NVIDIA

A10

A10 — illustration of the card's form factor
VRAM
24GB
FP32 TFLOPS
31.2 TFLOPS
TDP
150 W
Memory
GDDR6

Provider Marketplace

Cheapest
$0.13/hour
Starting from
OVHcloud logo
OVHcloudVisit
Best Value
$1.10/hour
Starting from
Modal logo
ModalVisit
Enterprise Choice
$1.42/hour
Starting from
Runcrate logo
Runcrate

All Cloud Providers

3 Options available
OVHcloud logo
OVHcloudCheapest
On-Demand
$0.13/ hour
Estimated Cost
Provision
Modal logo
Modal
On-Demand
$1.10/ hour
Estimated Cost
Provision
Runcrate logo
Runcrate
On-Demand
$1.42/ hour
Estimated Cost
View Provider

Estimates only — rates are collected automatically from public provider pages and may be out of date. Prices vary by region, commitment term, and availability, and typically exclude storage, egress, and tax. Confirm current pricing with the provider before purchasing. Last collected .

Compute Performance

FP3231.2 TFLOPS
TF3262.5 TFLOPS
FP16125 TFLOPS
BF16125 TFLOPS
INT8250 TOPS
INT4500 TOPS

Architecture

MicroarchitectureAmpere
Tensor Coresthird-generation Tensor Cores
RT Coressecond-generation, 72 RT Cores
Sparse AccelerationSupported (structural sparsity)
Dynamic PrecisionSupported (FP32/TF32/BF16/FP16/INT8/INT4)

Memory & VRAM

Memory TypeGDDR6
Total Capacity24GB
Bandwidth600GB/s

Connectivity & Scaling

InterconnectPCIe
GenerationPCIe Gen4
IB Bandwidth64 GB/s
PCIe InterfacePCIe Gen4

Virtualization

vGPU ReadinessSupported (NVIDIA Virtual PC, NVIDIA Virtual Applications, NVIDIA RTX Virtual Workstation, NVIDIA Virtual Compute Server, NVIDIA AI Enterprise)
Virt EfficiencyNear bare-metal (vendor claim)

Power & Efficiency

TDP150 W
ConnectorsPEX 8-pin

Physical Design

Form FactorPCIe
FHFLYes
Slot WidthSingle-slot

Software Ecosystem

DockerNVIDIA NGC catalog / NGC Containers (examples: 21.02-py3, 20.12-py3)
Driver StabilityRTX Enterprise drivers with extensive testing and ISV certifications (over 100 ISVs)

Server & Deployment

OEM Availabilitycompatible with a wide range of servers from worldwide OEMs.
PreconfiguredNVIDIA-Certified Systems™ available through leading server partners.
Edge DeployNVIDIA A10 is supported as part of NVIDIA-Certified Systems ™, in the on prem data center, in the cloud, and at the edge.

System Compatibility

Required PCIePCIe Gen4
Motherboardcompatible with a wide range of servers from worldwide OEMs.
OS CompatvGPU software support | NVIDIA Virtual PC, NVIDIA Virtual Applications, NVIDIA RTX Virtual Workstation, NVIDIA Virtual Compute Server, NVIDIA AI Enterprise

Benchmarks & Throughput

Structured Sparsity

Supported (up to 2x vs dense)

Training Benchmarks

TF32 provides up to 5X training throughput vs previous generation

Inference Benchmarks

A10 delivers up to 2.5X more inference performance compared to T4

Multi-GPU Scalability

Scaling Characteristics

ParallelismNVIDIA vPC/vApps, NVIDIA RTX ™ vWS, NVIDIA AI Enterprise

Workload Readiness

LLM Inference

A10 delivers up to 2.5X more inference performance compared to T4

Market Authority

Community Benchmarks

["MLPerf","SPECviewperf 2020"]

Key Strengths

Limitations

Also in the Lineup

NVIDIA
A100 80GB SXM
NVIDIA
A100 80GB PCIe
NVIDIA
A10G
NVIDIA
A16
NVIDIA
A30
NVIDIA
A40
NVIDIA
B200 SXM
NVIDIA
B300 SXM

Expert Insight

The A10 represents a strategic leap in AI compute. When comparing cloud providers, consider not just the hourly rate, but also the interconnect bandwidth (InfiniBand/NVLink) and regional availability which can significantly impact total cost of ownership for large-scale training.

Glossary Terms

FP32 TFLOPS
VRAM
TDP
Cores
Information updated daily. Cloud pricing subject to vendor availability.