p3.2xlarge
Amazon EC2 · P3 family
The p3.2xlarge is a gpu & accelerated Amazon EC2 instance type with 8 vCPUs and 61 GiB of memory, 1 × NVIDIA V100, running on Intel Xeon E5-2686 v4 (Broadwell) with up to 10 Gigabit networking, starting at $3.06 per hour on demand.
- vCPUs
- 8
- memory
- 61 GiB
- gpu
- 1× V100
- on-demand
- $3.06
- spot
- $1.381
- regions
- 12
dedicated
7.63 GiB per vCPU
16 GiB total
$2,234/mo · us-east-1
−55% vs on-demand
across 4 areas
p3.2xlarge price by region
12 regions · hourly list prices in USD · On-demand · Spot · 1-yr Reserved · 3-yr Reserved
| US East (N. Virginia)us-east-1cheapestreferenceNorth America | $3.06 | $1.381 |
| US East (Ohio)us-east-2cheapestNorth America | $3.06 | $1.02 |
| US West (Oregon)us-west-2cheapestNorth America | $3.06 | $0.7708 |
| EU (Ireland)eu-west-1Europe | $3.305 | $1.197 |
| Canada (Central)ca-central-1North America | $3.366 | $0.9017 |
| EU (London)eu-west-2Europe | $3.589 | $1.22 |
| AWS GovCloud (US)us-gov-west-1North America | $3.672 | – |
| EU (Frankfurt)eu-central-1Europe | $3.823 | $1.155 |
| Asia Pacific (Tokyo)ap-northeast-1Asia | $4.194 | $1.269 |
| Asia Pacific (Seoul)ap-northeast-2Asia | $4.234 | $0.7828 |
| Asia Pacific (Singapore)ap-southeast-1Asia | $4.234 | $1.208 |
| Asia Pacific (Sydney)ap-southeast-2Oceania | $4.234 | $1.589 |
Specifications
compute
- vCPUs
- 8
- Memory
- 61 GiB
- Memory per vCPU
- 7.63 GiB
- Processor
- Intel Xeon E5-2686 v4 (Broadwell)
- CPU vendor
- Intel
- Architecture
- x86-64
- Clock speed
- 2.3 GHz
- Burstable CPU
- No
storage & network
- Local storage
- Network storage only
- Storage detail
- EBS only
- Network
- Up to 10 Gigabit
- Peak bandwidth
- 10 Gbps
accelerators
- Model
- NVIDIA V100
- Count
- 1
- Memory (total)
- 16 GiB
- Architecture
- Volta
- Vendor
- NVIDIA
aws details
- API name
p3.2xlarge- Family
- P3
- Category
- GPU & accelerated
- Generation
- Current
- Regions
- 12 (Asia, Europe, North America, Oceania)
- Cheapest region
- US East (N. Virginia) · $3.06/hr
- Dedicated EBS throughput
- 1750 Mbps
- ECU
- 31
- Enhanced networking
- Yes
- Normalization size factor
- 16
- Processor features
- Intel AVX; Intel AVX2; Intel Turbo
- Spot interruption rate (us-east-1)
- >20%
- Spot savings vs on-demand (us-east-1)
- 86%
p3.2xlarge equivalents on other clouds
closest match by vCPU, memory, CPU architecture, accelerator and class · Linux on-demand in each provider's reference region
| Provider | Instance | vCPU | Memory | Processor | On-demand | vs p3.2xlarge |
|---|---|---|---|---|---|---|
| AWS | p3.2xlarge | 8 | 61 GiB | 1 × NVIDIA V100 | $3.06 | – |
| Azure | NC8as T4 v356 GiB (1.1× less) · NVIDIA T4 | 8 | 56 GiB | 1 × NVIDIA T4 | $0.752 | -75% |
| Vultr | vcg-a16-6c-60g-16vram6 vCPU (1.3× less) · NVIDIA A16 | 6 | 60 GiB | 1 × NVIDIA A16 | $0.471 | -85% |
P3 family sizes
| Size | vCPU | Memory | Accelerator | Local storage | On-demand | Monthly | Spot |
|---|---|---|---|---|---|---|---|
| p3.2xlarge | 8 | 61 GiB | 1 × NVIDIA V100 | – | $3.06 | $2,234 | $1.381 |
| p3.8xlarge | 32 | 244 GiB | 4 × NVIDIA V100 | – | $12.24 | $8,935 | $4.955 |
| p3.16xlarge | 64 | 488 GiB | 8 × NVIDIA V100 | – | $24.48 | $17,870 | $8.825 |
p3.2xlarge FAQ
How much does p3.2xlarge cost?
p3.2xlarge costs $3.06 per hour on demand with Linux in US East (N. Virginia) (us-east-1), about $2,234 per month. Across 12 regions the Linux on-demand price ranges from $3.06 in US East (N. Virginia) to $4.234 in Asia Pacific (Sydney). Spot capacity costs $1.381 per hour (55% less), but can be reclaimed. With Windows it costs $3.428 per hour.
How many vCPUs and how much memory does p3.2xlarge have?
p3.2xlarge has 8 vCPUs and 61 GiB of memory (7.63 GiB per vCPU), running on Intel Xeon E5-2686 v4 (Broadwell). It is a gpu & accelerated instance type in the P3 family.
What GPU does p3.2xlarge have?
p3.2xlarge has 1 × NVIDIA V100 with 16 GiB of accelerator memory in total.
What are the equivalents of p3.2xlarge on other clouds?
The closest matches by vCPU, memory, CPU architecture, accelerator and instance class are on Azure, NC8as T4 v3 (8 vCPU, 56 GiB, $0.752/hour in East US); on Vultr, vcg-a16-6c-60g-16vram (6 vCPU, 60 GiB, $0.471/hour in New Jersey).