This topic describes the features of compute optimized instance families with GPU capabilities and lists their instance types.

vgn6i, lightweight compute optimized instance family with GPU capabilities

vgn6i is under invitational preview. To use it, submit a ticket.

Features:
  • I/O optimized
  • Supports standard SSDs and ultra disks
  • Uses NVIDIA T4 GPU computing accelerators
  • Contains virtual GPUs (vGPUs), which are the result of GPU virtualization with mediated pass-through
    • Supports the 1/8, 1/4, and 1/2 computing capacity of NVIDIA® Tesla® T4 GPUs
    • Supports 2, 4, and 8 GB of GPU memory
  • CPU-to-memory ratio of 1:5
  • Equipped with 2.5 GHz Intel ® Xeon ® Platinum 8163 (Skylake) processors
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Real-time rendering for cloud gaming
    • Real-time rendering for AR and VR applications
    • AI (deep learning and machine learning) inference for the elastic deployment of Internet services
    • Educational environment of deep learning
    • Modeling experiment environment of deep learning
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.vgn6i-m4.xlarge 4 23.0 None 1/4 × T4 4 3.0 500 Yes 2 4 10
ecs.vgn6i-m8.2xlarge 10 46.0 None 1/2 × T4 8 4.0 800 Yes 4 5 20
Note

gn6i, compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • CPU-to-memory ratio of 1:4
  • Equipped with 2.5 GHz Intel ® Xeon ® Platinum 8163 (Skylake) processors
  • Supports standard SSDs, ultra disks, and enhanced SSDs (ESSDs) that deliver millions of IOPS
  • Uses NVIDIA T4 GPU computing accelerators
    • Powered by the new NVIDIA Turing architecture
    • 16 GB memory capacity (320 GB/s bandwidth)
    • 2,560 CUDA Cores
    • Up to 320 Turing Tensor Cores
    • Mixed-precision Tensor Cores support 65 FP16 TFLOPS, 130 INT8 TOPS, and 260 INT4 TOPS
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • AI (deep learning and machine learning) inference for computer vision, speech recognition, speech synthesis, natural language processing (NLP), machine translation, and recommendation systems
    • Real-time rendering for cloud gaming
    • Real-time rendering for AR and VR applications
    • Graphics workstations or overloaded graphics computing
    • GPU-accelerated databases
    • High-performance computing
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn6i-c4g1.xlarge 4 15.0 None 1 × T4 16 4.0 500 Yes 2 2 10
ecs.gn6i-c8g1.2xlarge 8 31.0 None 1 × T4 16 5.0 800 Yes 2 2 10
ecs.gn6i-c16g1.4xlarge 16 62.0 None 1 × T4 16 60 1,000 Yes 4 3 10
ecs.gn6i-c24g1.6xlarge 24 93.0 None 1 × T4 16 7.5 1,200 Yes 6 4 10
ecs.gn6i-c24g1.12xlarge 48 186.0 None 2 × T4 32 15.0 2,400 Yes 12 6 10
ecs.gn6i-c24g1.24xlarge 96 372.0 None 4 × T4 64 30.0 4,800 Yes 24 8 10
Note

gn6e, compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • Supports ESSDs, standard SSDs, and ultra disks
  • Uses NVIDIA V100 (32 GB NVLink) GPU processors
  • CPU-to-memory ratio of 1:4
  • Equipped with 2.5 GHz Intel ® Xeon ® Platinum 8163 (Skylake) processors
  • Uses NVIDIA V100 GPU computing accelerators (SXM2-based)
    • Powered by the new NVIDIA Volta architecture
    • 32 GB of HBM2 memory (900 GB/s bandwidth)
    • 5,120 CUDA Cores
    • 640 Tensor Cores
    • Supports up to six NVLink connections for a total bandwidth of 300 GB/s with 25 GB/s each
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Deep learning applications such as training and inference applications of AI algorithms used in image classification, autonomous vehicles, and speech recognition
    • Scientific computing applications, such as fluid dynamics, finance, molecular dynamics, and environmental analysis
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn6e-c12g1.3xlarge 12 92.0 None 1 × V100 32 5.0 800 Yes 8 6 10
ecs.gn6e-c12g1.12xlarge 48 368.0 None 4 × V100 128 16.0 2,400 Yes 8 8 20
ecs.gn6e-c12g1.24xlarge 96 736.0 None 8 × V100 256 32.0 4,800 Yes 16 8 20
Note

gn6v, compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • Supports ESSDs, standard SSDs, and ultra disks
  • Uses NVIDIA V100 GPU processors
  • CPU-to-memory ratio of 1:4
  • Equipped with 2.5 GHz Intel ® Xeon ® Platinum 8163 (Skylake) processors
  • Uses NVIDIA V100 GPU computing accelerators (SXM2-based)
    • Powered by the new NVIDIA Volta architecture
    • 16 GB of HBM2 memory (900 GB/s bandwidth)
    • 5,120 CUDA Cores
    • 640 Tensor Cores
    • Supports up to six NVLink connections for a total bandwidth of 300 GB/s with 25 GB/s each
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Deep learning applications such as training and inference applications of AI algorithms used in image classification, autonomous vehicles, and speech recognition
    • Scientific computing applications, such as fluid dynamics, finance, molecular dynamics, and environmental analysis
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn6v-c8g1.2xlarge 8 32.0 None 1 × NVIDIA V100 1 × 16 2.5 800 Yes 4 4 10
ecs.gn6v-c8g1.8xlarge 32 128.0 None 4 × NVIDIA V100 4 × 16 10.0 2,000 Yes 8 8 20
ecs.gn6v-c8g1.16xlarge 64 256.0 None 8 × NVIDIA V100 8 × 16 20.0 2,500 Yes 16 8 20
ecs.gn6v-c10g1.20xlarge 82 336.0 None 8 × NVIDIA V100 8 × 16 32.0 4,500 Yes 16 8 20
Note

vgn5i, lightweight compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • Supports standard SSDs and ultra disks
  • Uses NVIDIA P4 GPU computing accelerators
  • Contains virtual GPUs (vGPUs), which are the result of GPU virtualization with mediated pass-through
    • Supports the 1/8, 1/4, 1/2, and 1/1 computing capacity of NVIDIA® Tesla® P4 GPUs
    • Supports 1, 2, 4, and 8 GB of GPU memory
  • CPU-to-memory ratio of 1:3
  • Equipped with 2.5 GHz Intel ® Xeon ® E5-2682 v4 (Broadwell) processors
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Real-time rendering for cloud gaming
    • Real-time rendering for AR and VR applications
    • AI (deep learning and machine learning) inference for the elastic deployment of Internet services
    • Educational environment of deep learning
    • Modeling experiment environment of deep learning
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.vgn5i-m1.large 2 6.0 None 1/8 × P4 1 1.0 300 Yes 2 2 6
ecs.vgn5i-m2.xlarge 4 12.0 None 1/4 × P4 2 2.0 500 Yes 2 3 10
ecs.vgn5i-m4.2xlarge 8 24.0 None 1/2 × P4 4 3.0 800 Yes 2 4 10
ecs.vgn5i-m8.4xlarge 16 48.0 None 1 × P4 8 5.0 1,000 Yes 4 5 20
Note

gn5, compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • Supports standard SSDs and ultra disks
  • Uses NVIDIA P100 GPU processors
  • Multiple CPU-to-memory ratios
  • High-performance local NVMe SSDs
  • Equipped with 2.5 GHz Intel ® Xeon ® E5-2682 v4 (Broadwell) processors
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Deep learning
    • Scientific computing applications, such as fluid dynamics, finance, genomics, and environmental analysis
    • High-performance computing, rendering, multimedia encoding and decoding, and other server-side GPU compute workloads
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn5-c4g1.xlarge 4 30.0 440 1 × NVIDIA P100 1 × 16 3.0 300 No 1 3 10
ecs.gn5-c8g1.2xlarge 8 60.0 440 1 × NVIDIA P100 1 × 16 3.0 400 No 1 4 10
ecs.gn5-c4g1.2xlarge 8 60.0 880 2 × NVIDIA P100 2 × 16 5.0 1,000 No 2 4 10
ecs.gn5-c8g1.4xlarge 16 120.0 880 2 × NVIDIA P100 2 × 16 5.0 1,000 No 4 8 20
ecs.gn5-c28g1.7xlarge 28 112.0 440 1 × NVIDIA P100 1 × 16 5.0 1,000 No 8 8 20
ecs.gn5-c8g1.8xlarge 32 240.0 1,760 4 × NVIDIA P100 4 × 16 10.0 2,000 No 8 8 20
ecs.gn5-c28g1.14xlarge 56 224.0 880 2 × NVIDIA P100 2 × 16 10.0 2,000 No 14 8 20
ecs.gn5-c8g1.14xlarge 54 480.0 3,520 8 × NVIDIA P100 8 × 16 25.0 4,000 No 14 8 20
Note

gn5i, compute optimized instance family with GPU capabilities

Features:
  • I/O optimized
  • Supports standard SSDs and ultra disks
  • Uses NVIDIA P4 GPU processors
  • CPU-to-memory ratio of 1:4
  • Equipped with 2.5 GHz Intel ® Xeon ® E5-2682 v4 (Broadwell) processors
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Deep learning inference
    • Multimedia encoding and decoding, and other server-side GPU compute workloads
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn5i-c2g1.large 2 8.0 None 1 × NVIDIA P4 1 × 8 1.0 100 Yes 2 2 6
ecs.gn5i-c4g1.xlarge 4 16.0 None 1 × NVIDIA P4 1 × 8 1.5 200 Yes 2 3 10
ecs.gn5i-c8g1.2xlarge 8 32.0 None 1 × NVIDIA P4 1 × 8 2.0 400 Yes 4 4 10
ecs.gn5i-c16g1.4xlarge 16 64.0 None 1 × NVIDIA P4 1 × 8 3.0 800 Yes 4 8 20
ecs.gn5i-c16g1.8xlarge 32 128.0 None 2 × NVIDIA P4 2 × 8 6.0 1,200 Yes 8 8 20
ecs.gn5i-c28g1.14xlarge 56 224.0 None 2 × NVIDIA P4 2 × 8 10.0 2,000 Yes 14 8 20
Note

gn4, compute optimized family with GPU capabilities

Features:
  • I/O optimized
  • Supports standard SSDs and ultra disks
  • Uses NVIDIA M40 GPU processors
  • Multiple CPU-to-memory ratios
  • Equipped with 2.5 GHz Intel ® Xeon ® E5-2682 v4 (Broadwell) processors
  • Provides strong network performance based on large computing capacity
  • Scenarios:
    • Deep learning
    • Scientific computing applications, such as fluid dynamics, finance, genomics, and environmental analysis
    • High-performance computing, rendering, multimedia encoding and decoding, and other server-side GPU compute workloads
Instance types
Instance type vCPUs Memory (GiB) Local storage (GiB) GPUs GPU memory (GB) Bandwidth (Gbit/s) Packet forwarding rate (Kpps) IPv6 support NIC queues ENIs (including one primary ENI) Private IP addresses per ENI
ecs.gn4-c4g1.xlarge 4 30.0 None 1 × NVIDIA M40 1 × 12 3.0 300 No 1 3 10
ecs.gn4-c8g1.2xlarge 8 30.0 None 1 × NVIDIA M40 1 × 12 3.0 400 No 1 4 10
ecs.gn4.8xlarge 32 48.0 None 1 × NVIDIA M40 1 × 12 6.0 800 No 3 8 20
ecs.gn4-c4g1.2xlarge 8 60.0 None 2 × NVIDIA M40 2 × 12 5.0 500 No 1 4 10
ecs.gn4-c8g1.4xlarge 16 60.0 None 2 × NVIDIA M40 2 × 12 5.0 500 No 1 8 20
ecs.gn4.14xlarge 56 96.0 None 2 × NVIDIA M40 2 × 12 10.0 1,200 No 4 8 20
Note