Datasheet

NVIDIA RTX 4000

Ada Generation
Performance for endless possibilities.

Powering the Next Era of Innovation

Key Features
Industries are embracing accelerated computing and AI to tackle powerful
dynamics and unlock transformative possibilities. Generative AI is reshaping the > PCIe Gen4
way professionals create and innovate across various domains, from design and > Four DisplayPort 1.4a connectors
engineering to entertainment and healthcare.
> AV1 encode and decode support
The NVIDIA RTX™ 4000 Ada Generation is the most powerful single-slot GPU
for professionals, delivering remarkable acceleration for AI, real-time rendering, > DisplayPort with audio
graphics, and compute workloads to your desktop. Built on the NVIDIA Ada Lovelace > 3D stereo support with stereo
architecture, RTX 4000 combines 48 third-generation RT Cores, 192 fourth-generation connector
Tensor Cores, and 6,144 CUDA® cores with 20GB of graphics memory to effortlessly
> NVIDIA® GPUDirect® for Video
handle large datasets and complex visual workloads.
support
NVIDIA RTX professional graphics cards are certified with a broad range of
> NVIDIA GPUDirect remote direct
professional applications, tested by leading independent software vendors (ISVs)
memory access (RDMA) support
and workstation manufacturers, and backed by a global team of support specialists.
Get the peace of mind to focus on what matters with the premier visual computing > NVIDIA Quadro® Sync II¹
solution for mission-critical business. compatibility

> NVIDIA RTX Experience

Omniverse Rendering
3.0 2.0 > NVIDIA RTX Desktop Manager
2.5
software
2.6X 1.5
1.6X
2.0 > NVIDIA RTX IO support
1.5 1.0
> HDCP 2.2 support
1.0
0.5 > NVIDIA Mosaic² technology
0.5

0 0
NVIDIA RTX A4000 NVIDIA RTX 4000 NVIDIA RTX A4000 NVIDIA RTX 4000
Ada Generation Ada Generation
Relative Performance Relative Performance
Tests run on an Intel Core i9-12900K processor @ 3.2GHz (5.2GHz Tests run on an Intel Core i9-12900K processor @ 3.2GHz (5.2GHz
Turbo), 64GB RAM, Windows 11 Enterprise x64, NVIDIA Driver Turbo), 64GB RAM, Windows 11 Enterprise x64, Arnold v.6.0.2 RC2,
536.43. CAD application performance based on internal testing of NVIDIA Driver 528.42. Relative speedup for 1080p resolution, scene
NVIDIA Omniverse USD Composer with several models of varying sol subtest render time (seconds). Performance based on pre-
size and render complexity. Performance is measured as frames released build, subject to change.
rendered per second. NVIDIA DLSS 3 is enabled for NVIDIA RTX
4000 Ada Generation GPUs, DLSS 2 enabled for non-Ada Generation
GPUs. Performance based on pre-release build, subject to change.

NVIDIA RTX 4000 Ada Generation | Datasheet | 1

 Graphics Specifications
2.0
GPU Memory 20GB GDDR6
1.5 Memory Interface 160 bit
1.5X

1.0
Memory Bandwidth 360GB/s

Error Correcting Code (ECC) Yes

0.5
NVIDIA Ada Lovelace Architecture-Based 6,144
0 CUDA Cores
NVIDIA RTX A4000 NVIDIA RTX 4000
Ada Generation
Relative Performance
NVIDIA Fourth-Generation Tensor Cores 192
Tests run on an Intel Core i9-12900K processor @ 3.2GHz (5.2GHz
Turbo), 64GB RAM, Windows 11 Enterprise x64, NVIDIA Driver NVIDIA Third-Generation RT Cores 48
528.49. Relative speedup for 4K geomean score. Performance
based on pre-release build, subject to change.
Single-Precision Performance 26.7 TFLOPS³

RT Core Performance 61.8 TFLOPS³

Generative AI
Tensor Performance 327.6 TFLOPS4
1.5

System Interface PCIe 4.0 x16

1.0
1.2X Power Consumption Total board power: 130W

Thermal Solution Active

0.5
Form Factor 4.4” H x 9.5” L, single slot

Display Connectors 4x DisplayPort 1.4a5

0
NVIDIA RTX A4000 NVIDIA RTX 4000
Ada Generation
Max Simultaneous Displays 4x 4096 x 2160 @ 120Hz
Relative Performance
4x 5120 x 2880 @ 60Hz
Tests run on an Intel Core i9-12900K processor @ 3.2GHz (5.2GHz
Turbo), 64GB RAM, Windows 11 Enterprise x64, Stable Diffusion 2x 7680 x 4320 @ 60Hz
WebUI v1.3.1, NVIDIA Driver 536.15. Relative speedup for
512x512 image generation. Performance based on pre-released
build, subject to change. Encode/Decode Engines 2x encode, 2x decode (+AV1 encode
and decode)

Inference VR Ready Yes

2.0
Graphics APIs DirectX 12, Shader Model 6.7,
OpenGL 4.66, Vulkan 1.36
1.5
1.6X
Compute APIs CUDA 12.2, OpenCL 3.0,
1.0 DirectCompute

0.5
NVIDIA NVLink ® No

0
NVIDIA RTX A4000 NVIDIA RTX 4000
Ada Generation
Relative Performance
Tests run on an Intel Core i9-12900K processor @ 3.2GHz (5.2GHz
Turbo), 64GB RAM, Windows 11 Enterprise x64, TensorRT v8.5.1,
NVIDIA Driver 525.85. Relative speedup for ResNet-50 V1.5
Inference, mixed precision. Performance based on pre-release
build, subject to change.

Ready to Get Started?

To learn more about NVIDIA RTX 4000, visit:
www.nvidia.com/rtx-4000
1 Quadro Sync II card sold separately. I 2 Windows 10 and Linux. I 3 Peak rates based on GPU boost clock. I 4
Effective FP8 teraFLOPS (TFLOPS) using sparsity. I 5 Display ports are on by default for RTX 4000. | 6 Product is
based on a published Khronos specification and is expected to pass the Khronos conformance testing process
when available. Current conformance status can be found at www.khronos.org/conformance

© 2023 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, CUDA, GPUDirect, NVLink, Quadro, and
RTX 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. All
other trademarks are the property of their respective owners. 2895141. SEP23