g4dn.12xlarge-vs-g5.12xlarge
G4DN Family
When to use?G4dn instances are GPU-based instances optimized for graphics-intensive and machine learning inference workloads. They are commonly used for video streaming, game streaming, image processing, and inference workloads requiring NVIDIA T4 GPUs.
Machine Configuration (g4dn.12xlarge)
Region
us-east-1
Currency
USD ($)
| Pricing | |
|---|---|
| Monthly Cost | $2855.7600 |
| Hourly Cost | $3.9120 |
| Reserved - 1 year | $1679.1667 ($2.3002 per hour) |
| Reserved - 3 year | $1073.7778 ($1.4709 per hour) |
Compute | Value |
|---|---|
vCPUs | 48 |
Memory (Gib) | 192.00 |
Memory per vCPU (Gib) | 4 |
Clock Speed (GHz) | 2.5 |
CPU Architecture | x86_64 |
Networking | Value |
|---|---|
Network Performance | 50 Gigabit |
IPv6 | true |
Storage | Value |
|---|---|
Disk Space (GiB) | 900 |
EBS Optimized | True |
Max Bandwidth (Mbps) on EBS | 9500 |
Max Throughput (MB/s) on EBS | 1187.50 |
Max I/O Operations/second (IOPS) | 40000 |
G5 Family
When to use?G5 instances are next-generation GPU instances designed for demanding graphics and machine learning workloads. They deliver significantly higher GPU performance than G4dn and are ideal for ML training, advanced inference, high-end visualization, and graphics-heavy applications.
Machine Configuration (g5.12xlarge)
Region
us-east-1
Currency
USD ($)
| Pricing | |
|---|---|
| Monthly Cost | $4140.5600 |
| Hourly Cost | $5.6720 |
| Reserved - 1 year | $2434.6667 ($3.3352 per hour) |
| Reserved - 3 year | $1556.8611 ($2.1327 per hour) |