{{ 'fb_in_app_browser_popup.desc' | translate }} {{ 'fb_in_app_browser_popup.copy_link' | translate }}
{{ 'in_app_browser_popup.desc' | translate }}
The NVIDIA A100-40G, a data center-grade AI inference and training GPU, offers optimized performance for mid-range computational needs.
Not enough stock.
Your item was not added to your cart.
Not enough stock.
Please adjust your quantity.
{{'products.quick_cart.out_of_number_hint'| translate}}
{{'product.preorder_limit.hint'| translate}}
Limit {{ product.max_order_quantity }} per order.
Only {{ quantityOfStock }} item(s) left.
The NVIDIA A100-40G, a data center-grade AI inference and training GPU, offers optimized performance for mid-range computational needs. This powerful processor ensures efficient performance, making it ideal for engineers and data scientists tackling complex machine learning tasks and other demanding workflows.
Features | Nvidia A100-40G | Nvidia A100-80G | Nvidia V100 | AMD Instinct MI100 | Google TPU v4 | Intel Habana Gaudi | Nvidia T4 |
---|---|---|---|---|---|---|---|
Memory Size | 40 GB | 80 GB | 32 GB | 32 GB | 16 GB | 32 GB | 16 GB |
Memory Bandwidth | 1,555 GB/s | 2,039 GB/s | 900 GB/s | 1,232 GB/s | 700 GB/s | 1,023 GB/s | 300 GB/s |
Processing Power | 19.5 TFLOPS | 15.7 TFLOPS | 14 TFLOPS | 11.5 TFLOPS | N/A | N/A | 8.1 TFLOPS |
Inference Efficiency | High | High | Medium | Low | Medium | Medium | Medium |
Training Performance | Optimal | Optimal | Strong | Mid-tier | High | High | Low |
Use Case | Data Centers | Heavy Workloads | Research | Enterprise | Cloud Services | AI Development | Small Data Centers |
Accessory Model | Description |
---|---|
HGX A100 4-GPU Baseboard | Multi-GPU expansion for larger workloads |
Mellanox ConnectX-6 VPI | High-speed networking adapter |
NVIDIA NVSwitch | Interconnect for seamless multi-GPU communication |
Nvidia A100-40G Specifications | |
Model | A100-40G |
Memory | 40 GB HBM2 |
Memory Bandwidth | 1.6 TB/s |
CUDA Cores | 6,912 |
Tensor Cores | 432 |
TDP | 400W |
Architecture | Ampere GA100 |
NVLink Bandwidth | 600 GB/s |
Multi-Instance GPU (MIG) | Yes, Up to 7 instances |
Process Technology | 7nm |
Base Clock Speed | 765 MHz |
Boost Clock Speed | 1410 MHz |
PCI Express Generation | PCIe 4.0 |
DirectX | 12.0 |
OpenGL | 4.6 |
Form Factor | Dual-slot, full-height |
Interface | PCIe 4.0 x16 |
Number of GPUs | 1 |
Max GPU Temperature | 85°C |
Display Support | N/A (Data Center GPU) |
Cooling Solution | Passive |