NVIDIA A100 PCIe 80GB

Provider / plan	Price / GPU-hour	Rental type	Terms & configuration	Last checked	Provider pricing link
RunpodSecure Cloud Pods	$1.59Lowest flexible rate	On-demandNo reservation	1 GPU minimumSee provider View details Secure Cloud Pods published GPU-hour rate. CPU/RAM allocation varies; storage/networking extra. Not live capacity. Published rate $1.59 / GPU-hour CPU / vCPU Confirm with provider System RAM Confirm with provider Storage Confirm with provider Network / egress Confirm with provider GPU allocation Confirm with provider GPU topology Confirm with provider	2026-10-03Auto-collected	View pricing ↗

A100 PCIe 80GB offers Ampere compute and large HBM2e capacity in a PCIe card. It is a candidate for training, fine-tuning and inference when memory capacity matters more than access to newer precision formats. The PCIe and SXM editions should be compared separately.

CHOOSING THIS GPU

Is it right for your workload?

80 GB PCIe reference edition with 300 W maximum TDP. The SXM 80 GB edition uses a different power envelope and memory bandwidth.

Best suited to

80 GB-class model inference
BF16 or FP16 fine-tuning
Training in compatible PCIe server configurations

Strengths

80 GB memory in a PCIe form factor
Hardware MIG support
NVLink bridge capability

Things to consider

No native FP8 Tensor Core acceleration
Bridge and peer-to-peer access depend on the server
Lower memory bandwidth than A100 SXM 80GB

HARDWARE DETAILS

Inside the GPU

Manufacturer: NVIDIA
Architecture: Ampere
VRAM (GB): 80
Memory type: HBM2e
Form factor: PCIe
Memory bandwidth (GB/s): 1935
Maximum board power (W): 300
CUDA cores: 6912
Tensor cores: 432 / 3rd generation
Ray tracing cores: Not verified for this edition
CUDA capability: 8.0
Memory ECC: Not verified for this edition
Host interface: PCIe Gen4
GPU interconnect: 3rd generation; 600 GB/s aggregate bidirectional; bridge required
Hardware partitioning (MIG): Up to 7 hardware instances; provider must enable it
Cooling: Passive; server airflow required

Hardware capabilities are not a guarantee of access in a cloud instance. Check the exact edition, GPU allocation and server topology in the rental offer.

Theoretical compute performance

Precision	Dense peak	Sparse peak
FP32	19.5 TFLOPS	Not applicable
FP16 Tensor	312 TFLOPS	624 TFLOPS
BF16 Tensor	312 TFLOPS	624 TFLOPS
INT8 Tensor	624 TOPS	1248 TOPS
FP8 Tensor	Not supported	Not supported

Manufacturer peak rates. Sparse rates require supported structured sparsity. TFLOPS and TOPS are different units; these figures are not application benchmarks.

Unlisted rates are not verified here; that does not mean a precision is unsupported. TFLOPS measures floating-point operations, while TOPS measures integer operations in the INT8 rows.

MODEL MEMORY

Plan your model size

Estimate · not a benchmark

70B 8-bit weights alone are about 70 GB; a full runtime needs additional memory. Compare quantized single-GPU serving with supported multi-GPU execution. Full fine-tuning requires substantially more than inference weights.

Example model class	Weight format	Raw weights ≈	Minimum GPUs for weights only
Llama 3.1 8B ↗	4-bit	4 GB	1Runtime needs more memory
Llama 3.1 8B ↗	FP16	16 GB	1Runtime needs more memory
Qwen2.5 14B ↗	4-bit	7 GB	1Runtime needs more memory
Qwen2.5 32B ↗	4-bit	16 GB	1Runtime needs more memory
Llama 3.1 70B ↗	4-bit	35 GB	1Runtime needs more memory
Llama 3.1 70B ↗	8-bit	70 GB	1Runtime needs more memory
Llama 3.1 70B ↗	FP16	140 GB	2Runtime needs more memory

Arithmetic lower bound: rounded parameter count × bits per weight ÷ 8, in decimal GB. Excludes quantization metadata, KV cache, activations, CUDA workspaces and training states. Actual model sizes differ from their rounded names. No context length, batch size or concurrency is guaranteed. GPU counts assume supported model sharding and can be higher in practice. Quantized checkpoints and compatible kernels are required for 4-bit/8-bit execution.

SOFTWARE & PERFORMANCE

Before you deploy

Software compatibility

Use an NVIDIA CUDA-enabled framework/container compatible with this GPU and the host driver. Check PyTorch build and inference-engine kernel requirements before deployment. vLLM documents NVIDIA compute capability 7.5+ as a baseline; support for each model and quantization method still needs checking.

Measured benchmarks

A reproducible, comparable application benchmark has not yet been recorded for this edition. Peak TFLOPS above are not measured tokens per second or image-generation speed.

Compare tests using the same model, precision, GPU count, engine version, input/output length and batch or concurrency. A provider’s server configuration can change the result.

COMPARE YOUR OPTIONS

Alternatives to consider

NVIDIA A100 SXM 80GB ↗

80 GB · Ampere

Compare SXM topology and memory bandwidth.

NVIDIA H100 PCIe ↗

80 GB · Hopper

Compare Hopper features in a PCIe deployment.

Technical sources & review date

Specifications reviewed 2026-09-29. Workload recommendations are editorial guidance based on the hardware capabilities. Model-memory figures are arithmetic estimates.

Reference hardware; rental provider cooling configuration may differ. Hardware image source ↗

BUDGET YOUR RUN

GPU cost estimator

GPU compute only: hourly rate × billable hours × GPU count. CPU, RAM, storage, egress, taxes, billing increments and minimum terms may add charges. Serverless hours mean active GPU task time; spot capacity can be interrupted. Reserved commitments and quote-only offers are excluded from this simple hourly estimator.

Compare rental prices