Professional Documents
Culture Documents
Nvidia A40 Datasheet
Nvidia A40 Datasheet
NVIDIA A40
Powerful Data Center GPU For Visual Computing
The NVIDIA A40 accelerates the most demanding visual computing SPECIFICATIONS
workloads from the data center, combining the latest NVIDIA Ampere GPU architecture NVIDIA Ampere architecture
GPU memory 48 GB GDDR6 with ECC
architecture RT Cores, Tensor Cores, and CUDA® Cores with 48 GB Memory bandwidth 696 GB/s
of graphics memory. From powerful virtual workstations accessible Interconnect interface NVIDIA® NVLink® 112.5 GB/s
(bidirectional)3 PCIe Gen4: 64GB/s
from anywhere to dedicated render nodes, NVIDIA A40 brings next- NVIDIA Ampere architecture- 10,752
generation NVIDIA RTX ™ technology to the data center for the most based CUDA Cores
NVIDIA second-generation 84
advanced professional visualization workloads. RT Cores
NVIDIA third-generation 336
Tensor Cores
Peak FP32 TFLOPS (non-Tensor) 37.4
Up to 2X Faster Rendering Up to 40% Faster Graphics
Performance1 Performance1 Peak FP16 Tensor TFLOPS with 149.7 | 299.4*
FP16 Accumulate
Iray 2020.1 SPECviewperf 2020
Peak TF32 Tensor TFLOPS 74.8 | 149.6*
2.5X 1.6X
RT Core performance TFLOPS 73.1
1.4X
2.0X 1.4X Peak BF16 Tensor TFLOPS with 149.7 | 299.4*
2.0X 1.2X
FP32 Accumulate
1.5X 1.0X
0.8X
1X Peak INT8 Tensor TOPS 299.3 | 598.6*
Peak INT 4 Tensor TOPS 598.7 | 1,197.4*
1.0X 0.6X
1X Form factor 4.4" (H) x 10.5" (L) dual slot
0.4X
0.5X
0.2X Display ports 3x DisplayPort 1.4**; Supports
0 0
NVIDIA Mosaic and Quadro® Sync4
RTX 6000 A40 RTX 6000 A40 Max power consumption 300 W
Power connector 8-pin CPU
Thermal solution Passive
Up to 3X Faster AI Training Up to 50% Faster Single Precision Virtual GPU (vGPU) software NVIDIA vPC/vApps, NVIDIA RTX
Performance2 (FP32) HPC Performance2 support Virtual Workstation, NVIDIA Virtual
Compute Server
BERT pre-training throughput NAMD
vGPU profiles supported See the Virtual GPU Licensing Guide
8.0X 2.5X
NVENC | NVDEC 1x | 2x (includes AV1 decode)
2.0X
6.0X Secure and measured boot with Yes (optional)
6.4X 2.0X
hardware root of trust
1.5X
4.0X 1.5X NEBS ready Level 3
1.0X
3.4X Compute APIs CUDA, DirectCompute, OpenCL™,
1X
2.0X OpenACC®
0.5X
Graphics APIs DirectX 12.075, Shader Model 5.175,
1X
0 0 OpenGL 4.686, Vulkan 1.186
RTX 6000 A40 A100 RTX 6000 A40 A100
MIG support No
The NVIDIA A40 GPU delivers state-of-the-art visual computing capabilities, including real-time ray tracing, AI
acceleration, and multi-workload flexibility to accelerate deep learning, data science, and compute-based workloads.
Virtual workstations powered by NVIDIA A40 and NVIDIA RTX Virtual Workstation (vWS) and NVIDIA Virtual Compute
Server software benefit from extensive testing across a broad range of industry applications and professional software for
optimal performance and stability.
Learn more
© 2022 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, CUDA, GPUDirect, NVLink, OpenACC, 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 property of their respective owners. Mar22