A40 Specifications
Ampere architecture with dedicated ray tracing cores, Tensor Cores, and 48 GB GDDR6 for professional visualization and lightweight AI workloads.
48 GB
GDDR6 Memory
Large frame buffer for complex scenes and datasets
696 GB/s
Memory Bandwidth
Sustained throughput for rendering and simulation
84
RT Cores
Second-gen hardware ray tracing at 73 TFLOPS
149.7
FP16 TFLOPS
299.4 TFLOPS with sparsity enabled
NVIDIA A40 on Denvr AI Cloud
Simulation Environments
48 GB GDDR6 with dedicated ray tracing cores for 3D rendering, simulation, and visual computing workloads. Built for graphics-intensive pipelines.
vGPU Workstations
Powerful virtual workstation instances for remote users, enabling high-end remote design, AI, and compute workloads.
Lightweight Inference
Serve smaller models and embedding pipelines where HBM bandwidth isn't required. A40 delivers capable inference at a lower price point than A100.
Managed Storage
High-performance Weka filesystem and local NVMe available. No external storage to provision for datasets, checkpoints, or model artifacts.
Platform
GPUs
GPU VRAM
vCPUs
Memory
Local Storage
Interconnect
On-Demand
NVIDIA A40
4
48 GB
128
512 GB
2x 3.8TB NVMe
-
$0.65 / GPU
Configurations
Per-minute billing with on-demand and reserved options. All configurations available as bare metal, VM, or model endpoints.
Related GPUs
Compare Denvr GPU options by workload and performance requirements.
Optimized For
Very small model serving
Visualization, simulation, lightweight inference
Single-GPU inference, moderate training
VRAM
20 GB
48 GB
40 GB
Memory Bandwidth
800 GB/s
696 GB/s
1,600 GB/s
FP64/FP32
-
-
19.5 TFLOPS
FP16
156 TFLOPS
149.7 TFLOPS
312 TFLOPS
FP8
-
-
-
NVLink
-
-
-
On-Demand Pricing
$0.58 / GPU
$0.65 / GPU
$1.15 / GPU
Infrastructure you can trust at scale
As an NVIDIA Cloud Partner we build and operate AI clusters following NVIDIA Reference Architectures. Your models and data are supported via strict privacy safeguards and SOC 2 Type 2 security practices.
