0 | |
| enum |***`cudnnBatchNormOps_t`*** | 9.0 | |
| 0 |*`CUDNN_BATCHNORM_OPS_BN`* | 9.0 | |
| 1 |*`CUDNN_BATCHNORM_OPS_BN_ACTIVATION`* | 9.0 | |
| 2 |*`CUDNN_BATCHNORM_OPS_BN_ADD_ACTIVATION`* | 9.0 | |
| enum |***`cudnnRNNClipMode_t`*** | 9.0 | |
| 0 |*`CUDNN_RNN_CLIP_NONE`* | 9.0 | |
| 1 |*`CUDNN_RNN_CLIP_MINMAX`* | 9.0 | |
| struct |`cudnnRNNDataStruct` | 9.0 | |
| struct* |`cudnnRNNDataDescriptor_t` | 9.0 | |
| enum |***`cudnnRNNDataLayout_t`*** | 9.0 | |
| 0 |*`CUDNN_RNN_DATA_LAYOUT_SEQ_MAJOR_UNPACKED`* | 9.0 | |
| 1 |*`CUDNN_RNN_DATA_LAYOUT_SEQ_MAJOR_PACKED`* | 9.0 | |
| 2 |*`CUDNN_RNN_DATA_LAYOUT_BATCH_MAJOR_UNPACKED`* | 9.0 | |
| enum |***`cudnnRNNPaddingMode_t`*** | 9.0 | |
| 0 |*`CUDNN_RNN_PADDED_IO_DISABLED`* | 9.0 | |
| 1 |*`CUDNN_RNN_PADDED_IO_ENABLED`* | 9.0 | |
| enum |***`cudnnSeqDataAxis_t`*** | 9.0 | |
| 0 |*`CUDNN_SEQDATA_TIME_DIM`* | 9.0 | |
| 1 |*`CUDNN_SEQDATA_BATCH_DIM`* | 9.0 | |
| 2 |*`CUDNN_SEQDATA_BEAM_DIM`* | 9.0 | |
| 3 |*`CUDNN_SEQDATA_VECT_DIM`* | 9.0 | |
| define |`CUDNN_SEQDATA_DIM_COUNT` | 9.0 | |
| struct |`cudnnSeqDataStruct` | 9.0 | |
| struct* |`cudnnSeqDataDescriptor_t` | 9.0 | |
| unsigned |***`cudnnAttnQueryMap_t`*** | 9.0 | |
| 0 |*`CUDNN_ATTN_QUERYMAP_ALL_TO_ONE`* | 9.0 | |
| 1U << 0 |*`CUDNN_ATTN_QUERYMAP_ONE_TO_ONE`* | 9.0 | |
| 1 |*`CUDNN_ATTN_DISABLE_PROJ_BIASES`* | 10.1 Update 2 | |
| 1U << 1 |*`CUDNN_ATTN_ENABLE_PROJ_BIASES`* | 10.1 Update 2 | |
| struct |`cudnnAttnStruct` | 9.0 | |
| struct* |`cudnnAttnDescriptor_t` | 9.0 | |
| enum |***`cudnnMultiHeadAttnWeightKind_t`*** | 9.0 | |
| 0 |*`CUDNN_MH_ATTN_Q_WEIGHTS`* | 9.0 | |
| 1 |*`CUDNN_MH_ATTN_K_WEIGHTS`* | 9.0 | |
| 2 |*`CUDNN_MH_ATTN_V_WEIGHTS`* | 9.0 | |
| 3 |*`CUDNN_MH_ATTN_O_WEIGHTS`* | 9.0 | |
| 4 |*`CUDNN_MH_ATTN_Q_BIASES`* | 10.1 Update 2 | |
| 5 |*`CUDNN_MH_ATTN_K_BIASES`* | 10.1 Update 2 | |
| 6 |*`CUDNN_MH_ATTN_V_BIASES`* | 10.1 Update 2 | |
| 7 |*`CUDNN_MH_ATTN_O_BIASES`* | 10.1 Update 2 | |
| define 8 |`CUDNN_ATTN_WKIND_COUNT` | 10.1 Update 2 | |
| enum |***`cudnnWgradMode_t`*** | 9.0 | |
| 0 |*`CUDNN_WGRAD_MODE_ADD`* | 9.0 | |
| 1 |*`CUDNN_WGRAD_MODE_SET`* | 9.0 | |
| enum |***`cudnnReorderType_t`*** | 10.1 | |
| 0 |*`CUDNN_DEFAULT_REORDER`* | 10.1 | |
| 1 |*`CUDNN_NO_REORDER`* | 10.1 | |
| enum |***`cudnnLossNormalizationMode_t`*** | 10.1 | |
| 0 |*`CUDNN_LOSS_NORMALIZATION_NONE`* | 10.1 | |
| 1 |*`CUDNN_LOSS_NORMALIZATION_SOFTMAX`* | 10.1 | |
| struct |`cudnnFusedOpsConstParamStruct` | 10.1 | |
| struct* |`cudnnFusedOpsConstParamPack_t` | 10.1 | |
| struct |`cudnnFusedOpsVariantParamStruct` | 10.1 | |
| struct* |`cudnnFusedOpsVariantParamPack_t` | 10.1 | |
| struct |`cudnnFusedOpsPlanStruct` | 10.1 | |
| struct* |`cudnnFusedOpsPlan_t` | 10.1 | |
| enum |***`cudnnFusedOps_t`*** | 10.1 | |
| 0 |*`CUDNN_FUSED_SCALE_BIAS_ACTIVATION_CONV_BNSTATS`* | 10.1 | |
| 1 |*`CUDNN_FUSED_SCALE_BIAS_ACTIVATION_WGRAD`* | 10.1 | |
| 2 |*`CUDNN_FUSED_BN_FINALIZE_STATISTICS_TRAINING`* | 10.1 | |
| 3 |*`CUDNN_FUSED_BN_FINALIZE_STATISTICS_INFERENCE`* | 10.1 | |
| 4 |*`CUDNN_FUSED_CONV_SCALE_BIAS_ADD_ACTIVATION`* | 10.1 | |
| 5 |*`CUDNN_FUSED_SCALE_BIAS_ADD_ACTIVATION_GEN_BITMASK`* | 10.1 | |
| 6 |*`CUDNN_FUSED_DACTIVATION_FORK_DBATCHNORM`* | 10.1 | |
| enum |***`cudnnFusedOpsConstParamLabel_t`*** | 10.1 | |
| 0 |*`CUDNN_PARAM_XDESC`* | 10.1 | |
| 1 |*`CUDNN_PARAM_XDATA_PLACEHOLDER`* | 10.1 | |
| 2 |*`CUDNN_PARAM_BN_MODE`* | 10.1 | |
| 3 |*`CUDNN_PARAM_BN_EQSCALEBIAS_DESC`* | 10.1 | |
| 4 |*`CUDNN_PARAM_BN_EQSCALE_PLACEHOLDER`* | 10.1 | |
| 5 |*`CUDNN_PARAM_BN_EQBIAS_PLACEHOLDER`* | 10.1 | |
| 6 |*`CUDNN_PARAM_ACTIVATION_DESC`* | 10.1 | |
| 7 |*`CUDNN_PARAM_CONV_DESC`* | 10.1 | |
| 8 |*`CUDNN_PARAM_WDESC`* | 10.1 | |
| 9 |*`CUDNN_PARAM_WDATA_PLACEHOLDER`* | 10.1 | |
| 10 |*`CUDNN_PARAM_DWDESC`* | 10.1 | |
| 11 |*`CUDNN_PARAM_DWDATA_PLACEHOLDER`* | 10.1 | |
| 12 |*`CUDNN_PARAM_YDESC`* | 10.1 | |
| 13 |*`CUDNN_PARAM_YDATA_PLACEHOLDER`* | 10.1 | |
| 14 |*`CUDNN_PARAM_DYDESC`* | 10.1 | |
| 15 |*`CUDNN_PARAM_DYDATA_PLACEHOLDER`* | 10.1 | |
| 16 |*`CUDNN_PARAM_YSTATS_DESC`* | 10.1 | |
| 17 |*`CUDNN_PARAM_YSUM_PLACEHOLDER`* | 10.1 | |
| 18 |*`CUDNN_PARAM_YSQSUM_PLACEHOLDER`* | 10.1 | |
| 19 |*`CUDNN_PARAM_BN_SCALEBIAS_MEANVAR_DESC`* | 10.1 | |
| 20 |*`CUDNN_PARAM_BN_SCALE_PLACEHOLDER`* | 10.1 | |
| 21 |*`CUDNN_PARAM_BN_BIAS_PLACEHOLDER`* | 10.1 | |
| 22 |*`CUDNN_PARAM_BN_SAVED_MEAN_PLACEHOLDER`* | 10.1 | |
| 23 |*`CUDNN_PARAM_BN_SAVED_INVSTD_PLACEHOLDER`* | 10.1 | |
| 24 |*`CUDNN_PARAM_BN_RUNNING_MEAN_PLACEHOLDER`* | 10.1 | |
| 25 |*`CUDNN_PARAM_BN_RUNNING_VAR_PLACEHOLDER`* | 10.1 | |
| 26 |*`CUDNN_PARAM_ZDESC`* | 10.1 | |
| 27 |*`CUDNN_PARAM_ZDATA_PLACEHOLDER`* | 10.1 | |
| 28 |*`CUDNN_PARAM_BN_Z_EQSCALEBIAS_DESC`* | 10.1 | |
| 29 |*`CUDNN_PARAM_BN_Z_EQSCALE_PLACEHOLDER`* | 10.1 | |
| 30 |*`CUDNN_PARAM_BN_Z_EQBIAS_PLACEHOLDER`* | 10.1 | |
| 31 |*`CUDNN_PARAM_ACTIVATION_BITMASK_DESC`* | 10.1 | |
| 32 |*`CUDNN_PARAM_ACTIVATION_BITMASK_PLACEHOLDER`* | 10.1 | |
| 33 |*`CUDNN_PARAM_DXDESC`* | 10.1 | |
| 34 |*`CUDNN_PARAM_DXDATA_PLACEHOLDER`* | 10.1 | |
| 35 |*`CUDNN_PARAM_DZDESC`* | 10.1 | |
| 36 |*`CUDNN_PARAM_DZDATA_PLACEHOLDER`* | 10.1 | |
| 37 |*`CUDNN_PARAM_BN_DSCALE_PLACEHOLDER`* | 10.1 | |
| 38 |*`CUDNN_PARAM_BN_DBIAS_PLACEHOLDER`* | 10.1 | |
| enum |***`cudnnFusedOpsPointerPlaceHolder_t`*** | 10.1 | |
| 0 |*`CUDNN_PTR_NULL`* | 10.1 | |
| 1 |*`CUDNN_PTR_ELEM_ALIGNED`* | 10.1 | |
| 2 |*`CUDNN_PTR_16B_ALIGNED`* | 10.1 | |
| enum |***`cudnnFusedOpsVariantParamLabel_t`*** | 10.1 | |
| 0 |*`CUDNN_PTR_XDATA`* | 10.1 | |
| 1 |*`CUDNN_PTR_BN_EQSCALE`* | 10.1 | |
| 2 |*`CUDNN_PTR_BN_EQBIAS`* | 10.1 | |
| 3 |*`CUDNN_PTR_WDATA`* | 10.1 | |
| 4 |*`CUDNN_PTR_DWDATA`* | 10.1 | |
| 5 |*`CUDNN_PTR_YDATA`* | 10.1 | |
| 6 |*`CUDNN_PTR_DYDATA`* | 10.1 | |
| 7 |*`CUDNN_PTR_YSUM`* | 10.1 | |
| 8 |*`CUDNN_PTR_YSQSUM`* | 10.1 | |
| 9 |*`CUDNN_PTR_WORKSPACE`* | 10.1 | |
| 10 |*`CUDNN_PTR_BN_SCALE`* | 10.1 | |
| 11 |*`CUDNN_PTR_BN_BIAS`* | 10.1 | |
| 12 |*`CUDNN_PTR_BN_SAVED_MEAN`* | 10.1 | |
| 13 |*`CUDNN_PTR_BN_SAVED_INVSTD`* | 10.1 | |
| 14 |*`CUDNN_PTR_BN_RUNNING_MEAN`* | 10.1 | |
| 15 |*`CUDNN_PTR_BN_RUNNING_VAR`* | 10.1 | |
| 16 |*`CUDNN_PTR_ZDATA`* | 10.1 | |
| 17 |*`CUDNN_PTR_BN_Z_EQSCALE`* | 10.1 | |
| 18 |*`CUDNN_PTR_BN_Z_EQBIAS`* | 10.1 | |
| 19 |*`CUDNN_PTR_ACTIVATION_BITMASK`* | 10.1 | |
| 20 |*`CUDNN_PTR_DXDATA`* | 10.1 | |
| 21 |*`CUDNN_PTR_DZDATA`* | 10.1 | |
| 22 |*`CUDNN_PTR_BN_DSCALE`* | 10.1 | |
| 23 |*`CUDNN_PTR_BN_DBIAS`* | 10.1 | |
| 100 |*`CUDNN_SCALAR_SIZE_T_WORKSPACE_SIZE_IN_BYTES`* | 10.1 | |
| 101 |*`CUDNN_SCALAR_INT64_T_BN_ACCUMULATION_COUNT`* | 10.1 | |
| 102 |*`CUDNN_SCALAR_DOUBLE_BN_EXP_AVG_FACTOR`* | 10.1 | |
| 103 |*`CUDNN_SCALAR_DOUBLE_BN_EPSILON`* | 10.1 | |
## **2. CUDNN API functions**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:-----------------|
|`cudnnGetVersion` |`hipdnnGetVersion` |
|`cudnnGetCudartVersion` | | 7.5 |
|`cudnnGetErrorString` |`hipdnnGetErrorString` |
|`cudnnQueryRuntimeError` | | 8.0 |
|`cudnnGetProperty` | | 7.5 |
|`cudnnCreate` |`hipdnnCreate` |
|`cudnnDestroy` |`hipdnnDestroy` |
|`cudnnSetStream` |`hipdnnSetStream` |
|`cudnnGetStream` |`hipdnnGetStream` |
|`cudnnCreateTensorDescriptor` |`hipdnnCreateTensorDescriptor` |
|`cudnnSetTensor4dDescriptor` |`hipdnnSetTensor4dDescriptor` |
|`cudnnSetTensor4dDescriptorEx` |`hipdnnSetTensor4dDescriptorEx` |
|`cudnnGetTensor4dDescriptor` |`hipdnnGetTensor4dDescriptor` |
|`cudnnSetTensorNdDescriptor` |`hipdnnSetTensorNdDescriptor` |
|`cudnnSetTensorNdDescriptorEx` | | 7.5 |
|`cudnnGetTensorNdDescriptor` |`hipdnnGetTensorNdDescriptor` |
|`cudnnGetTensorSizeInBytes` | | 7.5 |
|`cudnnDestroyTensorDescriptor` |`hipdnnDestroyTensorDescriptor` |
|`cudnnTransformTensor` | |
|`cudnnTransformTensorEx` | | 9.0 |
|`cudnnInitTransformDest` | | 9.0 |
|`cudnnCreateTensorTransformDescriptor` | | 9.0 |
|`cudnnSetTensorTransformDescriptor` | | 9.0 |
|`cudnnGetTensorTransformDescriptor` | | 9.0 |
|`cudnnDestroyTensorTransformDescriptor` | | 9.0 |
|`cudnnAddTensor` |`hipdnnAddTensor` |
|`cudnnCreateOpTensorDescriptor` |`hipdnnCreateOpTensorDescriptor` | 7.5 |
|`cudnnSetOpTensorDescriptor` |`hipdnnSetOpTensorDescriptor` | 7.5 |
|`cudnnGetOpTensorDescriptor` |`hipdnnGetOpTensorDescriptor` | 7.5 |
|`cudnnDestroyOpTensorDescriptor` |`hipdnnDestroyOpTensorDescriptor` | 7.5 |
|`cudnnOpTensor` |`hipdnnOpTensor` | 7.5 |
|`cudnnGetFoldedConvBackwardDataDescriptors` | | 10.1 |
|`cudnnCreateReduceTensorDescriptor` |`hipdnnCreateReduceTensorDescriptor` | 7.5 |
|`cudnnSetReduceTensorDescriptor` |`hipdnnSetReduceTensorDescriptor` | 7.5 |
|`cudnnGetReduceTensorDescriptor` |`hipdnnGetReduceTensorDescriptor` | 7.5 |
|`cudnnDestroyReduceTensorDescriptor` |`hipdnnDestroyReduceTensorDescriptor` | 7.5 |
|`cudnnGetReductionIndicesSize` | | 7.5 |
|`cudnnGetReductionWorkspaceSize` |`hipdnnGetReductionWorkspaceSize` | 7.5 |
|`cudnnReduceTensor` |`hipdnnReduceTensor` | 7.5 |
|`cudnnSetTensor` |`hipdnnSetTensor` |
|`cudnnScaleTensor` |`hipdnnScaleTensor` |
|`cudnnCreateFilterDescriptor` |`hipdnnCreateFilterDescriptor` |
|`cudnnSetFilter4dDescriptor` |`hipdnnSetFilter4dDescriptor` |
|`cudnnGetFilter4dDescriptor` |`hipdnnGetFilter4dDescriptor` |
|`cudnnSetFilterNdDescriptor` |`hipdnnSetFilterNdDescriptor` |
|`cudnnGetFilterNdDescriptor` |`hipdnnGetFilterNdDescriptor` |
|`cudnnGetFilterSizeInBytes` | | 10.1 |
|`cudnnTransformFilter` | | 10.1 |
|`cudnnDestroyFilterDescriptor` |`hipdnnDestroyFilterDescriptor` |
|`cudnnReorderFilterAndBias` | | 10.1 |
|`cudnnCreateConvolutionDescriptor` |`hipdnnCreateConvolutionDescriptor` |
|`cudnnSetConvolutionMathType` |`hipdnnSetConvolutionMathType` | 8.0 |
|`cudnnGetConvolutionMathType` | | 8.0 |
|`cudnnSetConvolutionGroupCount` |`hipdnnSetConvolutionGroupCount` | 8.0 |
|`cudnnGetConvolutionGroupCount` | | 8.0 |
|`cudnnSetConvolutionReorderType` | | 10.1 |
|`cudnnGetConvolutionReorderType` | | 10.1 |
|`cudnnSetConvolution2dDescriptor` |`hipdnnSetConvolution2dDescriptor` |
|`cudnnGetConvolution2dDescriptor` |`hipdnnGetConvolution2dDescriptor` |
|`cudnnGetConvolution2dForwardOutputDim` |`hipdnnGetConvolution2dForwardOutputDim` |
|`cudnnSetConvolutionNdDescriptor` |`hipdnnSetConvolutionNdDescriptor` |
|`cudnnGetConvolutionNdDescriptor` | |
|`cudnnGetConvolutionNdForwardOutputDim` | |
|`cudnnDestroyConvolutionDescriptor` | |
|`cudnnGetConvolutionForwardAlgorithmMaxCount` | | 8.0 |
|`cudnnFindConvolutionForwardAlgorithm` |`hipdnnFindConvolutionForwardAlgorithm` |
|`cudnnFindConvolutionForwardAlgorithmEx` |`hipdnnFindConvolutionForwardAlgorithmEx` | 7.5 |
|`cudnnGetConvolutionForwardAlgorithm` |`hipdnnGetConvolutionForwardAlgorithm` |
|`cudnnGetConvolutionForwardAlgorithm_v7` | | 8.0 |
|`cudnnGetConvolutionForwardWorkspaceSize` |`hipdnnGetConvolutionForwardWorkspaceSize` |
|`cudnnConvolutionForward` |`hipdnnConvolutionForward` |
|`cudnnConvolutionBiasActivationForward` | | 7.5 |
|`cudnnConvolutionBackwardBias` |`hipdnnConvolutionBackwardBias` |
|`cudnnGetConvolutionBackwardFilterAlgorithmMaxCount` | | 8.0 |
|`cudnnFindConvolutionBackwardFilterAlgorithm` |`hipdnnFindConvolutionBackwardFilterAlgorithm` |
|`cudnnFindConvolutionBackwardFilterAlgorithmEx` |`hipdnnFindConvolutionBackwardFilterAlgorithmEx` | 7.5 |
|`cudnnGetConvolutionBackwardFilterAlgorithm` |`hipdnnGetConvolutionBackwardFilterAlgorithm` |
|`cudnnGetConvolutionBackwardFilterAlgorithm_v7` | | 8.0 |
|`cudnnGetConvolutionBackwardFilterWorkspaceSize` |`hipdnnGetConvolutionBackwardFilterWorkspaceSize`|
|`cudnnConvolutionBackwardFilter` |`hipdnnConvolutionBackwardFilter` |
|`cudnnGetConvolutionBackwardDataAlgorithmMaxCount` | | 8.0 |
|`cudnnFindConvolutionBackwardDataAlgorithm` |`hipdnnFindConvolutionBackwardDataAlgorithm` |
|`cudnnFindConvolutionBackwardDataAlgorithmEx` |`hipdnnFindConvolutionBackwardDataAlgorithmEx` | 7.5 |
|`cudnnGetConvolutionBackwardDataAlgorithm` |`hipdnnGetConvolutionBackwardDataAlgorithm` |
|`cudnnGetConvolutionBackwardDataAlgorithm_v7` | | 8.0 |
|`cudnnGetConvolutionBackwardDataWorkspaceSize` |`hipdnnGetConvolutionBackwardDataWorkspaceSize` |
|`cudnnConvolutionBackwardData` |`hipdnnConvolutionBackwardData` |
|`cudnnIm2Col` | |
|`cudnnSoftmaxForward` |`hipdnnSoftmaxForward` |
|`cudnnSoftmaxBackward` |`hipdnnSoftmaxBackward` |
|`cudnnCreatePoolingDescriptor` |`hipdnnCreatePoolingDescriptor` |
|`cudnnSetPooling2dDescriptor` |`hipdnnSetPooling2dDescriptor` |
|`cudnnGetPooling2dDescriptor` |`hipdnnGetPooling2dDescriptor` |
|`cudnnSetPoolingNdDescriptor` |`hipdnnSetPoolingNdDescriptor` |
|`cudnnGetPoolingNdDescriptor` | |
|`cudnnGetPoolingNdForwardOutputDim` | |
|`cudnnGetPooling2dForwardOutputDim` |`hipdnnGetPooling2dForwardOutputDim` |
|`cudnnDestroyPoolingDescriptor` |`hipdnnDestroyPoolingDescriptor` |
|`cudnnPoolingForward` |`hipdnnPoolingForward` |
|`cudnnPoolingBackward` |`hipdnnPoolingBackward` |
|`cudnnCreateActivationDescriptor` |`hipdnnCreateActivationDescriptor` |
|`cudnnSetActivationDescriptor` |`hipdnnSetActivationDescriptor` |
|`cudnnGetActivationDescriptor` |`hipdnnGetActivationDescriptor` |
|`cudnnDestroyActivationDescriptor` |`hipdnnDestroyActivationDescriptor` |
|`cudnnActivationForward` |`hipdnnActivationForward` |
|`cudnnActivationBackward` |`hipdnnActivationBackward` |
|`cudnnCreateLRNDescriptor` |`hipdnnCreateLRNDescriptor` |
|`cudnnSetLRNDescriptor` |`hipdnnSetLRNDescriptor` |
|`cudnnGetLRNDescriptor` |`hipdnnGetLRNDescriptor` |
|`cudnnDestroyLRNDescriptor` |`hipdnnDestroyLRNDescriptor` |
|`cudnnLRNCrossChannelForward` |`hipdnnLRNCrossChannelForward` |
|`cudnnLRNCrossChannelBackward` |`hipdnnLRNCrossChannelBackward` |
|`cudnnDivisiveNormalizationForward` | |
|`cudnnDivisiveNormalizationBackward` | |
|`cudnnDeriveBNTensorDescriptor` |`hipdnnDeriveBNTensorDescriptor` |
|`cudnnBatchNormalizationForwardTraining` |`hipdnnBatchNormalizationForwardTraining` |
|`cudnnBatchNormalizationForwardTrainingEx` | | 9.0 |
|`cudnnBatchNormalizationForwardInference` |`hipdnnBatchNormalizationForwardInference` |
|`cudnnBatchNormalizationBackward` |`hipdnnBatchNormalizationBackward` |
|`cudnnBatchNormalizationBackwardEx` | | 9.0 |
|`cudnnCreateSpatialTransformerDescriptor` | | 7.5 |
|`cudnnSetSpatialTransformerNdDescriptor` | | 7.5 |
|`cudnnDestroySpatialTransformerDescriptor` | | 7.5 |
|`cudnnSpatialTfGridGeneratorForward` | | 7.5 |
|`cudnnSpatialTfGridGeneratorBackward` | | 7.5 |
|`cudnnSpatialTfSamplerForward` | | 7.5 |
|`cudnnSpatialTfSamplerBackward` | | 7.5 |
|`cudnnCreateDropoutDescriptor` |`hipdnnCreateDropoutDescriptor` | 7.5 |
|`cudnnDestroyDropoutDescriptor` |`hipdnnDestroyDropoutDescriptor` | 7.5 |
|`cudnnDropoutGetStatesSize` |`hipdnnDropoutGetStatesSize` | 7.5 |
|`cudnnDropoutGetReserveSpaceSize` | | 7.5 |
|`cudnnSetDropoutDescriptor` |`hipdnnSetDropoutDescriptor` | 7.5 |
|`cudnnGetDropoutDescriptor` | | 8.0 |
|`cudnnRestoreDropoutDescriptor` | | 8.0 |
|`cudnnDropoutForward` | | 7.5 |
|`cudnnDropoutBackward` | | 7.5 |
|`cudnnCreateRNNDescriptor` |`hipdnnCreateRNNDescriptor` | 7.5 |
|`cudnnDestroyRNNDescriptor` |`hipdnnDestroyRNNDescriptor` | 7.5 |
|`cudnnGetRNNForwardInferenceAlgorithmMaxCount` | | 8.0 |
|`cudnnFindRNNForwardInferenceAlgorithmEx` | | 8.0 |
|`cudnnGetRNNForwardTrainingAlgorithmMaxCount` | | 8.0 |
|`cudnnFindRNNForwardTrainingAlgorithmEx` | | 8.0 |
|`cudnnGetRNNBackwardDataAlgorithmMaxCount` | | 8.0 |
|`cudnnFindRNNBackwardDataAlgorithmEx` | | 8.0 |
|`cudnnGetRNNBackwardWeightsAlgorithmMaxCount` | | 8.0 |
|`cudnnFindRNNBackwardWeightsAlgorithmEx` | | 8.0 |
|`cudnnCreatePersistentRNNPlan` |`hipdnnCreatePersistentRNNPlan` | 7.5 |
|`cudnnSetPersistentRNNPlan` |`hipdnnSetPersistentRNNPlan` | 7.5 |
|`cudnnDestroyPersistentRNNPlan` |`hipdnnDestroyPersistentRNNPlan` | 7.5 |
|`cudnnSetRNNDescriptor` |`hipdnnSetRNNDescriptor` | 7.5 |
|`cudnnGetRNNDescriptor` |`hipdnnGetRNNDescriptor` | 8.0 |
|`cudnnSetRNNProjectionLayers` | | 8.0 |
|`cudnnGetRNNProjectionLayers` | | 8.0 |
|`cudnnSetRNNAlgorithmDescriptor` | | 8.0 |
|`cudnnSetRNNMatrixMathType` | | 8.0 |
|`cudnnGetRNNMatrixMathType` | | 8.0 |
|`cudnnGetRNNWorkspaceSize` |`hipdnnGetRNNWorkspaceSize` | 7.5 |
|`cudnnGetRNNTrainingReserveSize` |`hipdnnGetRNNTrainingReserveSize` | 7.5 |
|`cudnnGetRNNParamsSize` |`hipdnnGetRNNParamsSize` | 7.5 |
|`cudnnGetRNNLinLayerMatrixParams` |`hipdnnGetRNNLinLayerMatrixParams` | 7.5 |
|`cudnnGetRNNLinLayerBiasParams` |`hipdnnGetRNNLinLayerBiasParams` | 7.5 |
|`cudnnRNNForwardInference` |`hipdnnRNNForwardInference` | 7.5 |
|`cudnnRNNForwardInferenceEx` | | 9.0 |
|`cudnnRNNForwardTraining` |`hipdnnRNNForwardTraining` | 7.5 |
|`cudnnRNNForwardTrainingEx` | | 9.0 |
|`cudnnRNNBackwardData` |`hipdnnRNNBackwardData` | 7.5 |
|`cudnnRNNBackwardDataEx` | | 9.0 |
|`cudnnRNNBackwardWeights` |`hipdnnRNNBackwardWeights` | 7.5 |
|`cudnnRNNBackwardWeightsEx` | | 9.0 |
|`cudnnSetRNNPaddingMode` | | 9.0 |
|`cudnnGetRNNPaddingMode` | | 9.0 |
|`cudnnCreateRNNDataDescriptor` | | 9.0 |
|`cudnnDestroyRNNDataDescriptor` | | 9.0 |
|`cudnnSetRNNDataDescriptor` | | 9.0 |
|`cudnnGetRNNDataDescriptor` | | 9.0 |
|`cudnnSetRNNBiasMode` | | 9.0 |
|`cudnnGetRNNBiasMode` | | 9.0 |
|`cudnnCreateCTCLossDescriptor` | | 8.0 |
|`cudnnSetCTCLossDescriptor` | | 8.0 |
|`cudnnSetCTCLossDescriptorEx` | | 10.1 |
|`cudnnGetCTCLossDescriptor` | | 8.0 |
|`cudnnGetCTCLossDescriptorEx` | | 10.1 |
|`cudnnDestroyCTCLossDescriptor` | | 8.0 |
|`cudnnCTCLoss` | | 8.0 |
|`cudnnGetCTCLossWorkspaceSize` | | 8.0 |
|`cudnnCreateAlgorithmDescriptor` | | 8.0 |
|`cudnnSetAlgorithmDescriptor` | | 8.0 |
|`cudnnGetAlgorithmDescriptor` | | 8.0 |
|`cudnnCopyAlgorithmDescriptor` | | 8.0 |
|`cudnnDestroyAlgorithmDescriptor` | | 8.0 |
|`cudnnCreateAlgorithmPerformance` | | 8.0 |
|`cudnnSetAlgorithmPerformance` | | 8.0 |
|`cudnnGetAlgorithmPerformance` | | 8.0 |
|`cudnnDestroyAlgorithmPerformance` | | 8.0 |
|`cudnnGetAlgorithmSpaceSize` | | 8.0 |
|`cudnnSaveAlgorithm` | | 8.0 |
|`cudnnRestoreAlgorithm` | | 8.0 |
|`cudnnSetRNNDescriptor_v5` |`hipdnnSetRNNDescriptor_v5` | 8.0 |
|`cudnnSetRNNDescriptor_v6` |`hipdnnSetRNNDescriptor_v6` | 7.5 |
|`cudnnSetCallback` | | 8.0 |
|`cudnnGetCallback` | | 8.0 |
|`cudnnGetBatchNormalizationForwardTrainingExWorkspaceSize` | | 9.0 |
|`cudnnGetBatchNormalizationBackwardExWorkspaceSize` | | 9.0 |
|`cudnnGetBatchNormalizationTrainingExReserveSpaceSize` | | 9.0 |
|`cudnnRNNSetClip` | | 9.0 |
|`cudnnRNNGetClip` | | 9.0 |
|`cudnnCreateSeqDataDescriptor` | | 9.0 |
|`cudnnDestroySeqDataDescriptor` | | 9.0 |
|`cudnnSetSeqDataDescriptor` | | 9.0 |
|`cudnnGetSeqDataDescriptor` | | 9.0 |
|`cudnnCreateAttnDescriptor` | | 9.0 |
|`cudnnDestroyAttnDescriptor` | | 9.0 |
|`cudnnSetAttnDescriptor` | | 9.0 |
|`cudnnGetAttnDescriptor` | | 9.0 |
|`cudnnGetMultiHeadAttnBuffers` | | 9.0 |
|`cudnnGetMultiHeadAttnWeights` | | 9.0 |
|`cudnnMultiHeadAttnForward` | | 9.0 |
|`cudnnMultiHeadAttnBackwardData` | | 9.0 |
|`cudnnMultiHeadAttnBackwardWeights` | | 9.0 |
|`cudnnCreateFusedOpsConstParamPack` | | 10.1 |
|`cudnnDestroyFusedOpsConstParamPack` | | 10.1 |
|`cudnnSetFusedOpsConstParamPackAttribute` | | 10.1 |
|`cudnnGetFusedOpsConstParamPackAttribute` | | 10.1 |
|`cudnnCreateFusedOpsVariantParamPack` | | 10.1 |
|`cudnnDestroyFusedOpsVariantParamPack` | | 10.1 |
|`cudnnSetFusedOpsVariantParamPackAttribute` | | 10.1 |
|`cudnnGetFusedOpsVariantParamPackAttribute` | | 10.1 |
|`cudnnCreateFusedOpsPlan` | | 10.1 |
|`cudnnDestroyFusedOpsPlan` | | 10.1 |
|`cudnnMakeFusedOpsPlan` | | 10.1 |
|`cudnnFusedOpsExecute` | | 10.1 |
\* CUDA version, in which API has appeared and (optional) last version before abandoning it; no value in case of earlier versions < 7.5.
docs/markdown/CUFFT_API_supported_by_HIP.md
-81
View File
@@ -1,81 +0,0 @@
# CUFFT API supported by HIP
## **1. CUFFT Data types**
| **type** | **CUDA** |**CUDA version\***| **HIP** |**HIP value** (if differs) |
|-------------:|---------------------------------------------------------------|:----------------:|------------------------------------------------------------|---------------------------|
| enum |***`cufftResult_t`*** | |***`hipfftResult_t`*** |
| enum |***`cufftResult`*** | |***`hipfftResult`*** |
| 0x0 |*`CUFFT_SUCCESS`* | |*`HIPFFT_SUCCESS`* | 0 |
| 0x1 |*`CUFFT_INVALID_PLAN`* | |*`HIPFFT_INVALID_PLAN`* | 1 |
| 0x2 |*`CUFFT_ALLOC_FAILED`* | |*`HIPFFT_ALLOC_FAILED`* | 2 |
| 0x3 |*`CUFFT_INVALID_TYPE`* | |*`HIPFFT_INVALID_TYPE`* | 3 |
| 0x4 |*`CUFFT_INVALID_VALUE`* | |*`HIPFFT_INVALID_VALUE`* | 4 |
| 0x5 |*`CUFFT_INTERNAL_ERROR`* | |*`HIPFFT_INTERNAL_ERROR`* | 5 |
| 0x6 |*`CUFFT_EXEC_FAILED`* | |*`HIPFFT_EXEC_FAILED`* | 6 |
| 0x7 |*`CUFFT_SETUP_FAILED`* | |*`HIPFFT_SETUP_FAILED`* | 7 |
| 0x8 |*`CUFFT_INVALID_SIZE`* | |*`HIPFFT_INVALID_SIZE`* | 8 |
| 0x9 |*`CUFFT_UNALIGNED_DATA`* | |*`HIPFFT_UNALIGNED_DATA`* | 9 |
| 0xA |*`CUFFT_INCOMPLETE_PARAMETER_LIST`* | |*`HIPFFT_INCOMPLETE_PARAMETER_LIST`* | 10 |
| 0xB |*`CUFFT_INVALID_DEVICE`* | |*`HIPFFT_INVALID_DEVICE`* | 11 |
| 0xC |*`CUFFT_PARSE_ERROR`* | |*`HIPFFT_PARSE_ERROR`* | 12 |
| 0xD |*`CUFFT_NO_WORKSPACE`* | |*`HIPFFT_NO_WORKSPACE`* | 13 |
| 0xE |*`CUFFT_NOT_IMPLEMENTED`* | |*`HIPFFT_NOT_IMPLEMENTED`* | 14 |
| 0xF |*`CUFFT_LICENSE_ERROR`* | | |
| 0x10 |*`CUFFT_NOT_SUPPORTED`* | 8.0 |*`HIPFFT_NOT_SUPPORTED`* | 16 |
| float |***`cufftReal`*** | |***`hipfftReal`*** |
| double |***`cufftDoubleReal`*** | |***`hipfftDoubleReal`*** |
| float2 |***`cufftComplex`*** | |***`hipfftComplex`*** |
| double2 |***`cufftDoubleComplex`*** | |***`hipfftDoubleComplex`*** |
| define |`CUFFT_FORWARD` | |`HIPFFT_FORWARD` |
| define |`CUFFT_INVERSE` | |`HIPFFT_BACKWARD` |
| enum |***`cufftType_t`*** | |***`hipfftType_t`*** |
| enum |***`cufftType`*** | |***`hipfftType`*** |
| 0x2a |*`CUFFT_R2C`* | |*`HIPFFT_R2C`* |
| 0x2c |*`CUFFT_C2R`* | |*`HIPFFT_C2R`* |
| 0x29 |*`CUFFT_C2C`* | |*`HIPFFT_C2C`* |
| 0x6a |*`CUFFT_D2Z`* | |*`HIPFFT_D2Z`* |
| 0x6c |*`CUFFT_Z2D`* | |*`HIPFFT_Z2D`* |
| 0x69 |*`CUFFT_Z2Z`* | |*`HIPFFT_Z2Z`* |
| enum |***`cufftCompatibility_t`*** | | |
| enum |***`cufftCompatibility`*** | | |
| 0x01 |*`CUFFT_COMPATIBILITY_FFTW_PADDING`* | | |
| define |`CUFFT_COMPATIBILITY_DEFAULT` | | |
| int |***`cufftHandle`*** | |***`hipfftHandle`*** |
## **2. CUFFT API functions**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cufftPlan1d` |`hipfftPlan1d` |
|`cufftPlan2d` |`hipfftPlan2d` |
|`cufftPlan3d` |`hipfftPlan3d` |
|`cufftPlanMany` |`hipfftPlanMany` |
|`cufftMakePlan1d` |`hipfftMakePlan1d` |
|`cufftMakePlan2d` |`hipfftMakePlan2d` |
|`cufftMakePlan3d` |`hipfftMakePlan3d` |
|`cufftMakePlanMany` |`hipfftMakePlanMany` |
|`cufftMakePlanMany64` |`hipfftMakePlanMany64` | 7.5 |
|`cufftGetSizeMany64` |`hipfftGetSizeMany64` | 7.5 |
|`cufftEstimate1d` |`hipfftEstimate1d` |
|`cufftEstimate2d` |`hipfftEstimate2d` |
|`cufftEstimate3d` |`hipfftEstimate3d` |
|`cufftEstimateMany` |`hipfftEstimateMany` |
|`cufftCreate` |`hipfftCreate` |
|`cufftGetSize1d` |`hipfftGetSize1d` |
|`cufftGetSize2d` |`hipfftGetSize2d` |
|`cufftGetSize3d` |`hipfftGetSize3d` |
|`cufftGetSizeMany` |`hipfftGetSizeMany` |
|`cufftGetSize` |`hipfftGetSize` |
|`cufftSetWorkArea` |`hipfftSetWorkArea` |
|`cufftSetAutoAllocation` |`hipfftSetAutoAllocation` |
|`cufftExecC2C` |`hipfftExecC2C` |
|`cufftExecR2C` |`hipfftExecR2C` |
|`cufftExecC2R` |`hipfftExecC2R` |
|`cufftExecZ2Z` |`hipfftExecZ2Z` |
|`cufftExecD2Z` |`hipfftExecD2Z` |
|`cufftExecZ2D` |`hipfftExecZ2D` |
|`cufftSetStream` |`hipfftSetStream` |
|`cufftDestroy` |`hipfftDestroy` |
|`cufftGetVersion` |`hipfftGetVersion` |
|`cufftGetProperty` | | 8.0 |
docs/markdown/CURAND_API_supported_by_HIP.md
-172
View File
@@ -1,172 +0,0 @@
# CURAND API supported by HIP
## **1. CURAND Data types**
| **type** | **CUDA** |**CUDA version\***| **HIP** | **HIP value** (if differs) |
|-------------:|---------------------------------------------------------------|:----------------:|------------------------------------------------------------|----------------------------|
| define |`CURAND_VER_MAJOR` | 10.1 Update 2 | |
| define |`CURAND_VER_MINOR` | 10.1 Update 2 | |
| define |`CURAND_VER_PATCH` | 10.1 Update 2 | |
| define |`CURAND_VER_BUILD` | 10.1 Update 2 | |
| define |`CURAND_VERSION` | 10.1 Update 2 | |
| enum |***`curandStatus`*** | |***`hiprandStatus`*** |
| enum |***`curandStatus_t`*** | |***`hiprandStatus_t`*** |
| 0 |*`CURAND_STATUS_SUCCESS`* | |*`HIPRAND_STATUS_SUCCESS`* |
| 100 |*`CURAND_STATUS_VERSION_MISMATCH`* | |*`HIPRAND_STATUS_VERSION_MISMATCH`* |
| 101 |*`CURAND_STATUS_NOT_INITIALIZED`* | |*`HIPRAND_STATUS_NOT_INITIALIZED`* |
| 102 |*`CURAND_STATUS_ALLOCATION_FAILED`* | |*`HIPRAND_STATUS_ALLOCATION_FAILED`* |
| 103 |*`CURAND_STATUS_TYPE_ERROR`* | |*`HIPRAND_STATUS_TYPE_ERROR`* |
| 104 |*`CURAND_STATUS_OUT_OF_RANGE`* | |*`HIPRAND_STATUS_OUT_OF_RANGE`* |
| 105 |*`CURAND_STATUS_LENGTH_NOT_MULTIPLE`* | |*`HIPRAND_STATUS_LENGTH_NOT_MULTIPLE`* |
| 106 |*`CURAND_STATUS_DOUBLE_PRECISION_REQUIRED`* | |*`HIPRAND_STATUS_DOUBLE_PRECISION_REQUIRED`* |
| 201 |*`CURAND_STATUS_LAUNCH_FAILURE`* | |*`HIPRAND_STATUS_LAUNCH_FAILURE`* |
| 202 |*`CURAND_STATUS_PREEXISTING_FAILURE`* | |*`HIPRAND_STATUS_PREEXISTING_FAILURE`* |
| 203 |*`CURAND_STATUS_INITIALIZATION_FAILED`* | |*`HIPRAND_STATUS_INITIALIZATION_FAILED`* |
| 204 |*`CURAND_STATUS_ARCH_MISMATCH`* | |*`HIPRAND_STATUS_ARCH_MISMATCH`* |
| 999 |*`CURAND_STATUS_INTERNAL_ERROR`* | |*`HIPRAND_STATUS_INTERNAL_ERROR`* |
| enum |***`curandRngType`*** | |***`hiprandRngType`*** |
| enum |***`curandRngType_t`*** | |***`hiprandRngType_t`*** |
| 0 |*`CURAND_RNG_TEST`* | |*`HIPRAND_RNG_TEST`* |
| 100 |*`CURAND_RNG_PSEUDO_DEFAULT`* | |*`HIPRAND_RNG_PSEUDO_DEFAULT`* | 400 |
| 101 |*`CURAND_RNG_PSEUDO_XORWOW`* | |*`HIPRAND_RNG_PSEUDO_XORWOW`* | 401 |
| 121 |*`CURAND_RNG_PSEUDO_MRG32K3A`* | |*`HIPRAND_RNG_PSEUDO_MRG32K3A`* | 402 |
| 141 |*`CURAND_RNG_PSEUDO_MTGP32`* | |*`HIPRAND_RNG_PSEUDO_MTGP32`* | 403 |
| 142 |*`CURAND_RNG_PSEUDO_MT19937`* | |*`HIPRAND_RNG_PSEUDO_MT19937`* | 404 |
| 161 |*`CURAND_RNG_PSEUDO_PHILOX4_32_10`* | |*`HIPRAND_RNG_PSEUDO_PHILOX4_32_10`* | 405 |
| 200 |*`CURAND_RNG_QUASI_DEFAULT`* | |*`HIPRAND_RNG_QUASI_DEFAULT`* | 500 |
| 201 |*`CURAND_RNG_QUASI_SOBOL32`* | |*`HIPRAND_RNG_QUASI_SOBOL32`* | 501 |
| 202 |*`CURAND_RNG_QUASI_SCRAMBLED_SOBOL32`* | |*`HIPRAND_RNG_QUASI_SCRAMBLED_SOBOL32`* | 502 |
| 203 |*`CURAND_RNG_QUASI_SOBOL64`* | |*`HIPRAND_RNG_QUASI_SOBOL64`* | 503 |
| 204 |*`CURAND_RNG_QUASI_SCRAMBLED_SOBOL64`* | |*`HIPRAND_RNG_QUASI_SCRAMBLED_SOBOL64`* | 504 |
| enum |***`curandOrdering`*** | | |
| enum |***`curandOrdering_t`*** | | |
| 100 |*`CURAND_ORDERING_PSEUDO_BEST`* | | |
| 101 |*`CURAND_ORDERING_PSEUDO_DEFAULT`* | | |
| 102 |*`CURAND_ORDERING_PSEUDO_SEEDED`* | | |
| 201 |*`CURAND_ORDERING_QUASI_DEFAULT`* | | |
| enum |***`curandDirectionVectorSet`*** | | |
| enum |***`curandDirectionVectorSet_t`*** | | |
| 101 |*`CURAND_DIRECTION_VECTORS_32_JOEKUO6`* | | |
| 102 |*`CURAND_SCRAMBLED_DIRECTION_VECTORS_32_JOEKUO6`* | | |
| 103 |*`CURAND_DIRECTION_VECTORS_64_JOEKUO6`* | | |
| 104 |*`CURAND_SCRAMBLED_DIRECTION_VECTORS_64_JOEKUO6`* | | |
| uint | `curandDirectionVectors32_t` | | `hiprandDirectionVectors32_t` |
| uint | `curandDirectionVectors64_t` | | |
| struct | `curandGenerator_st` | | `hiprandGenerator_st` |
| struct* | `curandGenerator_t` | | `hiprandGenerator_t` |
| double | `curandDistribution_st` | | |
| double | `curandHistogramM2V_st` | | |
| double* | `curandDistribution_t` | | |
| double* | `curandHistogramM2V_t` | | |
| struct | `curandDistributionShift_st` | | |
| struct* | `curandDistributionShift_t` | | |
| struct | `curandDistributionM2Shift_st` | | |
| struct* | `curandDistributionM2Shift_t` | | |
| struct | `curandHistogramM2_st` | | |
| struct* | `curandHistogramM2_t` | | |
| uint | `curandHistogramM2K_st` | | |
| uint* | `curandHistogramM2K_t` | | |
| struct | `curandDiscreteDistribution_st` | | `hiprandDiscreteDistribution_st` |
| struct* | `curandDiscreteDistribution_t` | | `hiprandDiscreteDistribution_t` |
| enum |***`curandMethod`*** | | |
| enum |***`curandMethod_t`*** | | |
| 0 |*`CURAND_CHOOSE_BEST`* | | |
| 1 |*`CURAND_ITR`* | | |
| 2 |*`CURAND_KNUTH`* | | |
| 3 |*`CURAND_HITR`* | | |
| 4 |*`CURAND_M1`* | | |
| 5 |*`CURAND_M2`* | | |
| 6 |*`CURAND_BINARY_SEARCH`* | | |
| 7 |*`CURAND_DISCRETE_GAUSS`* | | |
| 8 |*`CURAND_REJECTION`* | | |
| 9 |*`CURAND_DEVICE_API`* | | |
| 10 |*`CURAND_FAST_REJECTION`* | | |
| 11 |*`CURAND_3RD`* | | |
| 12 |*`CURAND_DEFINITION`* | | |
| 13 |*`CURAND_POISSON`* | | |
| struct | `curandStateMtgp32` | | `hiprandStateMtgp32` |
| typedef | `curandStateMtgp32_t` | | `hiprandStateMtgp32_t` |
| struct | `curandStateScrambledSobol64` | | |
| typedef | `curandStateScrambledSobol64_t` | | |
| struct | `curandStateSobol64` | | |
| typedef | `curandStateSobol64_t` | | |
| struct | `curandStateScrambledSobol32` | | |
| typedef | `curandStateScrambledSobol32_t` | | |
| struct | `curandStateSobol32` | | `hiprandStateSobol32` |
| typedef | `curandStateSobol32_t` | | `hiprandStateSobol32_t` |
| struct | `curandStateMRG32k3a` | | `hiprandStateMRG32k3a` |
| typedef | `curandStateMRG32k3a_t` | | `hiprandStateMRG32k3a_t` |
| struct | `curandStatePhilox4_32_10` | | `hiprandStatePhilox4_32_10` |
| typedef | `curandStatePhilox4_32_10_t` | | `hiprandStatePhilox4_32_10_t` |
| struct | `curandStateXORWOW` | | `hiprandStateXORWOW` |
| typedef | `curandStateXORWOW_t` | | `hiprandStateXORWOW_t` |
| struct | `curandState` | | `hiprandState` |
| typedef | `curandState_t` | | `hiprandState_t` |
## **2. Host API Functions**
| **CUDA** | **HIP** |
|-----------------------------------------------------------|--------------------------------------------|
| `curandCreateGenerator` | `hiprandCreateGenerator` |
| `curandCreateGeneratorHost` | `hiprandCreateGeneratorHost` |
| `curandCreatePoissonDistribution` | `hiprandCreatePoissonDistribution` |
| `curandDestroyDistribution` | `hiprandDestroyDistribution` |
| `curandDestroyGenerator` | `hiprandDestroyGenerator` |
| `curandGenerate` | `hiprandGenerate` |
| `curandGenerateLogNormal` | `hiprandGenerateLogNormal` |
| `curandGenerateLogNormalDouble` | `hiprandGenerateLogNormalDouble` |
| `curandGenerateLongLong` | |
| `curandGenerateNormal` | `hiprandGenerateNormal` |
| `curandGenerateNormalDouble` | `hiprandGenerateNormalDouble` |
| `curandGeneratePoisson` | `hiprandGeneratePoisson` |
| `curandGenerateSeeds` | `hiprandGenerateSeeds` |
| `curandGenerateUniform` | `hiprandGenerateUniform` |
| `curandGenerateUniformDouble` | `hiprandGenerateUniformDouble` |
| `curandGetDirectionVectors32` | |
| `curandGetDirectionVectors64` | |
| `curandGetProperty` | |
| `curandGetScrambleConstants32` | |
| `curandGetScrambleConstants64` | |
| `curandGetVersion` | `hiprandGetVersion` |
| `curandSetGeneratorOffset` | `hiprandSetGeneratorOffset` |
| `curandSetGeneratorOrdering` | |
| `curandSetPseudoRandomGeneratorSeed` | `hiprandSetPseudoRandomGeneratorSeed` |
| `curandSetQuasiRandomGeneratorDimensions` | `hiprandSetQuasiRandomGeneratorDimensions` |
| `curandSetStream` | `hiprandSetStream` |
| `curandMakeMTGP32Constants` | `hiprandMakeMTGP32Constants` |
| `curandMakeMTGP32KernelState` | `hiprandMakeMTGP32KernelState` |
## **3. Device API Functions**
| **CUDA** | **HIP** |
|-----------------------------------------------------------|--------------------------------------------|
| `curand` | `hiprand` |
| `curand_init` | `hiprand_init` |
| `curand_log_normal` | `hiprand_log_normal` |
| `curand_log_normal_double` | `hiprand_log_normal_double` |
| `curand_log_normal2` | `hiprand_log_normal2` |
| `curand_log_normal2_double` | `hiprand_log_normal2_double` |
| `curand_log_normal4` | `hiprand_log_normal4` |
| `curand_log_normal4_double` | `hiprand_log_normal4_double` |
| `curand_mtgp32_single` | |
| `curand_mtgp32_single_specific` | |
| `curand_mtgp32_specific` | |
| `curand_normal` | `hiprand_normal` |
| `curand_normal_double` | `hiprand_normal_double` |
| `curand_normal2` | `hiprand_normal2` |
| `curand_normal2_double` | `hiprand_normal2_double` |
| `curand_normal4` | `hiprand_normal4` |
| `curand_normal4_double` | `hiprand_normal4_double` |
| `curand_uniform` | `hiprand_uniform` |
| `curand_uniform_double` | `hiprand_uniform_double` |
| `curand_uniform2_double` | `hiprand_uniform2_double` |
| `curand_uniform4` | `hiprand_uniform4` |
| `curand_uniform4_double` | `hiprand_uniform4_double` |
| `curand_discrete` | `hiprand_discrete` |
| `curand_discrete4` | `hiprand_discrete4` |
| `curand_poisson` | `hiprand_poisson` |
| `curand_poisson4` | `hiprand_poisson4` |
| `curand_Philox4x32_10` | |
| `skipahead` | `skipahead` |
| `skipahead_sequence` | `skipahead_sequence` |
| `skipahead_subsequence` | `skipahead_subsequence` |
docs/markdown/CUSPARSE_API_supported_by_HIP.md
-797
View File
@@ -1,797 +0,0 @@
# CUSPARSE API supported by HIP
## **1. cuSPARSE Data types**
| **type** | **CUDA** |**CUDA version\***| **HIP** |
|-------------:|---------------------------------------------------------------|:-----------------|------------------------------------------------------------|
| define |`CUSPARSE_VER_MAJOR` | 10.1 Update 2 | |
| define |`CUSPARSE_VER_MINOR` | 10.1 Update 2 | |
| define |`CUSPARSE_VER_PATCH` | 10.1 Update 2 | |
| define |`CUSPARSE_VER_BUILD` | 10.1 Update 2 | |
| define |`CUSPARSE_VERSION` | 10.1 Update 2 | |
| enum |***`cusparseAction_t`*** | |***`hipsparseAction_t`*** |
| 0 |*`CUSPARSE_ACTION_SYMBOLIC`* | |*`HIPSPARSE_ACTION_SYMBOLIC`* |
| 1 |*`CUSPARSE_ACTION_NUMERIC`* | |*`HIPSPARSE_ACTION_NUMERIC`* |
| enum |***`cusparseDirection_t`*** | |***`hipsparseDirection_t`*** |
| 0 |*`CUSPARSE_DIRECTION_ROW`* | |*`HIPSPARSE_DIRECTION_ROW`* |
| 1 |*`CUSPARSE_DIRECTION_COLUMN`* | |*`HIPSPARSE_DIRECTION_COLUMN`* |
| enum |***`cusparseHybPartition_t`*** | |***`hipsparseHybPartition_t`*** |
| 0 |*`CUSPARSE_HYB_PARTITION_AUTO`* | |*`HIPSPARSE_HYB_PARTITION_AUTO`* |
| 1 |*`CUSPARSE_HYB_PARTITION_USER`* | |*`HIPSPARSE_HYB_PARTITION_USER`* |
| 2 |*`CUSPARSE_HYB_PARTITION_MAX`* | |*`HIPSPARSE_HYB_PARTITION_MAX`* |
| enum |***`cusparseDiagType_t`*** | |***`hipsparseDiagType_t`*** |
| 0 |*`CUSPARSE_DIAG_TYPE_NON_UNIT`* | |*`HIPSPARSE_DIAG_TYPE_NON_UNIT`* |
| 1 |*`CUSPARSE_DIAG_TYPE_UNIT`* | |*`HIPSPARSE_DIAG_TYPE_UNIT`* |
| enum |***`cusparseFillMode_t`*** | |***`hipsparseFillMode_t`*** |
| 0 |*`CUSPARSE_FILL_MODE_LOWER`* | |*`HIPSPARSE_FILL_MODE_LOWER`* |
| 1 |*`CUSPARSE_FILL_MODE_UPPER`* | |*`HIPSPARSE_FILL_MODE_UPPER`* |
| enum |***`cusparseIndexBase_t`*** | |***`hipsparseIndexBase_t`*** |
| 0 |*`CUSPARSE_INDEX_BASE_ZERO`* | |*`HIPSPARSE_INDEX_BASE_ZERO`* |
| 1 |*`CUSPARSE_INDEX_BASE_ONE`* | |*`HIPSPARSE_INDEX_BASE_ONE`* |
| enum |***`cusparseMatrixType_t`*** | |***`hipsparseMatrixType_t`*** |
| 0 |*`CUSPARSE_MATRIX_TYPE_GENERAL`* | |*`HIPSPARSE_MATRIX_TYPE_GENERAL`* |
| 1 |*`CUSPARSE_MATRIX_TYPE_SYMMETRIC`* | |*`HIPSPARSE_MATRIX_TYPE_SYMMETRIC`* |
| 2 |*`CUSPARSE_MATRIX_TYPE_HERMITIAN`* | |*`HIPSPARSE_MATRIX_TYPE_HERMITIAN`* |
| 3 |*`CUSPARSE_MATRIX_TYPE_TRIANGULAR`* | |*`HIPSPARSE_MATRIX_TYPE_TRIANGULAR`* |
| enum |***`cusparseOperation_t`*** | |***`hipsparseOperation_t`*** |
| 0 |*`CUSPARSE_OPERATION_NON_TRANSPOSE`* | |*`HIPSPARSE_OPERATION_NON_TRANSPOSE`* |
| 1 |*`CUSPARSE_OPERATION_TRANSPOSE`* | |*`HIPSPARSE_OPERATION_TRANSPOSE`* |
| 2 |*`CUSPARSE_OPERATION_CONJUGATE_TRANSPOSE`* | |*`HIPSPARSE_OPERATION_CONJUGATE_TRANSPOSE`* |
| enum |***`cusparsePointerMode_t`*** | |***`hipsparsePointerMode_t`*** |
| 0 |*`CUSPARSE_POINTER_MODE_HOST`* | |*`HIPSPARSE_POINTER_MODE_HOST`* |
| 1 |*`CUSPARSE_POINTER_MODE_DEVICE`* | |*`HIPSPARSE_POINTER_MODE_DEVICE`* |
| enum |***`cusparseAlgMode_t`*** | 8.0 | |
| 0 |*`CUSPARSE_ALG0`* | 8.0 | |
| 1 |*`CUSPARSE_ALG1`* | 8.0 | |
| 0 |*`CUSPARSE_ALG_NAIVE`* | 9.2 | |
| 1 |*`CUSPARSE_ALG_MERGE_PATH`* | 9.2 | |
| enum |***`cusparseSolvePolicy_t`*** | |***`hipsparseSolvePolicy_t`*** |
| 0 |*`CUSPARSE_SOLVE_POLICY_NO_LEVEL`* | |*`HIPSPARSE_SOLVE_POLICY_NO_LEVEL`* |
| 1 |*`CUSPARSE_SOLVE_POLICY_USE_LEVEL`* | |*`HIPSPARSE_SOLVE_POLICY_USE_LEVEL`* |
| enum |***`cusparseStatus_t`*** | |***`hipsparseMatrixType_t`*** |
| 0 |*`CUSPARSE_STATUS_SUCCESS`* | |*`HIPSPARSE_STATUS_SUCCESS`* |
| 1 |*`CUSPARSE_STATUS_NOT_INITIALIZED`* | |*`HIPSPARSE_STATUS_NOT_INITIALIZED`* |
| 2 |*`CUSPARSE_STATUS_ALLOC_FAILED`* | |*`HIPSPARSE_STATUS_ALLOC_FAILED`* |
| 3 |*`CUSPARSE_STATUS_INVALID_VALUE`* | |*`HIPSPARSE_STATUS_INVALID_VALUE`* |
| 4 |*`CUSPARSE_STATUS_ARCH_MISMATCH`* | |*`HIPSPARSE_STATUS_ARCH_MISMATCH`* |
| 5 |*`CUSPARSE_STATUS_MAPPING_ERROR`* | |*`HIPSPARSE_STATUS_MAPPING_ERROR`* |
| 6 |*`CUSPARSE_STATUS_EXECUTION_FAILED`* | |*`HIPSPARSE_STATUS_EXECUTION_FAILED`* |
| 7 |*`CUSPARSE_STATUS_INTERNAL_ERROR`* | |*`HIPSPARSE_STATUS_INTERNAL_ERROR`* |
| 8 |*`CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED`* | |*`HIPSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED`* |
| 9 |*`CUSPARSE_STATUS_ZERO_PIVOT`* | |*`HIPSPARSE_STATUS_ZERO_PIVOT`* |
| struct |`cusparseContext` | | |
| typedef |`cusparseHandle_t` | |`hipsparseHandle_t` |
| struct |`cusparseHybMat` | | |
| typedef |`cusparseHybMat_t` | |`hipsparseHybMat_t` |
| struct |`cusparseMatDescr` | | |
| typedef |`cusparseMatDescr_t` | |`hipsparseMatDescr_t` |
| struct |`cusparseSolveAnalysisInfo` | | |
| typedef |`cusparseSolveAnalysisInfo_t` | | |
| struct |`csrsv2Info` | | |
| typedef |`csrsv2Info_t` | |`csrsv2Info_t` |
| struct |`csrsm2Info` | 9.2 |`csrsm2Info` |
| typedef |`csrsm2Info_t` | |`csrsm2Info_t` |
| struct |`bsrsv2Info` | | |
| typedef |`bsrsv2Info_t` | | |
| struct |`bsrsm2Info` | | |
| typedef |`bsrsm2Info_t` | | |
| struct |`bsric02Info` | | |
| typedef |`bsric02Info_t` | | |
| struct |`csrilu02Info` | | |
| typedef |`csrilu02Info_t` | |`csrilu02Info_t` |
| struct |`bsrilu02Info` | | |
| typedef |`bsrilu02Info_t` | | |
| struct |`csru2csrInfo` | | |
| typedef |`csru2csrInfo_t` | | |
| struct |`csrgemm2Info` | |`csrgemm2Info` |
| typedef |`csrgemm2Info_t` | |`csrgemm2Info_t` |
| struct |`cusparseColorInfo` | | |
| typedef |`cusparseColorInfo_t` | | |
| struct |`pruneInfo` | 9.0 | |
| typedef |`pruneInfo_t` | 9.0 | |
| enum |***`cusparseCsr2CscAlg_t`*** | 10.1 | |
| 1 |*`CUSPARSE_CSR2CSC_ALG1`* | 10.1 | |
| 2 |*`CUSPARSE_CSR2CSC_ALG2`* | 10.1 | |
| enum |***`cusparseFormat_t`*** | 10.1 | |
| 1 |*`CUSPARSE_FORMAT_CSR`* | 10.1 | |
| 2 |*`CUSPARSE_FORMAT_CSC`* | 10.1 | |
| 3 |*`CUSPARSE_FORMAT_COO`* | 10.1 | |
| 4 |*`CUSPARSE_FORMAT_COO_AOS`* | 10.1 | |
| enum |***`cusparseOrder_t`*** | 10.1 | |
| 1 |*`CUSPARSE_ORDER_COL`* | 10.1 | |
| 2 |*`CUSPARSE_ORDER_ROW`* | 10.1 | |
| enum |***`cusparseSpMVAlg_t`*** | 10.1 | |
| 0 |*`CUSPARSE_MV_ALG_DEFAULT`* | 10.1 | |
| 1 |*`CUSPARSE_COOMV_ALG`* | 10.1 | |
| 2 |*`CUSPARSE_CSRMV_ALG1`* | 10.1 | |
| 3 |*`CUSPARSE_CSRMV_ALG2`* | 10.1 | |
| enum |***`cusparseSpMMAlg_t`*** | 10.1 | |
| 0 |*`CUSPARSE_MM_ALG_DEFAULT`* | 10.1 | |
| 1 |*`CUSPARSE_COOMM_ALG1`* | 10.1 | |
| 2 |*`CUSPARSE_COOMM_ALG2`* | 10.1 | |
| 3 |*`CUSPARSE_COOMM_ALG3`* | 10.1 | |
| 4 |*`CUSPARSE_CSRMM_ALG1`* | 10.1 | |
| enum |***`cusparseIndexType_t`*** | 10.1 | |
| 1 |*`CUSPARSE_INDEX_16U`* | 10.1 | |
| 2 |*`CUSPARSE_INDEX_32I`* | 10.1 | |
| 3 |*`CUSPARSE_INDEX_64I`* | 10.1 | |
| struct |`cusparseSpMatDescr` | 10.1 | |
| typedef |`cusparseSpMatDescr_t` | 10.1 | |
| struct |`cusparseDnMatDescr` | 10.1 | |
| typedef |`cusparseDnMatDescr_t` | 10.1 | |
| struct |`cusparseSpVecDescr` | 10.1 | |
| typedef |`cusparseSpVecDescr_t` | 10.1 | |
| struct |`cusparseDnVecDescr` | 10.1 | |
| typedef |`cusparseDnVecDescr_t` | 10.1 | |
## **2. cuSPARSE Helper Function Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseCreate` |`hipsparseCreate` |
|`cusparseCreateSolveAnalysisInfo` | |
|`cusparseCreateHybMat` |`hipsparseCreateHybMat` |
|`cusparseCreateMatDescr` |`hipsparseCreateMatDescr` |
|`cusparseDestroy` |`hipsparseDestroy` |
|`cusparseDestroySolveAnalysisInfo` | |
|`cusparseDestroyHybMat` |`hipsparseDestroyHybMat` |
|`cusparseDestroyMatDescr` |`hipsparseDestroyMatDescr` |
|`cusparseGetLevelInfo` | |
|`cusparseGetMatDiagType` |`hipsparseGetMatDiagType` |
|`cusparseGetMatFillMode` |`hipsparseGetMatFillMode` |
|`cusparseGetMatIndexBase` |`hipsparseGetMatIndexBase` |
|`cusparseGetMatType` |`hipsparseGetMatType` |
|`cusparseGetPointerMode` |`hipsparseGetPointerMode` |
|`cusparseGetVersion` |`hipsparseGetVersion` |
|`cusparseSetMatDiagType` |`hipsparseSetMatDiagType` |
|`cusparseSetMatFillMode` |`hipsparseSetMatFillMode` |
|`cusparseSetMatType` |`hipsparseSetMatType` |
|`cusparseSetPointerMode` |`hipsparseSetPointerMode` |
|`cusparseSetStream` |`hipsparseSetStream` |
|`cusparseGetStream` |`hipsparseGetStream` | 8.0 |
|`cusparseCreateCsrsv2Info` |`hipsparseCreateCsrsv2Info` |
|`cusparseDestroyCsrsv2Info` |`hipsparseDestroyCsrsv2Info` |
|`cusparseCreateCsrsm2Info` |`hipsparseCreateCsrsm2Info` | 9.2 |
|`cusparseDestroyCsrsm2Info` |`hipsparseDestroyCsrsm2Info` | 9.2 |
|`cusparseCreateCsric02Info` | |
|`cusparseDestroyCsric02Info` | |
|`cusparseCreateCsrilu02Info` |`hipsparseCreateCsrilu02Info` |
|`cusparseDestroyCsrilu02Info` |`hipsparseDestroyCsrilu02Info` |
|`cusparseCreateBsrsv2Info` | |
|`cusparseDestroyBsrsv2Info` | |
|`cusparseCreateBsrsm2Info` | |
|`cusparseDestroyBsrsm2Info` | |
|`cusparseCreateBsric02Info` | |
|`cusparseDestroyBsric02Info` | |
|`cusparseCreateBsrilu02Info` | |
|`cusparseDestroyBsrilu02Info` | |
|`cusparseCreateCsrgemm2Info` |`hipsparseCreateCsrgemm2Info` |
|`cusparseDestroyCsrgemm2Info` |`hipsparseDestroyCsrgemm2Info` |
|`cusparseCreatePruneInfo` | | 9.0 |
|`cusparseDestroyPruneInfo` | | 9.0 |
## **3. cuSPARSE Level 1 Function Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSaxpyi` |`hipsparseSaxpyi` |
|`cusparseDaxpyi` |`hipsparseDaxpyi` |
|`cusparseCaxpyi` |`hipsparseCaxpyi` |
|`cusparseZaxpyi` |`hipsparseZaxpyi` |
|`cusparseSdoti` |`hipsparseSdoti` |
|`cusparseDdoti` |`hipsparseDdoti` |
|`cusparseCdoti` |`hipsparseCdoti` |
|`cusparseZdoti` |`hipsparseZdoti` |
|`cusparseCdotci` |`hipsparseCdotci` |
|`cusparseZdotci` |`hipsparseZdotci` |
|`cusparseSgthr` |`hipsparseSgthr` |
|`cusparseDgthr` |`hipsparseDgthr` |
|`cusparseCgthr` |`hipsparseCgthr` |
|`cusparseZgthr` |`hipsparseZgthr` |
|`cusparseSgthrz` |`hipsparseSgthrz` |
|`cusparseDgthrz` |`hipsparseDgthrz` |
|`cusparseCgthrz` |`hipsparseCgthrz` |
|`cusparseZgthrz` |`hipsparseZgthrz` |
|`cusparseSroti` |`hipsparseSroti` |
|`cusparseDroti` |`hipsparseDroti` |
|`cusparseSsctr` |`hipsparseSsctr` |
|`cusparseDsctr` |`hipsparseDsctr` |
|`cusparseCsctr` |`hipsparseCsctr` |
|`cusparseZsctr` |`hipsparseZsctr` |
## **4. cuSPARSE Level 2 Function Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSbsrmv` |`hipsparseSbsrmv` |
|`cusparseDbsrmv` |`hipsparseDbsrmv` |
|`cusparseCbsrmv` |`hipsparseCbsrmv` |
|`cusparseZbsrmv` |`hipsparseZbsrmv` |
|`cusparseSbsrxmv` | |
|`cusparseDbsrxmv` | |
|`cusparseCbsrxmv` | |
|`cusparseZbsrxmv` | |
|`cusparseScsrmv` |`hipsparseScsrmv` |
|`cusparseDcsrmv` |`hipsparseDcsrmv` |
|`cusparseCcsrmv` |`hipsparseCcsrmv` |
|`cusparseZcsrmv` |`hipsparseZcsrmv` |
|`cusparseCsrmvEx` | | 8.0 |
|`cusparseCsrmvEx_bufferSize` | | 8.0 |
|`cusparseScsrmv_mp` | | 8.0 |
|`cusparseDcsrmv_mp` | | 8.0 |
|`cusparseCcsrmv_mp` | | 8.0 |
|`cusparseZcsrmv_mp` | | 8.0 |
|`cusparseSgemvi` | | 7.5 |
|`cusparseDgemvi` | | 7.5 |
|`cusparseCgemvi` | | 7.5 |
|`cusparseZgemvi` | | 7.5 |
|`cusparseSgemvi_bufferSize` | | 7.5 |
|`cusparseDgemvi_bufferSize` | | 7.5 |
|`cusparseCgemvi_bufferSize` | | 7.5 |
|`cusparseZgemvi_bufferSize` | | 7.5 |
|`cusparseSbsrsv2_bufferSize` | |
|`cusparseSbsrsv2_bufferSizeExt` | |
|`cusparseDbsrsv2_bufferSize` | |
|`cusparseDbsrsv2_bufferSizeExt` | |
|`cusparseCbsrsv2_bufferSize` | |
|`cusparseCbsrsv2_bufferSizeExt` | |
|`cusparseZbsrsv2_bufferSize` | |
|`cusparseZbsrsv2_bufferSizeExt` | |
|`cusparseSbsrsv2_analysis` | |
|`cusparseDbsrsv2_analysis` | |
|`cusparseCbsrsv2_analysis` | |
|`cusparseZbsrsv2_analysis` | |
|`cusparseXbsrsv2_zeroPivot` | |
|`cusparseSbsrsv2_solve | |
|`cusparseDbsrsv2_solve | |
|`cusparseCbsrsv2_solve | |
|`cusparseZbsrsv2_solve | |
|`cusparseScsrsv_analysis` | |
|`cusparseDcsrsv_analysis` | |
|`cusparseCcsrsv_analysis` | |
|`cusparseZcsrsv_analysis` | |
|`cusparseCsrsv_analysisEx` | | 8.0 |
|`cusparseScsrsv_solve` | |
|`cusparseDcsrsv_solve` | |
|`cusparseCcsrsv_solve` | |
|`cusparseZcsrsv_solve` | |
|`cusparseCsrsv_solveEx` | | 8.0 |
|`cusparseScsrsv2_bufferSize` |`hipsparseScsrsv2_bufferSize` |
|`cusparseScsrsv2_bufferSizeExt` |`hipsparseScsrsv2_bufferSizeExt` |
|`cusparseDcsrsv2_bufferSize` |`hipsparseDcsrsv2_bufferSize` |
|`cusparseDcsrsv2_bufferSizeExt` |`hipsparseDcsrsv2_bufferSizeExt` |
|`cusparseCcsrsv2_bufferSize` |`hipsparseCcsrsv2_bufferSize` |
|`cusparseCcsrsv2_bufferSizeExt` |`hipsparseCcsrsv2_bufferSizeExt` |
|`cusparseZcsrsv2_bufferSize` |`hipsparseZcsrsv2_bufferSize` |
|`cusparseZcsrsv2_bufferSizeExt` |`hipsparseZcsrsv2_bufferSizeExt` |
|`cusparseScsrsv2_analysis` |`hipsparseScsrsv2_analysis` |
|`cusparseDcsrsv2_analysis` |`hipsparseDcsrsv2_analysis` |
|`cusparseCcsrsv2_analysis` |`hipsparseCcsrsv2_analysis` |
|`cusparseZcsrsv2_analysis` |`hipsparseZcsrsv2_analysis` |
|`cusparseScsrsv2_solve` |`hipsparseScsrsv2_solve` |
|`cusparseDcsrsv2_solve` |`hipsparseDcsrsv2_solve` |
|`cusparseCcsrsv2_solve` |`hipsparseCcsrsv2_solve` |
|`cusparseZcsrsv2_solve` |`hipsparseZcsrsv2_solve` |
|`cusparseXcsrsv2_zeroPivot` |`hipsparseXcsrsv2_zeroPivot` |
|`cusparseShybmv` |`hipsparseShybmv` |
|`cusparseDhybmv` |`hipsparseDhybmv` |
|`cusparseChybmv` |`hipsparseChybmv` |
|`cusparseZhybmv` |`hipsparseZhybmv` |
|`cusparseShybsv_analysis` | |
|`cusparseDhybsv_analysis` | |
|`cusparseChybsv_analysis` | |
|`cusparseZhybsv_analysis` | |
|`cusparseShybsv_solve` | |
|`cusparseDhybsv_solve` | |
|`cusparseChybsv_solve` | |
|`cusparseZhybsv_solve` | |
## **5. cuSPARSE Level 3 Function Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseScsrmm` |`hipsparseScsrmm` |
|`cusparseDcsrmm` |`hipsparseDcsrmm` |
|`cusparseCcsrmm` |`hipsparseCcsrmm` |
|`cusparseZcsrmm` |`hipsparseZcsrmm` |
|`cusparseScsrmm2` |`hipsparseScsrmm2` |
|`cusparseDcsrmm2` |`hipsparseDcsrmm2` |
|`cusparseCcsrmm2` |`hipsparseCcsrmm2` |
|`cusparseZcsrmm2` |`hipsparseZcsrmm2` |
|`cusparseScsrsm_analysis` | |
|`cusparseDcsrsm_analysis` | |
|`cusparseCcsrsm_analysis` | |
|`cusparseZcsrsm_analysis` | |
|`cusparseScsrsm_solve` | |
|`cusparseDcsrsm_solve` | |
|`cusparseCcsrsm_solve` | |
|`cusparseZcsrsm_solve` | |
|`cusparseScsrsm2_bufferSizeExt` |`hipsparseScsrsm2_bufferSizeExt` | 9.2 |
|`cusparseDcsrsm2_bufferSizeExt` |`hipsparseDcsrsm2_bufferSizeExt` | 9.2 |
|`cusparseCcsrsm2_bufferSizeExt` |`hipsparseCcsrsm2_bufferSizeExt` | 9.2 |
|`cusparseZcsrsm2_bufferSizeExt` |`hipsparseZcsrsm2_bufferSizeExt` | 9.2 |
|`cusparseScsrsm2_analysis` |`hipsparseScsrsm2_analysis` | 9.2 |
|`cusparseDcsrsm2_analysis` |`hipsparseDcsrsm2_analysis` | 9.2 |
|`cusparseCcsrsm2_analysis` |`hipsparseCcsrsm2_analysis` | 9.2 |
|`cusparseZcsrsm2_analysis` |`hipsparseZcsrsm2_analysis` | 9.2 |
|`cusparseScsrsm2_solve` |`hipsparseScsrsm2_solve` | 9.2 |
|`cusparseDcsrsm2_solve` |`hipsparseDcsrsm2_solve` | 9.2 |
|`cusparseCcsrsm2_solve` |`hipsparseCcsrsm2_solve` | 9.2 |
|`cusparseZcsrsm2_solve` |`hipsparseZcsrsm2_solve` | 9.2 |
|`cusparseXcsrsm2_zeroPivot` |`hipsparseXcsrsm2_zeroPivot` | 9.2 |
|`cusparseSbsrmm` | |
|`cusparseDbsrmm` | |
|`cusparseCbsrmm` | |
|`cusparseZbsrmm` | |
|`cusparseSbsrsm2_bufferSize` | |
|`cusparseSbsrsm2_bufferSizeExt` | |
|`cusparseDbsrsm2_bufferSize` | |
|`cusparseDbsrsm2_bufferSizeExt` | |
|`cusparseCbsrsm2_bufferSize` | |
|`cusparseCbsrsm2_bufferSizeExt` | |
|`cusparseZbsrsm2_bufferSize` | |
|`cusparseZbsrsm2_bufferSizeExt` | |
|`cusparseSbsrsm2_analysis` | |
|`cusparseDbsrsm2_analysis` | |
|`cusparseCbsrsm2_analysis` | |
|`cusparseZbsrsm2_analysis` | |
|`cusparseSbsrsm2_solve` | |
|`cusparseDbsrsm2_solve` | |
|`cusparseCbsrsm2_solve` | |
|`cusparseZbsrsm2_solve` | |
|`cusparseXbsrsm2_zeroPivot` | |
|`cusparseSgemmi` | | 8.0 |
|`cusparseDgemmi` | | 8.0 |
|`cusparseCgemmi` | | 8.0 |
|`cusparseZgemmi` | | 8.0 |
## **6. cuSPARSE Extra Function Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseXcsrgeamNnz` |`hipsparseXcsrgeamNnz` |
|`cusparseScsrgeam` |`hipsparseScsrgeam` |
|`cusparseDcsrgeam` |`hipsparseDcsrgeam` |
|`cusparseCcsrgeam` |`hipsparseCcsrgeam` |
|`cusparseZcsrgeam` |`hipsparseZcsrgeam` |
|`cusparseXcsrgeam2Nnz` |`hipsparseXcsrgeam2Nnz` | 9.2 |
|`cusparseScsrgeam2` |`hipsparseScsrgeam2` | 9.2 |
|`cusparseDcsrgeam2` |`hipsparseDcsrgeam2` | 9.2 |
|`cusparseCcsrgeam2` |`hipsparseCcsrgeam2` | 9.2 |
|`cusparseZcsrgeam2` |`hipsparseZcsrgeam2` | 9.2 |
|`cusparseScsrgeam2_bufferSizeExt` |`hipsparseScsrgeam2_bufferSizeExt` | 9.2 |
|`cusparseDcsrgeam2_bufferSizeExt` |`hipsparseDcsrgeam2_bufferSizeExt` | 9.2 |
|`cusparseCcsrgeam2_bufferSizeExt` |`hipsparseCcsrgeam2_bufferSizeExt` | 9.2 |
|`cusparseZcsrgeam2_bufferSizeExt` |`hipsparseZcsrgeam2_bufferSizeExt` | 9.2 |
|`cusparseXcsrgemmNnz` |`hipsparseXcsrgemmNnz` |
|`cusparseScsrgemm` |`hipsparseScsrgemm` |
|`cusparseDcsrgemm` |`hipsparseDcsrgemm` |
|`cusparseCcsrgemm` |`hipsparseCcsrgemm` |
|`cusparseZcsrgemm` |`hipsparseZcsrgemm` |
|`cusparseXcsrgemm2Nnz` |`hipsparseXcsrgemm2Nnz` |
|`cusparseScsrgemm2` |`hipsparseScsrgemm2` |
|`cusparseDcsrgemm2` |`hipsparseDcsrgemm2` |
|`cusparseCcsrgemm2` |`hipsparseCcsrgemm2` |
|`cusparseZcsrgemm2` |`hipsparseZcsrgemm2` |
|`cusparseScsrgemm2_bufferSizeExt` |`hipsparseScsrgemm2_bufferSizeExt` |
|`cusparseDcsrgemm2_bufferSizeExt` |`hipsparseDcsrgemm2_bufferSizeExt` |
|`cusparseCcsrgemm2_bufferSizeExt` |`hipsparseCcsrgemm2_bufferSizeExt` |
|`cusparseZcsrgemm2_bufferSizeExt` |`hipsparseZcsrgemm2_bufferSizeExt` |
## **7. cuSPARSE Preconditioners Reference**
## ***7.1. Incomplete Cholesky Factorization: level 0***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseScsric0` | |
|`cusparseDcsric0` | |
|`cusparseCcsric0` | |
|`cusparseZcsric0` | |
|`cusparseScsric02_bufferSize` | |
|`cusparseScsric02_bufferSizeExt` | |
|`cusparseDcsric02_bufferSize` | |
|`cusparseDcsric02_bufferSizeExt` | |
|`cusparseCcsric02_bufferSize` | |
|`cusparseCcsric02_bufferSizeExt` | |
|`cusparseZcsric02_bufferSize` | |
|`cusparseZcsric02_bufferSizeExt` | |
|`cusparseScsric02_analysis` | |
|`cusparseDcsric02_analysis` | |
|`cusparseCcsric02_analysis` | |
|`cusparseZcsric02_analysis` | |
|`cusparseScsric02` | |
|`cusparseDcsric02` | |
|`cusparseCcsric02` | |
|`cusparseZcsric02` | |
|`cusparseXcsric02_zeroPivot` | |
|`cusparseSbsric02_bufferSize` | |
|`cusparseSbsric02_bufferSizeExt` | |
|`cusparseDbsric02_bufferSize` | |
|`cusparseDbsric02_bufferSizeExt` | |
|`cusparseCbsric02_bufferSize` | |
|`cusparseCbsric02_bufferSizeExt` | |
|`cusparseZbsric02_bufferSize` | |
|`cusparseZbsric02_bufferSizeExt` | |
|`cusparseSbsric02_analysis` | |
|`cusparseDbsric02_analysis` | |
|`cusparseCbsric02_analysis` | |
|`cusparseZbsric02_analysis` | |
|`cusparseSbsric02` | |
|`cusparseDbsric02` | |
|`cusparseCbsric02` | |
|`cusparseZbsric02` | |
|`cusparseXbsric02_zeroPivot` | |
## ***7.2. Incomplete LU Factorization: level 0***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseScsrilu0` | |
|`cusparseDcsrilu0` | |
|`cusparseCcsrilu0` | |
|`cusparseZcsrilu0` | |
|`cusparseCsrilu0Ex` | | 8.0 |
|`cusparseScsrilu02_numericBoost` | |
|`cusparseDcsrilu02_numericBoost` | |
|`cusparseCcsrilu02_numericBoost` | |
|`cusparseZcsrilu02_numericBoost` | |
|`cusparseXcsrilu02_zeroPivot` |`hipsparseXcsrilu02_zeroPivot` |
|`cusparseScsrilu02_bufferSize` |`hipsparseScsrilu02_bufferSize` |
|`cusparseScsrilu02_bufferSizeExt` |`hipsparseScsrilu02_bufferSizeExt` |
|`cusparseDcsrilu02_bufferSize` |`hipsparseDcsrilu02_bufferSize` |
|`cusparseDcsrilu02_bufferSizeExt` |`hipsparseDcsrilu02_bufferSizeExt` |
|`cusparseCcsrilu02_bufferSize` |`hipsparseCcsrilu02_bufferSize` |
|`cusparseCcsrilu02_bufferSizeExt` |`hipsparseCcsrilu02_bufferSizeExt` |
|`cusparseZcsrilu02_bufferSize` |`hipsparseZcsrilu02_bufferSize` |
|`cusparseZcsrilu02_bufferSizeExt` |`hipsparseZcsrilu02_bufferSizeExt` |
|`cusparseScsrilu02_analysis` |`hipsparseScsrilu02_analysis` |
|`cusparseDcsrilu02_analysis` |`hipsparseDcsrilu02_analysis` |
|`cusparseCcsrilu02_analysis` |`hipsparseCcsrilu02_analysis` |
|`cusparseZcsrilu02_analysis` |`hipsparseZcsrilu02_analysis` |
|`cusparseScsrilu02` |`hipsparseScsrilu02` |
|`cusparseDcsrilu02` |`hipsparseDcsrilu02` |
|`cusparseCcsrilu02` |`hipsparseCcsrilu02` |
|`cusparseZcsrilu02` |`hipsparseZcsrilu02` |
|`cusparseXbsric02_zeroPivot` |`hipsparseXcsrilu02_zeroPivot` |
|`cusparseSbsrilu02_numericBoost` | |
|`cusparseDbsrilu02_numericBoost` | |
|`cusparseCbsrilu02_numericBoost` | |
|`cusparseZbsrilu02_numericBoost` | |
|`cusparseSbsrilu02_bufferSize` | |
|`cusparseSbsrilu02_bufferSizeExt` | |
|`cusparseDbsrilu02_bufferSize` | |
|`cusparseDbsrilu02_bufferSizeExt` | |
|`cusparseCbsrilu02_bufferSize` | |
|`cusparseCbsrilu02_bufferSizeExt` | |
|`cusparseZbsrilu02_bufferSize` | |
|`cusparseZbsrilu02_bufferSizeExt` | |
|`cusparseSbsrilu02_analysis` | |
|`cusparseDbsrilu02_analysis` | |
|`cusparseCbsrilu02_analysis` | |
|`cusparseZbsrilu02_analysis` | |
|`cusparseSbsrilu02` | |
|`cusparseDbsrilu02` | |
|`cusparseCbsrilu02` | |
|`cusparseZbsrilu02` | |
|`cusparseXbsrilu02_zeroPivot` | |
## ***7.3. Tridiagonal Solve***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSgtsv` | |
|`cusparseDgtsv` | |
|`cusparseCgtsv` | |
|`cusparseZgtsv` | |
|`cusparseSgtsv_nopivot` | |
|`cusparseDgtsv_nopivot` | |
|`cusparseCgtsv_nopivot` | |
|`cusparseZgtsv_nopivot` | |
|`cusparseSgtsv2_bufferSizeExt` | | 9.0 |
|`cusparseDgtsv2_bufferSizeExt` | | 9.0 |
|`cusparseCgtsv2_bufferSizeExt` | | 9.0 |
|`cusparseZgtsv2_bufferSizeExt` | | 9.0 |
|`cusparseSgtsv2` | | 9.0 |
|`cusparseDgtsv2` | | 9.0 |
|`cusparseCgtsv2` | | 9.0 |
|`cusparseZgtsv2` | | 9.0 |
|`cusparseSgtsv2_nopivot_bufferSizeExt` | | 9.0 |
|`cusparseDgtsv2_nopivot_bufferSizeExt` | | 9.0 |
|`cusparseCgtsv2_nopivot_bufferSizeExt` | | 9.0 |
|`cusparseZgtsv2_nopivot_bufferSizeExt` | | 9.0 |
|`cusparseSgtsv2_nopivot` | | 9.0 |
|`cusparseDgtsv2_nopivot` | | 9.0 |
|`cusparseCgtsv2_nopivot` | | 9.0 |
|`cusparseZgtsv2_nopivot` | | 9.0 |
## ***7.4. Batched Tridiagonal Solve***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSgtsvStridedBatch` | |
|`cusparseDgtsvStridedBatch` | |
|`cusparseCgtsvStridedBatch` | |
|`cusparseZgtsvStridedBatch` | |
|`cusparseSgtsv2StridedBatch_bufferSizeExt` | | 9.0 |
|`cusparseDgtsv2StridedBatch_bufferSizeExt` | | 9.0 |
|`cusparseCgtsv2StridedBatch_bufferSizeExt` | | 9.0 |
|`cusparseZgtsv2StridedBatch_bufferSizeExt` | | 9.0 |
|`cusparseSgtsv2StridedBatch` | | 9.0 |
|`cusparseDgtsv2StridedBatch` | | 9.0 |
|`cusparseCgtsv2StridedBatch` | | 9.0 |
|`cusparseZgtsv2StridedBatch` | | 9.0 |
|`cusparseSgtsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseDgtsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseCgtsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseZgtsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseSgtsvInterleavedBatch` | | 9.2 |
|`cusparseDgtsvInterleavedBatch` | | 9.2 |
|`cusparseCgtsvInterleavedBatch` | | 9.2 |
|`cusparseZgtsvInterleavedBatch` | | 9.2 |
## ***7.5. Batched Pentadiagonal Solve***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSgpsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseDgpsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseCgpsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseZgpsvInterleavedBatch_bufferSizeExt` | | 9.2 |
|`cusparseSgpsvInterleavedBatch` | | 9.2 |
|`cusparseDgpsvInterleavedBatch` | | 9.2 |
|`cusparseCgpsvInterleavedBatch` | | 9.2 |
|`cusparseZgpsvInterleavedBatch` | | 9.2 |
## **8. cuSPARSE Matrix Reorderings Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseScsrcolor` | |
|`cusparseDcsrcolor` | |
|`cusparseCcsrcolor` | |
|`cusparseZcsrcolor` | |
## **9. cuSPARSE Format Conversion Reference**
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSbsr2csr` | |
|`cusparseDbsr2csr` | |
|`cusparseCbsr2csr` | |
|`cusparseZbsr2csr` | |
|`cusparseSgebsr2gebsc_bufferSize` | |
|`cusparseSgebsr2gebsc_bufferSizeExt` | |
|`cusparseDgebsr2gebsc_bufferSize` | |
|`cusparseDgebsr2gebsc_bufferSizeExt` | |
|`cusparseCgebsr2gebsc_bufferSize` | |
|`cusparseCgebsr2gebsc_bufferSizeExt` | |
|`cusparseZgebsr2gebsc_bufferSize` | |
|`cusparseZgebsr2gebsc_bufferSizeExt` | |
|`cusparseSgebsr2gebsc` | |
|`cusparseDgebsr2gebsc` | |
|`cusparseCgebsr2gebsc` | |
|`cusparseZgebsr2gebsc` | |
|`cusparseSgebsr2gebsr_bufferSize` | |
|`cusparseSgebsr2gebsr_bufferSizeExt` | |
|`cusparseDgebsr2gebsr_bufferSize` | |
|`cusparseDgebsr2gebsr_bufferSizeExt` | |
|`cusparseCgebsr2gebsr_bufferSize` | |
|`cusparseCgebsr2gebsr_bufferSizeExt` | |
|`cusparseZgebsr2gebsr_bufferSize` | |
|`cusparseZgebsr2gebsr_bufferSizeExt` | |
|`cusparseXgebsr2gebsrNnz` | |
|`cusparseSgebsr2gebsr` | |
|`cusparseDgebsr2gebsr` | |
|`cusparseCgebsr2gebsr` | |
|`cusparseZgebsr2gebsr` | |
|`cusparseXgebsr2csr` | |
|`cusparseSgebsr2csr` | |
|`cusparseDgebsr2csr` | |
|`cusparseCgebsr2csr` | |
|`cusparseZgebsr2csr` | |
|`cusparseScsr2gebsr_bufferSize` | |
|`cusparseScsr2gebsr_bufferSizeExt` | |
|`cusparseDcsr2gebsr_bufferSize` | |
|`cusparseDcsr2gebsr_bufferSizeExt` | |
|`cusparseCcsr2gebsr_bufferSize` | |
|`cusparseCcsr2gebsr_bufferSizeExt` | |
|`cusparseZcsr2gebsr_bufferSize` | |
|`cusparseZcsr2gebsr_bufferSizeExt` | |
|`cusparseXcsr2gebsrNnz` | |
|`cusparseScsr2gebsr` | |
|`cusparseDcsr2gebsr` | |
|`cusparseCcsr2gebsr` | |
|`cusparseZcsr2gebsr` | |
|`cusparseXcoo2csr` |`hipsparseXcoo2csr` |
|`cusparseScsc2dense` | |
|`cusparseDcsc2dense` | |
|`cusparseCcsc2dense` | |
|`cusparseZcsc2dense` | |
|`cusparseScsc2hyb` | |
|`cusparseDcsc2hyb` | |
|`cusparseCcsc2hyb` | |
|`cusparseZcsc2hyb` | |
|`cusparseXcsr2bsrNnz` | |
|`cusparseScsr2bsr` | |
|`cusparseDcsr2bsr` | |
|`cusparseCcsr2bsr` | |
|`cusparseZcsr2bsr` | |
|`cusparseXcsr2coo` |`hipsparseXcsr2coo` |
|`cusparseScsr2csc` |`hipsparseScsr2csc` |
|`cusparseDcsr2csc` |`hipsparseDcsr2csc` |
|`cusparseCcsr2csc` |`hipsparseCcsr2csc` |
|`cusparseZcsr2csc` |`hipsparseZcsr2csc` |
|`cusparseCsr2cscEx` | | 8.0 |
|`cusparseCsr2cscEx2` | | 10.1 |
|`cusparseCsr2cscEx2_bufferSize` | | 10.1 |
|`cusparseScsr2dense` | |
|`cusparseDcsr2dense` | |
|`cusparseCcsr2dense` | |
|`cusparseZcsr2dense` | |
|`cusparseScsr2csr_compress` | | 8.0 |
|`cusparseDcsr2csr_compress` | | 8.0 |
|`cusparseCcsr2csr_compress` | | 8.0 |
|`cusparseZcsr2csr_compress` | | 8.0 |
|`cusparseScsr2hyb` |`hipsparseScsr2hyb` |
|`cusparseDcsr2hyb` |`hipsparseDcsr2hyb` |
|`cusparseCcsr2hyb` |`hipsparseCcsr2hyb` |
|`cusparseZcsr2hyb` |`hipsparseZcsr2hyb` |
|`cusparseSdense2csc` | |
|`cusparseDdense2csc` | |
|`cusparseCdense2csc` | |
|`cusparseZdense2csc` | |
|`cusparseSdense2csr` | |
|`cusparseDdense2csr` | |
|`cusparseCdense2csr` | |
|`cusparseZdense2csr` | |
|`cusparseSdense2hyb` | |
|`cusparseDdense2hyb` | |
|`cusparseCdense2hyb` | |
|`cusparseZdense2hyb` | |
|`cusparseShyb2csc` | |
|`cusparseDhyb2csc` | |
|`cusparseChyb2csc` | |
|`cusparseZhyb2csc` | |
|`cusparseShyb2csr` | |
|`cusparseDhyb2csr` | |
|`cusparseChyb2csr` | |
|`cusparseZhyb2csr` | |
|`cusparseShyb2dense` | |
|`cusparseDhyb2dense` | |
|`cusparseChyb2dense` | |
|`cusparseZhyb2dense` | |
|`cusparseSnnz` |`cusparseSnnz` |
|`cusparseDnnz` |`cusparseDnnz` |
|`cusparseCnnz` |`cusparseCnnz` |
|`cusparseZnnz` |`cusparseZnnz` |
|`cusparseCreateIdentityPermutation` |`hipsparseCreateIdentityPermutation` |
|`cusparseXcoosort_bufferSizeExt` |`hipsparseXcoosort_bufferSizeExt` |
|`cusparseXcoosortByRow` |`hipsparseXcoosortByRow` |
|`cusparseXcoosortByColumn` |`hipsparseXcoosortByColumn` |
|`cusparseXcsrsort_bufferSizeExt` |`hipsparseXcsrsort_bufferSizeExt` |
|`cusparseXcsrsort` |`hipsparseXcsrsort` |
|`cusparseXcscsort_bufferSizeExt` |`hipsparseXcscsort_bufferSizeExt` |
|`cusparseXcscsort` |`hipsparseXcscsort` |
|`cusparseCreateCsru2csrInfo` | |
|`cusparseDestroyCsru2csrInfo` | |
|`cusparseScsru2csr_bufferSizeExt` | |
|`cusparseDcsru2csr_bufferSizeExt` | |
|`cusparseCcsru2csr_bufferSizeExt` | |
|`cusparseZcsru2csr_bufferSizeExt` | |
|`cusparseScsru2csr` | |
|`cusparseDcsru2csr` | |
|`cusparseCcsru2csr` | |
|`cusparseZcsru2csr` | |
|`cusparseScsr2csru` | |
|`cusparseDcsr2csru` | |
|`cusparseCcsr2csru` | |
|`cusparseZcsr2csru` | |
|`cusparseHpruneDense2csr` | | 9.0 |
|`cusparseSpruneDense2csr` | | 9.0 |
|`cusparseDpruneDense2csr` | | 9.0 |
|`cusparseHpruneDense2csr_bufferSizeExt` | | 9.0 |
|`cusparseSpruneDense2csr_bufferSizeExt` | | 9.0 |
|`cusparseDpruneDense2csr_bufferSizeExt` | | 9.0 |
|`cusparseHpruneDense2csrNnz` | | 9.0 |
|`cusparseSpruneDense2csrNnz` | | 9.0 |
|`cusparseDpruneDense2csrNnz` | | 9.0 |
|`cusparseHpruneCsr2csr` | | 9.0 |
|`cusparseSpruneCsr2csr` | | 9.0 |
|`cusparseDpruneCsr2csr` | | 9.0 |
|`cusparseHpruneCsr2csr_bufferSizeExt` | | 9.0 |
|`cusparseSpruneCsr2csr_bufferSizeExt` | | 9.0 |
|`cusparseDpruneCsr2csr_bufferSizeExt` | | 9.0 |
|`cusparseHpruneCsr2csrNnz` | | 9.0 |
|`cusparseSpruneCsr2csrNnz` | | 9.0 |
|`cusparseDpruneCsr2csrNnz` | | 9.0 |
|`cusparseHpruneDense2csrByPercentage` | | 9.0 |
|`cusparseSpruneDense2csrByPercentage` | | 9.0 |
|`cusparseDpruneDense2csrByPercentage` | | 9.0 |
|`cusparseHpruneDense2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseSpruneDense2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseDpruneDense2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseHpruneDense2csrNnzByPercentage` | | 9.0 |
|`cusparseSpruneDense2csrNnzByPercentage` | | 9.0 |
|`cusparseDpruneDense2csrNnzByPercentage` | | 9.0 |
|`cusparseHpruneCsr2csrByPercentage` | | 9.0 |
|`cusparseSpruneCsr2csrByPercentage` | | 9.0 |
|`cusparseDpruneCsr2csrByPercentage` | | 9.0 |
|`cusparseHpruneCsr2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseSpruneCsr2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseDpruneCsr2csrByPercentage_bufferSizeExt` | | 9.0 |
|`cusparseHpruneCsr2csrNnzByPercentage` | | 9.0 |
|`cusparseSpruneCsr2csrNnzByPercentage` | | 9.0 |
|`cusparseDpruneCsr2csrNnzByPercentage` | | 9.0 |
|`cusparseSnnz_compress` |`hipsparseSnnz_compress` | 8.0 |
|`cusparseDnnz_compress` |`hipsparseDnnz_compress` | 8.0 |
|`cusparseCnnz_compress` |`hipsparseCnnz_compress` | 8.0 |
|`cusparseZnnz_compress` |`hipsparseZnnz_compress` | 8.0 |
## **10. cuSPARSE Generic API Reference**
## ***10.1. Generic Sparse API helper functions***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseCreateCoo` | | 10.1 |
|`cusparseCreateCooAoS` | | 10.1 |
|`cusparseCreateCsr` | | 10.1 |
|`cusparseDestroySpMat` | | 10.1 |
|`cusparseCooGet` | | 10.1 |
|`cusparseCooAoSGet` | | 10.1 |
|`cusparseCsrGet` | | 10.1 |
|`cusparseSpMatGetFormat` | | 10.1 |
|`cusparseSpMatGetIndexBase` | | 10.1 |
|`cusparseSpMatGetValues` | | 10.1 |
|`cusparseSpMatSetValues` | | 10.1 |
|`cusparseSpMatGetStridedBatch` | | 10.1 |
|`cusparseSpMatSetStridedBatch` | | 10.1 |
|`cusparseSpMatGetNumBatches` | | 10.1 |
|`cusparseSpMatSetNumBatches` | | 10.1 |
|`cusparseCreateSpVec` | | 10.1 |
|`cusparseDestroySpVec` | | 10.1 |
|`cusparseSpVecGet` | | 10.1 |
|`cusparseSpVecGetIndexBase` | | 10.1 |
|`cusparseSpVecGetValues` | | 10.1 |
|`cusparseSpVecSetValues` | | 10.1 |
## ***10.2. Generic Dense API helper functions***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseCreateDnMat` | | 10.1 |
|`cusparseDestroyDnMat` | | 10.1 |
|`cusparseDnMatGet` | | 10.1 |
|`cusparseDnMatGetValues` | | 10.1 |
|`cusparseDnMatSetValues` | | 10.1 |
|`cusparseDnMatSetStridedBatch` | | 10.1 |
|`cusparseDnMatGetStridedBatch` | | 10.1 |
|`cusparseCreateDnVec` | | 10.1 |
|`cusparseDestroyDnVec` | | 10.1 |
|`cusparseDnVecGet` | | 10.1 |
|`cusparseDnVecGetValues` | | 10.1 |
|`cusparseDnVecSetValues` | | 10.1 |
## ***10.3. Generic SpMM API functions***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSpMM` | | 10.1 |
|`cusparseSpMM_bufferSize` | | 10.1 |
## ***10.4. Generic SpVV API functions [Undocumented]***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSpVV` | | 10.1 |
|`cusparseSpVV_bufferSize` | | 10.1 |
## ***10.5. Generic SpMV API functions [Undocumented]***
| **CUDA** | **HIP** |**CUDA version\***|
|-----------------------------------------------------------|-------------------------------------------------|:----------------:|
|`cusparseSpMV` | | 10.1 |
|`cusparseSpMV_bufferSize` | | 10.1 |
\* CUDA version, in which API has appeared and (optional) last version before abandoning it; no value in case of earlier versions < 7.5.