cuda10.1

Version:: 10.1.243
Category:: tools
Cluster:: Loki

Author / Distributor

https://developer.nvidia.com/cuda-toolkit-archive

Description

The NVIDIA CUDA® 10.1 suite is organized into modular components to allow fine-grained loading of GPU development tools and libraries.

Available modules:

cuda10.1/toolkit/10.1.243 The core development environment for compiling and running CUDA programs.
cuda10.1/blas/10.1.243 A GPU-accelerated implementation of Basic Linear Algebra Subprograms (cuBLAS).
cuda10.1/fft/10.1.243 A GPU-accelerated Fast Fourier Transform library (cuFFT).
cuda10.1/nsight/10.1.243 A low-overhead performance analysis tool for optimizing GPU/CPU applications.
cuda10.1/profiler/10.1.243 A visualization and profiling tool (nvprof) for analyzing CUDA applications.

Each module can be loaded individually based on your development needs. These components target GPU acceleration and scientific computation workflows.

Documentation

Toolkit: https://docs.nvidia.com/cuda/archive/10.1/
cuBLAS Guide: https://docs.nvidia.com/cuda/archive/10.1/cublas
cuFFT Guide: https://docs.nvidia.com/cuda/archive/10.1/cufft
Nsight Compute: https://docs.nvidia.com/nsight-compute/
CUDA Profiler: https://docs.nvidia.com/cuda/archive/10.1/profiler-users-guide/

Examples/Usage

Load the CUDA toolkit:

$ module load cuda10.1/toolkit/10.1.243

Compile a CUDA program:

$ nvcc -o vecAdd vecAdd.cu

Run the program:

$ ./vecAdd

Load the cuBLAS or cuFFT modules for linking against specific libraries:

$ module load cuda10.1/blas/10.1.243
$ module load cuda10.1/fft/10.1.243

Use profiling tools:

$ module load cuda10.1/profiler/10.1.243
$ nvprof ./vecAdd

Use Nsight Compute:

$ module load cuda10.1/nsight/10.1.243
$ ncu ./vecAdd

Unload all modules:

$ module unload cuda10.1/toolkit/10.1.243
$ module unload cuda10.1/blas/10.1.243
$ module unload cuda10.1/fft/10.1.243
$ module unload cuda10.1/nsight/10.1.243
$ module unload cuda10.1/profiler/10.1.243

Installation

Source code is obtained from CUDA 10.1