1. What's New
This is the first release of the NVIDIA HPC SDK, a comprehensive suite of compilers and libraries enabling developers to program the entire HPC platform, from the GPU foundation to the CPU and through the interconnect.
Key features of the 20.7 GA release of the NVIDIA HPC SDK for Linux include:
- Support for NVIDIA Ampere Architecture GPUs with FP16, TF32 and FP64 tensor cores, and NVIDIA Volta tensor cores
- Supported on CUDA 11.0, 10.2 and 10.1 for compute capability 3.5 and up
- Cross-platform support for x86-64, OpenPOWER and Arm Server multicore CPUs
- nvc++ ISO C++17 compiler with Parallel Algorithms acceleration on GPUs, OpenACC and OpenMP
- nvfortran ISO Fortran 2003 compiler with array intrinsics acceleration on GPUs, CUDA Fortran, OpenACC and OpenMP
- nvc ISO C11 compiler with OpenACC and OpenMP
- nvcc NVIDIA CUDA C++ compiler
- cuBLAS GPU-accelerated basic linear algebra subroutine (BLAS) library
- cuSOLVER GPU-accelerated dense and sparse direct solvers
- cuSPARSE GPU-accelerated BLAS for sparse matrices
- cuFFT GPU-accelerated library for Fast Fourier Transforms
- cuTENSOR GPU-accelerated tensor linear algebra library
- cuRAND GPU-accelerated random number generation (RNG)
- Thrust GPU-accelerated library of C++ parallel algorithms and data structures
- CUB cooperative threadblock primitives and utilities for CUDA kernel programming
- libcu++ opt-in heterogeneous CUDA C++ Standard Libarary
- NCCL library for fast multi-GPU/multi-node collective communications
- NVSHMEM library for fast GPU memory-to-memory transfers (OpenSHMEM compatible)
- OpenMPI GPU-aware message passing interface library
- NVIDIA Nsight Systems interactive HPC applications performance profiler
- NVIDIA Nsight Compute interactive GPU compute kernel performance profiler
- NVIDIA cuda-gdb for interactive GPU compute kernel performance debugging
- NVIDIA compute-sanitizer for memory safety checks
2. Release Component Versions
The NVIDIA HPC SDK 20.7 release contains the following versions of each component:
Linux_x86_64 | Linux_ppc64le | Linux_aarch64 | |||||||
---|---|---|---|---|---|---|---|---|---|
CUDA 10.1 | CUDA 10.2 | CUDA 11.0 | CUDA 10.1 | CUDA 10.2 | CUDA 11.0 | CUDA 10.1 | CUDA 10.2 | CUDA 11.0 | |
nvc++ | 20.7 | 20.7 | 20.7 | ||||||
nvc | 20.7 | 20.7 | 20.7 | ||||||
nvfortran | 20.7 | 20.7 | 20.7 | ||||||
nvcc | 10.1.243 | 10.2.89 | 11.0.109 | 10.1.243 | 10.2.89 | 11.0.109 | N/A | N/A | 11.0.109 |
NCCL | 2.7.3-1 | 2.7.3-1 | 2.7.3-1 | 2.7.3-1 | 2.7.3-1 | 2.7.3-1 | N/A | N/A | 2.7.3-1 |
NVSHMEM | 1.0.0-0 | 1.0.0-0 | 1.0.0-0 | 1.0.0-0 | 1.0.0-0 | 1.0.0-0 | N/A | N/A | N/A |
cuBLAS | 10.2.1.243 | 10.2.2.89 | 11.1.0.229 | 10.2.1.243 | 10.2.2.89 | 11.1.0.229 | N/A | N/A | 11.1.0.229 |
cuFFT | 10.1.1.243 | 10.1.2.89 | 10.2.0.218 | 10.1.1.243 | 10.1.2.89 | 10.2.0.218 | N/A | N/A | 10.2.0.218 |
cuRAND | 10.1.1.243 | 10.1.2.89 | 10.2.1.218 | 10.1.1.243 | 10.1.2.89 | 10.2.1.218 | N/A | N/A | 10.2.1.218 |
cuSOLVER | 10.2.0.243 | 10.3.0.89 | 10.5.0.218 | 10.2.0.243 | 10.3.0.89 | 10.5.0.218 | N/A | N/A | 10.5.0.218 |
cuSPARSE | 10.3.0.243 | 10.3.1.89 | 11.1.0.218 | 10.3.0.243 | 10.3.1.89 | 11.1.0.218 | N/A | N/A | 11.1.0.218 |
cuTENSOR | 1.0.2 | 1.0.2 | 1.1.0 | 1.0.2 | 1.0.2 | 1.1.0 | N/A | N/A | 1.1.0 |
Nsight Compute | 2020.1.0.33-28294165 | 2020.1.0.33-28294165 | 2020.1.0.33-28294165 | ||||||
Nsight Systems | 2020.3.1.54 CLI and GUI | 2020.3.1.54 CLI | 2020.3.1.54 CLI | ||||||
OpenMPI | 3.1.5 | 3.1.5 | 3.1.5 | ||||||
OpenBLAS | 0.3.7 | 0.3.7 | 0.3.7 | ||||||
Scalapack | 2.1.0 | 2.1.0 | 2.1.0 | ||||||
Thrust | 1.9.6-1 | 1.9.7 | 1.9.9 | 1.9.6-1 | 1.9.7 | 1.9.9 | 1.9.6-1 | 1.9.7 | 1.9.9 |
CUB | N/A | N/A | 1.9.9 | N/A | N/A | 1.9.9 | N/A | N/A | 1.9.9 |
libcu++ | N/A | 1.0.0 | 2.0.0 | N/A | 1.0.0 | 2.0.0 | N/A | 1.0.0 | 2.0.0 |
3. Supported Platforms
3.1. Platform Requirements for the HPC SDK
Architecture | Linux Distributions | Minimum gcc/glibc Toolchain | Minimum CUDA Driver |
---|---|---|---|
x86_64 |
CentOS 6.6, 6.7, 6.8, 6.9, 6.10 |
C99: 4.4 |
418.39 |
ppc64le |
RHEL 7.3, 7.4, 7.5, 7.6, 7.7, 8.0, 8.1 |
C99: 4.4 |
410.45 |
aarch64 |
RHEL 8.1 |
C99: 4.4 |
450.36 |
3.2. Supported CUDA Toolkit Versions
The NVIDIA HPC SDK uses elements of the CUDA toolchain when building programs for execution with NVIDIA GPUs. Every HPC SDK installation package puts the required CUDA components into an installation directory called [install-prefix]/[arch]/[nvhpc-version]/cuda.
An NVIDIA CUDA GPU device driver must be installed on a system with a GPU before you can run a program compiled for the GPU on that system. The NVIDIA HPC SDK does not contain CUDA Drivers. You must download and install the appropriate CUDA Driver from NVIDIA .
The nvaccelinfo tool prints the Driver version as its first line of output. You can use it to find out which version of the CUDA Driver is installed on your system.
- CUDA 10.1
- CUDA 10.2
- CUDA 11.0
4. Known Limitations
- When the -stdpar option is used on a machine with multiple GPUs installed with differing compute capabilities, a target compute capability must be selected by the user at compile time by using the flag '-gpu=ccXY' where 'XY' specifies the two-digit compute capability version. This requirement will be lifted in the upcoming 20.9 release.
- The cuda-gdb debugger is included in this release. Currently, Fortran arrays with non-constant bounds are not handled correctly and querying values will yield incorrect results. Stepping through cuda-fortran and OpenACC kernels is partially supported, but incorrect line numbers are displayed. For additional general limitations with cuda-gdb, please refer to its documentation.
- When using -stdpar to accelerate C++ parallel algorithms, the algorithm calls cannot include virtual function calls or function calls through a function pointer, cannot use C++ exceptions, can only dereference pointers that point to the heap, and must use random access iterators (raw pointers as iterators work best)
Notices
Notice
ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND OTHER DOCUMENTS (TOGETHER AND SEPARATELY, "MATERIALS") ARE BEING PROVIDED "AS IS." NVIDIA MAKES NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO THE MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT, MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE.
Information furnished is believed to be accurate and reliable. However, NVIDIA Corporation assumes no responsibility for the consequences of use of such information or for any infringement of patents or other rights of third parties that may result from its use. No license is granted by implication of otherwise under any patent rights of NVIDIA Corporation. Specifications mentioned in this publication are subject to change without notice. This publication supersedes and replaces all other information previously supplied. NVIDIA Corporation products are not authorized as critical components in life support devices or systems without express written approval of NVIDIA Corporation.
Trademarks
NVIDIA, the NVIDIA logo, CUDA, CUDA-X, GPUDirect, HPC SDK, NGC, NVIDIA Volta, NVIDIA DGX, NVIDIA Nsight, NVLink, NVSwitch, and Tesla are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries. Other company and product names may be trademarks of the respective companies with which they are associated.