p4d.24xlarge
⚡ 8x GPU • 320.0 GB HBM2Part of the P4D Accelerated Computing family — Hardware accelerators (GPUs, FPGAs) for specialized compute.
Name:
pGPU Deep Learning & HPCp = GPU Deep Learning & HPCHigh-end NVIDIA GPUs (Blackwell, Hopper, Ampere, Volta) built for frontier AI/LLM training, HPC simulations, and generative AI.44th Generation4 = 4th GenerationAWS 4th generation hardware platform with updated CPU/system architecture.dDirect-Attached NVMe SSDd = Direct-Attached NVMe SSDEquipped with physical high-speed NVMe SSD instance store volumes attached locally to the host server for temporary scratch and caching.24xlarge24 Extra Large24xlarge = 24 Extra Large96 vCPUs • 1152.0 GiB RAM • 8000 GB SSD • 400 Gigabit Network • 8x NVIDIA A100 (320.0 GB VRAM)Price (Linux)
$21.9576/ hr
~$16,029.08/mo
Compute
96vCPUs
48 cores • x86_64
Memory
1152.0GiB
12.0 GiB / vCPU
Value
$0.2287/ vCPU-hr
$0.0191/GiB-hr
Network
100Gbps
Baseline: 100 Gbps
Storage & EBS
2,375MB/s
Max: 80,000 IOPS
Technical Specifications & Limits
Hardware specs, container thresholds, and system limitsCompute & Processor
ProcessorIntel Xeon Platinum 8275L
Architecturex86_64 (x86_64)
vCPUs / Threads96 vCPUs
Physical Cores48
Clock Speed3 GHz
Intel AVX / AVX2AVX & AVX2 Supported
Intel Turbo BoostYes
Bare MetalNo (Virtualized)
Performance & Benchmarks
CoreMark Score1,593,361
CoreMark / vCPU16,597.5 / core
FFmpeg Transcoding—
CoreMark per Dollar72,565 score / $
Cost per vCPU-hr$0.2287
Cost per GiB-hr$0.0191
Containers & Kubernetes
EKS Max Pods737 pods
Max ECS Tasks120 tasks
Max ENIs60
IPv4/v6 per ENI50
Total Addressable IPs3000 total IPs
CNI Prefix DelegationSupported (High Density)
Networking & Bandwidth
Network Performance400 Gigabit
Baseline Bandwidth100 Gbps
Burst / Peak Bandwidth100 Gbps
Enhanced NetworkingNo
EFA SupportNo
IPv6 SupportNative Dual-Stack Supported
Storage & EBS Optimization
Storage ConfigurationSSD (8000 GB)
Local NVMe Disk8000 GB
EBS Baseline Throughput2375 MB/s
EBS Max Throughput2375 MB/s
EBS Baseline IOPS80,000
EBS Max / Burst IOPS80,000
Dedicated EBS Throughput19000 Mbps
GPU & Hardware Acceleration
GPU ModelNVIDIA A100
Accelerator Count8x GPU(s)
Total GPU Memory320.0 GB
Memory ArchitectureHBM2
CUDA Compute Capabilityv8.0
Acceleration RuntimesCUDA, TensorRT, cuDNN, PyTorch, PyJAX
Platform & Architecture Details
Hypervisor / SystemAWS Nitro System (PCIe Offload)
Nitro EnclavesSupported (Isolated security enclaves)
Normalization Factor192
Generation StatusCurrent Generation (Recommended)
Date Introduced2020-11-02
Regional Pricing Matrix
Live on-demand rates across all global AWS cloud regionsLowest Price Region
$21.9576/ hr
US East (Ohio) (us-east-2)Highest Price Region
$35.5569/ hr
South America (Sao Paulo) (sa-east-1)Global Fleet Average
$26.5459/ hr
Across all active regionsRegional Coverage
15regions
Available worldwide| Region Name | Region Code | On-Demand Price |
|---|---|---|
Asia Pacific (Mumbai) | ap-south-1 | $26.3444 |
Asia Pacific (Seoul) | ap-northeast-2 | $30.4103 |
Asia Pacific (Singapore) | ap-southeast-1 | $26.3492 |
Asia Pacific (Sydney) | ap-southeast-2 | $28.5449 |
Asia Pacific (Tokyo) | ap-northeast-1 | $30.0978 |
AWS GovCloud (US-West) | us-gov-west-1 | $26.3511 |
Canada (Central) | ca-central-1 | $25.2517 |
EU (Frankfurt) | eu-central-1 | $27.4330 |
EU (Ireland) | eu-west-1 | $23.7157 |
EU (London) | eu-west-2 | $28.5449 |
EU (Stockholm) | eu-north-1 | $23.7157 |
South America (Sao Paulo)Highest | sa-east-1 | $35.5569 |
US East (N. Virginia) | us-east-1 | $21.9576 |
US East (Ohio)Lowest | us-east-2 | $21.9576 |
US West (Oregon)Lowest | us-west-2 | $21.9576 |
No regions matched your search filter.