Support Matrix#

GPU, CUDA Toolkit, and CUDA Driver Requirements#

The following sections highlight the compatibility of NVIDIA cuDNN versions with the various supported NVIDIA CUDA Toolkit, CUDA driver, and NVIDIA hardware versions.

Supported NVIDIA Hardware and CUDA Version#

cuDNN Package[1]

CUDA Toolkit Version

Supports static linking?[2]

NVIDIA Driver Version for Linux

NVIDIA Driver Version for Windows

CUDA Compute Capability

Supported NVIDIA GPU Architectures

cuDNN 9.17.0 for CUDA 13.x

  • 13.1

  • 13.0

Yes

>=580.65.06

N/A

  • 12.0

  • 10.0

  • 9.0

  • 8.9

  • 8.6

  • 8.0

  • 7.5

  • NVIDIA Blackwell

  • NVIDIA Hopper

  • NVIDIA Ada Lovelace

  • NVIDIA Ampere

  • NVIDIA Turing

cuDNN 9.17.0 for CUDA 12.x

  • 12.9

  • 12.8

  • 12.6

  • 12.5

  • 12.4

  • 12.3

  • 12.2

  • 12.1

  • 12.0

Yes

>=525.60.13

>=527.41

  • NVIDIA Blackwell[3]

  • NVIDIA Hopper

  • NVIDIA Ada Lovelace

  • NVIDIA Ampere

  • NVIDIA Turing

Note

For best performance, the recommended configuration is cuDNN 9.17.0 with CUDA 13.1. This is the configuration used for tuning heuristics.

CPU Architecture and OS Requirements#

The following tables highlight the compatibility of cuDNN versions with the various supported OS versions.

Linux#

Refer to the following table to view the list of supported Linux versions for cuDNN.

Linux Versions for cuDNN#

Architecture

OS Name

OS Version

Distro Information: Kernel

Distro Information: GCC

Distro Information: Glibc

x86_64

RHEL

10.x

5.4.0

14.2.1

2.39

x86_64

RHEL

9.x

5.14.0

11.4.1

2.34

x86_64

RHEL

8.x

4.18.0

8.5.0

2.28

x86_64

Rocky

10.0

5.4.0

14.2.1

2.39

x86_64

Rocky

9.3

5.14.0

11.4.1

2.34

x86_64

Rocky

8.6

4.18

8.5.0

2.28

x86_64

Ubuntu

24.04

6.8.0

13.2.0

2.39

x86_64

Ubuntu

22.04

6.2.0

11.4.0

2.35

x86_64

Debian

12.1

6.1.0

12.2.0

2.36

x86_64

SLES

15

6.4.0

7.5.0

2.38

x86_64

OpenSuse

15

6.4.0

7.5.0

2.38

AArch64 SBSA

RHEL

10.x

5.4.0

14.2.1

2.39

AArch64 SBSA

RHEL

9.x

5.14.0

11.4.1

2.34

AArch64 SBSA

RHEL

8.x

4.18

8.5.0

2.28

AArch64 SBSA

Ubuntu

24.04

6.8.0

13.2.0

2.39

AArch64 SBSA

Ubuntu

22.04

5.15.0

11.4.0

2.35

AArch64 SBSA

Debian

12.1

6.1.0

12.2.0

2.36

AArch64 for Jetson

Ubuntu

24.04

6.8.0

13.2.0

2.39

AArch64 for Jetson

Ubuntu

22.04

4.15.0

7.3.1

2.35

Note

  • For platforms that ship a compiler version older than GCC 6 by default, linking to static cuDNN using the default compiler is not supported.

  • For RHEL 8.9 and Rocky 8.9 Linux, the R525 and later display driver is needed.

  • The Conda package for cuDNN can be installed on any combination of architecture and OS in the preceding table.

Windows#

Windows 10, Windows 11, and Windows Server 2022 are supported. Refer to the following table to view the list of supported Visual Studio versions for cuDNN.

Visual Studio Versions Based on Your Version of CUDA#

CUDA 13.x, 12.x

Visual Studio

2022, 2019

Note

The Conda package for cuDNN can be installed on any supported Windows version.

Footnotes

[1]

The cuDNN build for CUDA 12.x is compatible with CUDA 12.x for all x, including future CUDA 12.x releases that ship after this cuDNN release. This applies to both the dynamic and static builds of cuDNN.

[2]

This column specifies whether the given cuDNN library can be statically linked against the CUDA toolkit for the given CUDA version. Dynamic linking is supported in all cases.

[3](1,2,3)

Requires CUDA >=12.8, Linux driver >=570.26, and Windows driver >=570.65.