NVIDIA A100 40 GB SXM
40 GB of HBM on the SXM4 form factor. A100 remains useful for BF16 training and supported inference, but cannot execute Hopper’s native FP8 Tensor Core path.
HBM2
Published peak, not measured application throughput
SXM requires a compatible server baseboard and cooling. An A100 PCIe listing is not interchangeable with this eight-GPU HGX configuration.
When this is a sensible choice
Start here if…
Choose the 40 GB part for established training jobs whose per-rank allocation fits. Four 40 GB devices are four separate memory budgets, not one 160 GB allocation.
Choose another configuration if…
Move to 80 GB when per-rank memory forces excessive sharding or checkpointing. An advertised 80 GB variant also changes memory bandwidth, not just capacity.
Specifications with their boundaries attached
- Architecture
- Ampere
- Memory
- 40 GB HBM2
- Memory bandwidth
- 1.56 TB/s per accelerator
- Peak compute
- 312 TFLOPS dense BF16/FP16; no native FP8 Tensor Cores
- Scale-up interconnect
- NVLink 3: 600 GB/s bidirectional; HGX fabric
- Host attachment
- PCIe 4.0 x16
- Power
- 400 W
- Partitioning
- Up to seven MIG instances
- Catalogue status
- Documented product
Compute figures are theoretical peaks at the stated precision. Dense and structured-sparse rates must not be mixed. Bandwidth labelled bidirectional combines both directions. See the source documents.
Three bandwidths, three different jobs
NVLink 3: 600 GB/s bidirectional; HGX fabric
InfiniBand, RoCE, EFA or provider-specific transport
Where it appears in provider documentation
| Provider / machine | Network scope | What changes the decision |
|---|---|---|
AWS P4d / P4de ↗ | EFA; verify size and cluster placement | P4d has A100 40GB; P4de has A100 80GB. Instance RAM is separate from accelerator memory. |
Google Cloud A2 Standard / A2 Ultra ↗ | VM-size-specific network | Standard uses 40GB A100; Ultra uses 80GB A100. GPU-to-GPU bandwidth is not the VM’s network bandwidth. |
Lambda On-demand GPU instances ↗ | Ordinary instance listing does not guarantee cluster fabric | B200 and H100 SXM come in multiple GPU counts. H100 PCIe is separately listed: choose the full product name. |
Model fit and software support
These publisher or serving-engine documents mention this hardware family. They have not been reproduced on our machines.
No model-specific recipe in our reviewed set certifies this exact hardware. The memory calculator can narrow candidates, but it cannot establish software support. Read the model register.
What is the memory floor?
Start with total parameters, then add the memory the workload needs. This arithmetic does not certify a serving configuration. All output sizes below are decimal GB.
Override device capacity with the memory exposed by your allocation, particularly for cloud B300 and partitioned devices. The starting 32 GB budget and 15% reserve are editable teaching assumptions. They are not measurements for the selected model. Mixed-precision tensors, quantization scales, vision encoders and draft models can increase the weight payload.
Raw weights only
320B × 8 bits ÷ 8
Budget per device
40 GB × (1 − 15%)
Arithmetic lower bound
round up ((weights + 32) ÷ budget)
This assumes perfectly balanced sharding. A result of three does not prove that a three-device parallel layout is supported. Check layer/expert divisibility, actual allocatable memory, precision kernels and the fabric before renting.
320B is the model card’s total; 18B active is not its storage size. The vLLM recipe reports about 306 GiB for the native FP8 checkpoint. Hopper requires BF16 KV for this model; the documented ROCm path targets gfx950, not every Instinct GPU. Read the model source ↗
For full training, also budget gradients, optimizer states, activations and communication buffers. The inference weight estimate above is insufficient.
Tokens per second, TTFT and TPOT are not certified for this hardware in our reference. Use the benchmark checklist to compare an exact model, software revision and workload.
Nearby memory capacities, different tradeoffs
These are comparison candidates selected by memory capacity, not performance rankings or drop-in replacements.
NVIDIA L40S
A 48 GB server card for inference, fine-tuning and graphics. The extra capacity is useful; it does not bring an HGX interconnect.
NVIDIA A10
A 24 GB graphics-capable server GPU. It offers more memory than T4, without the HBM or switched links of A100.
NVIDIA L4
A low-power 24 GB inference and media GPU. More compute does not imply more memory bandwidth: its published 300 GB/s is below A10.
Sources and review scope
Manufacturer and cloud documentation checked 2026-09-12. We reviewed specifications and the stated product boundaries; we did not run training or serving benchmarks on this device. Cloud memory and availability can differ by configuration.
- NVIDIA A100 datasheet ↗ · checked 2026-09-12
Cite this reference
AI Infra Interviews. NVIDIA A100 40 GB SXM: specifications and workload fit. Reviewed . https://aiinfrainterviews.com/hardware/a100-40-sxm
Include your access date when citing a changing specification. Link to the specification section for hardware figures or the explanation for a sizing or workload decision.
Research method and limits
We compile manufacturer specifications, cloud documentation, model cards and serving recipes. The reference preserves source units and distinguishes individual devices from nodes and racks. Conflicting figures and unknown fields remain labelled.
Our contribution is the comparison, unit reconciliation, worked arithmetic and workload explanation. Published peaks are vendor specifications. Calculator results are estimates under the displayed assumptions. Neither is a measurement from our own accelerator lab.
For a manufacturer’s specification, consult the original source documents. Cite our page when using its analysis, and retain primary-source attribution for underlying figures. This curated reference does not establish market share, live availability or a universal performance ranking.
Found a discrepancy? Send a correction with the page URL, exact variant, disputed figure and supporting primary document.