| title | gpu_math: gpu_math |
|---|---|
| description | NVidia CUDA implementation. |
| keywords | neural network, math, gpu |
| author | WilliamDAssafMSFT |
| ms.author | wiassaf |
| manager | cgronlun |
| ms.date | 07/15/2019 |
| ms.topic | reference |
| ms.prod | sql |
| ms.technology | machine-learning-services |
| ms.service | |
| ms.assetid | |
| ROBOTS | |
| audience | |
| ms.devlang | Python |
| ms.reviewer | |
| ms.suite | |
| ms.tgt_pltfrm | |
| ms.custom | |
| monikerRange | >=sql-server-2017||>=sql-server-linux-ver15 |
microsoftml.gpu_math(gpu_id: numbers.Real = -1,
cu_dnn: bool = False, cu_dnn_algo: str = 'ImplicitPrecompGemm')
NVidia CUDA implementation.
GPU device id (settings).
Use cuDNN on GPU (settings).
cuDNN optimization options (settings).