g5.8xlarge
Amazon EC2 · G5 family
The g5.8xlarge is a gpu & accelerated Amazon EC2 instance type with 32 vCPUs and 128 GiB of memory, 1 × NVIDIA A10G, running on AMD EPYC 7R32 with 25 Gigabit networking, starting at $2.448 per hour on demand.
- vCPUs
- 32
- memory
- 128 GiB
- gpu
- 1× A10G
- on-demand
- $2.448
- spot
- $1.276
- 1-yr reserved
- $1.542
dedicated
4 GiB per vCPU
24 GiB total
$1,787/mo · us-east-1
−48% vs on-demand
−37% vs on-demand
g5.8xlarge price by region
19 regions · hourly list prices in USD · On-demand · Spot · 1-yr Reserved · 3-yr Reserved
| US East (N. Virginia)us-east-1cheapestreferenceNorth America | $2.448 | $1.276 | $1.542 | $1.058 |
| US East (Ohio)us-east-2cheapestNorth America | $2.448 | $0.9043 | $1.542 | $1.058 |
| US West (Oregon)us-west-2cheapestNorth America | $2.448 | $0.6403 | $1.542 | $1.058 |
| EU (Spain)eu-south-2Europe | $2.579 | – | – | – |
| EU (Stockholm)eu-north-1Europe | $2.596 | $0.6069 | $1.636 | $1.122 |
| Canada (Central)ca-central-1North America | $2.718 | $1.012 | $1.712 | $1.174 |
| EU (Ireland)eu-west-1Europe | $2.733 | $1.097 | $1.722 | $1.181 |
| Israel (Tel Aviv)il-central-1Middle East | $2.869 | $1.082 | $1.808 | $1.24 |
| Asia Pacific (Mumbai)ap-south-1Asia | $2.94 | $1.052 | $1.852 | $1.27 |
| Middle East (UAE)me-central-1Middle East | $3.003 | $0.7936 | $1.892 | $1.297 |
| Asia Pacific (Seoul)ap-northeast-2Asia | $3.01 | $0.7515 | $1.896 | $1.30 |
| EU (Frankfurt)eu-central-1Europe | $3.061 | $0.7863 | $1.929 | $1.322 |
| EU (London)eu-west-2Europe | $3.107 | $1.026 | $1.958 | $1.342 |
| EU (Paris)eu-west-3Europe | $3.107 | – | – | – |
| Asia Pacific (Sydney)ap-southeast-2Oceania | $3.183 | $0.7554 | $2.005 | $1.375 |
| Asia Pacific (Jakarta)ap-southeast-3Asia | $3.426 | $0.5055 | $2.159 | $1.48 |
| Asia Pacific (Tokyo)ap-northeast-1Asia | $3.55 | $1.533 | $2.237 | $1.534 |
| Asia Pacific (Hong Kong)ap-east-1Asia | $3.769 | – | $2.374 | $1.628 |
| South America (Sao Paulo)sa-east-1South America | $4.161 | $0.7891 | $2.622 | $1.798 |
Specifications
compute
- vCPUs
- 32
- Memory
- 128 GiB
- Memory per vCPU
- 4 GiB
- Processor
- AMD EPYC 7R32
- CPU vendor
- AMD
- Architecture
- x86-64
- Clock speed
- 2.8 GHz
- Burstable CPU
- No
storage & network
- Local storage
- 900 GB NVME
- Storage detail
- 1 x 900 GB NVMe SSD
- Network
- 25 Gigabit
- Peak bandwidth
- 25 Gbps
accelerators
- Model
- NVIDIA A10G
- Count
- 1
- Memory (total)
- 24 GiB
- Architecture
- Ampere
- Vendor
- NVIDIA
aws details
- API name
g5.8xlarge- Family
- G5
- Category
- GPU & accelerated
- Generation
- Current
- Regions
- 19 (Asia, Europe, Middle East, North America, Oceania, South America)
- Cheapest region
- US East (N. Virginia) · $2.448/hr
- Dedicated EBS throughput
- 16000 Mbps
- Enhanced networking
- Yes
- Normalization size factor
- 2.43
- Processor features
- AMD Turbo; AVX; AVX2
- Spot interruption rate (us-east-1)
- >20%
- Spot savings vs on-demand (us-east-1)
- 58%
g5.8xlarge 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 g5.8xlarge |
|---|---|---|---|---|---|---|
| AWS | g5.8xlarge | 32 | 128 GiB | 1 × NVIDIA A10G | $2.448 | – |
| Azure | NV24ads V710 v524 vCPU (1.3× less) · AMD Radeon Pro V710 | 24 | 128 GiB | 1 × AMD Radeon Pro V710 | $2.00 | -18% |
| NG32adms V620 v1176 GiB (1.38×) · AMD Radeon Pro V620 | 32 | 176 GiB | 1 × AMD Radeon Pro V620 | $3.302 | +35% | |
| Google Cloud | g2-standard-32NVIDIA L4 | 32 | 128 GiB | 1 × NVIDIA L4 | $1.734 | -29% |
| Akamai | RTX4000 Ada x1 X-LargeNVIDIA RTX 4000 Ada | 32 | 128 GiB | 1 × NVIDIA RTX 4000 Ada | $1.53 | -37% |
G5 family sizes
| Size | vCPU | Memory | Accelerator | Local storage | On-demand | Monthly | Spot | 1-yr |
|---|---|---|---|---|---|---|---|---|
| g5.xlarge | 4 | 16 GiB | 1 × NVIDIA A10G | 250 GB NVME | $1.006 | $734 | $0.5006 | $0.6338 |
| g5.2xlarge | 8 | 32 GiB | 1 × NVIDIA A10G | 450 GB NVME | $1.212 | $885 | $0.5954 | $0.7636 |
| g5.4xlarge | 16 | 64 GiB | 1 × NVIDIA A10G | 600 GB NVME | $1.624 | $1,186 | $0.7147 | $1.023 |
| g5.8xlarge | 32 | 128 GiB | 1 × NVIDIA A10G | 900 GB NVME | $2.448 | $1,787 | $1.276 | $1.542 |
| g5.12xlarge | 48 | 192 GiB | 4 × NVIDIA A10G | 3.8 TB NVME | $5.672 | $4,141 | $2.469 | $3.573 |
| g5.16xlarge | 64 | 256 GiB | 1 × NVIDIA A10G | 1.9 TB NVME | $4.096 | $2,990 | $1.75 | $2.58 |
| g5.24xlarge | 96 | 384 GiB | 4 × NVIDIA A10G | 3.8 TB NVME | $8.144 | $5,945 | $3.955 | $5.131 |
| g5.48xlarge | 192 | 768 GiB | 8 × NVIDIA A10G | 7.6 TB NVME (2 disks) | $16.29 | $11,890 | $6.499 | $10.26 |
Similar instance types on AWS
other AWS families with a similar shape, cheapest first
g5.8xlarge FAQ
How much does g5.8xlarge cost?
g5.8xlarge costs $2.448 per hour on demand with Linux in US East (N. Virginia) (us-east-1), about $1,787 per month. Across 19 regions the Linux on-demand price ranges from $2.448 in US East (N. Virginia) to $4.161 in South America (Sao Paulo). Spot capacity costs $1.276 per hour (48% less), but can be reclaimed. A 1-yr Reserved commitment brings it to $1.542 per hour. With Windows it costs $3.92 per hour.
How many vCPUs and how much memory does g5.8xlarge have?
g5.8xlarge has 32 vCPUs and 128 GiB of memory (4 GiB per vCPU), running on AMD EPYC 7R32. It is a gpu & accelerated instance type in the G5 family.
What GPU does g5.8xlarge have?
g5.8xlarge has 1 × NVIDIA A10G with 24 GiB of accelerator memory in total.
What are the equivalents of g5.8xlarge on other clouds?
The closest matches by vCPU, memory, CPU architecture, accelerator and instance class are on Azure, NV24ads V710 v5 (24 vCPU, 128 GiB, $2.00/hour in East US); on Google Cloud, g2-standard-32 (32 vCPU, 128 GiB, $1.734/hour in Iowa); on Akamai, RTX4000 Ada x1 X-Large (32 vCPU, 128 GiB, $1.53/hour in Newark, NJ).