GPU-accelerated ECSs

GPU-accelerated ECSs

Available now: All GPU models except the recommended ones. If available ECSs are sold out, use the recommended ones.

• G series
  • GPU-accelerated Enhancement G7r
  • GPU-accelerated Enhancement G7v
  • Graphics-accelerated Enhancement G7
  • GPU-accelerated Enhancement G6
  • GPU-accelerated Enhancement G5r
  • Graphics-accelerated Enhancement G5
• P series
  • Computing-accelerated P3v
  • Computing-accelerated P2s (recommended)
  • Inference-accelerated Pi3
  • Inference-accelerated Pi2 (recommended)
  • Inference-accelerated Pi2nl

Images Supported by GPU-accelerated ECSs

GPU-accelerated Enhancement G7r

Overview

G7r ECSs use NVIDIA Quadro RTX A6000 graphics card with up to 48 GiB GDDR6 GPU memory and support DirectX, Shader Model, OpenGL, and Vulkan. Theoretically, the FP32 is 38.7 TFLOPS and the tensor is 309.7 TFLOPS (sparsity enabled). More tensor cores deliver more powerful performance to meet diverse graphics processing requirements.

Select your desired GPU-accelerated ECS type and specifications.

Specifications

G7r ECS Features

• CPU: 3rd Generation Intel® Xeon® Scalable 8378A processors (3.0 GHz of base frequency and 3.5 GHz of turbo frequency)
• Graphics acceleration APIs
  • DirectX 12.0, Direct2D, DirectX Video Acceleration (DXVA)
  • Shader Model 5.1
  • OpenGL 4.6
  • Vulkan 1.1
• CUDA, DirectCompute, and OpenCL
• A single card is equipped with 10,752 CUDA cores, 84 second-generation RT cores, and 576 third-generation Tensor cores.
• Graphics applications accelerated
• Heavy-load CPU inference
• Application flow identical to common ECSs
• Automatic scheduling of G7r ECSs to AZs where NVIDIA Quadro RTX A6000 GPUs are used
• One NVENC (encoding) engine and two NVDEC (decoding) engines (including AV1 decoding) embedded

Supported Common Software

G7r ECSs are used in graphics acceleration scenarios, such as video rendering, cloud desktop, and 3D visualization. If the software relies on GPU DirectX and OpenGL hardware acceleration, use G7r ECSs. G7r ECSs support the following commonly used graphics processing software:

• AutoCAD
• 3ds Max
• MAYA
• Agisoft PhotoScan
• ContextCapture

Notes

• After a G7r ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G7r ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• G7r ECSs created using a public image have had the GRID driver of a specific version installed by default. However, you need to purchase and configure a GRID license by yourself. Ensure that the GRID driver version meets service requirements.

  For details about how to configure a GRID license, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• If a G7r ECS is created using a private image, make sure that the GRID driver was installed during the private image creation. If the GRID driver has not been installed, install the driver for graphics acceleration after the ECS is created.

  For details, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

GPU-accelerated Enhancement G7v

Overview

G7v ECSs use NVIDIA A40 GPUs and support DirectX, Shader Model, OpenGL, and Vulkan. Each GPU provides 48 GiB of GPU memory. Theoretically, the peak FP32 is 37.4 TFLOPS and the peak TF32 tensor is 74.8 TFLOPS | 149.6 TFLOPS (sparsity enabled). They deliver two times the rendering performance and 1.4 times the graphics processing performance of RTX6000 GPUs to meet professional graphics processing requirements.

Select your desired GPU-accelerated ECS type and specifications.

Specifications

G7v ECS Features

• CPU: 3rd Generation Intel® Xeon® Scalable 8378A processors (3.0 GHz of base frequency and 3.5 GHz of turbo frequency)
• Graphics acceleration APIs
  • DirectX 12.07, Direct2D, DirectX Video Acceleration (DXVA)
  • Shader Model 5.17
  • OpenGL 4.68
  • Vulkan 1.18
• CUDA, DirectCompute, OpenACC, and OpenCL
• A single card is equipped with 10,752 CUDA cores, 84 second-generation RT cores, and 336 third-generation Tensor cores.
• Graphics applications accelerated
• Heavy-load CPU inference
• Application flow identical to common ECSs
• Automatic scheduling of G7v ECSs to AZs where NVIDIA A40 GPUs are used
• One NVENC (encoding) engine and two NVDEC (decoding) engines (including AV1 decoding) embedded

Supported Common Software

G7v ECSs are used in graphics acceleration scenarios, such as video rendering, cloud desktop, and 3D visualization. If the software relies on GPU DirectX and OpenGL hardware acceleration, use G7v ECSs. G7v ECSs support the following commonly used graphics processing software:

• AutoCAD
• 3ds Max
• MAYA
• Agisoft PhotoScan
• ContextCapture
• Adobe Premiere Pro
• Solidworks
• Unreal Engine
• Blender
• Vray

Notes

• After a G7v ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G7v ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• G7v ECSs created using a public image have had the GRID driver of a specific version installed by default. However, you need to purchase and configure a GRID license by yourself. Ensure that the GRID driver version meets service requirements.

  For details about how to configure a GRID license, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• If a G7v ECS is created using a private image, make sure that the GRID driver was installed during the private image creation. If the GRID driver has not been installed, install the driver for graphics acceleration after the ECS is created.

  For details, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

Graphics-accelerated Enhancement G7

Overview

G7 ECSs use NVIDIA A40 GPUs and support DirectX, Shader Model, OpenGL, and Vulkan. Each GPU provides 48 GiB of GPU memory. Theoretically, the peak FP32 is 37.4 TFLOPS and the peak TF32 tensor is 74.8 TFLOPS | 149.6 TFLOPS (sparsity enabled). They deliver two times the rendering performance and 1.4 times the graphics processing performance of RTX6000 GPUs to meet professional graphics processing requirements.

Select your desired GPU-accelerated ECS type and specifications.

Specifications

G7 ECS Features

• CPU: 3rd Generation Intel® Xeon® Scalable 8378A processors (3.0 GHz of base frequency and 3.5 GHz of turbo frequency)
• Graphics acceleration APIs
  • DirectX 12.07, Direct2D, DirectX Video Acceleration (DXVA)
  • Shader Model 5.17
  • OpenGL 4.68
  • Vulkan 1.18
• CUDA, DirectCompute, OpenACC, and OpenCL
• A single card is equipped with 10,752 CUDA cores, 84 second-generation RT cores, and 336 third-generation Tensor cores.
• Graphics applications accelerated
• Heavy-load CPU inference
• Application flow identical to common ECSs
• Automatic scheduling of G7 ECSs to AZs where NVIDIA A40 GPUs are used
• One NVENC (encoding) engine and two NVDEC (decoding) engines (including AV1 decoding) embedded

Supported Common Software

G7 ECSs are used in graphics acceleration scenarios, such as video rendering, cloud desktop, and 3D visualization. If the software relies on GPU DirectX and OpenGL hardware acceleration, use G7 ECSs. G7 ECSs support the following commonly used graphics processing software:

• AutoCAD
• 3ds Max
• MAYA
• Agisoft PhotoScan
• ContextCapture
• Adobe Premiere Pro
• Solidworks
• Unreal Engine
• Blender
• Vray

Notes

• After a G7 ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G7 ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• G7 ECSs created using a public image have had the GRID driver of a specific version installed by default. However, you need to purchase and configure a GRID license by yourself. Ensure that the GRID driver version meets service requirements.

  For details about how to configure a GRID license, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• If a G7 ECS is created using a private image, make sure that the GRID driver was installed during the private image creation. If the GRID driver has not been installed, install the driver for graphics acceleration after the ECS is created.

  For details, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

GPU-accelerated Enhancement G6

Overview

G6 ECSs use NVIDIA Tesla T4 GPUs to support DirectX, OpenGL, and Vulkan and provide 16 GiB of GPU memory. The theoretical Pixel rate is 101.8 Gpixel/s and Texture rate 254.4 GTexel/s, meeting professional graphics processing requirements.

Select your desired GPU-accelerated ECS type and specifications.

Specifications

G6 ECS Features

• CPU: 2nd Generation Intel® Xeon® Scalable 6266 processors (3.0 GHz of base frequency and 3.4 GHz of turbo frequency)
• Graphics acceleration APIs
  • DirectX 12, Direct2D, and DirectX Video Acceleration (DXVA)
  • OpenGL 4.5
  • Vulkan 1.0
• CUDA and OpenCL
• NVIDIA T4 GPUs
• Graphics applications accelerated
• Heavy-load CPU inference
• Automatic scheduling of G6 ECSs to AZs where NVIDIA T4 GPUs are used
• One NVENC engine and two NVDEC engines embedded

Supported Common Software

G6 ECSs are used in graphics acceleration scenarios, such as video rendering, cloud desktop, and 3D visualization. If the software relies on GPU DirectX and OpenGL hardware acceleration, use G6 ECSs. G6 ECSs support the following commonly used graphics processing software:

• AutoCAD
• 3ds Max
• MAYA
• Agisoft PhotoScan
• ContextCapture

Notes

• After a G6 ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G6 ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• G6 ECSs created using a public image have had the GRID driver of a specific version installed by default. However, you need to purchase and configure a GRID license by yourself. Ensure that the GRID driver version meets service requirements.

  For details about how to configure a GRID license, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• If a G6 ECS is created using a private image, make sure that the GRID driver was installed during the private image creation. If not, install the driver for graphics acceleration after the ECS is created.

  For details, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

GPU-accelerated Enhancement G5r

Overview

G5r ECSs are based on PCI passthrough and exclusively use GPUs for professional graphics acceleration. G5r ECSs equipped with NVIDIA Quadro RTX5000 GPUs support DirectX and OpenGL and provide a maximum GPU memory of 16 GiB for rendering, cloud gaming, and graphics workstations.

Specifications

NVIDIA Quadro RTX5000 GPUs use the latest-generation Turing architecture and work with the latest NVIDIA RTX platform.

G5r ECS Features

• CPU: 2nd Generation Intel® Xeon® Scalable 6278 processors (2.6 GHz of base frequency and 3.5 GHz of turbo frequency), or Intel® Xeon® Scalable 6151 processors (3.0 GHz of base frequency and 3.4 GHz of turbo frequency)
• Grating acceleration based on 48 RT cores
• NVIDIA RTX5000 GPUs
• Rendering graphics acceleration
• Deep learning based on 3,072 CUDA cores and 384 Tensor cores
• GPU passthrough
• A maximum GPU memory of 16 GiB

Notes

• After a G5r ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G5r ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• G5r ECSs do not support specifications modification.
• G5r ECSs are in open beta testing. Contact customer service for the test.
• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

GPU-accelerated Enhancement G5

Overview

G5 ECSs use NVIDIA GRID vGPUs and provide comprehensive, professional graphics acceleration. They use NVIDIA Tesla V100 GPUs and support DirectX, OpenGL, and Vulkan. These ECSs provide 1, 2, 4, 8, or 16 GiB of GPU memory and up to 4096 x 2160 resolution, meeting requirements from entry-level through professional graphics processing.

Select your desired GPU-accelerated ECS type and specifications.

V100-xQ indicates that V100 GPUs are virtualized to vGPUs with different specifications and models using GRID. x specifies the vGPU memory, and Q indicates that the vGPU of this type is designed to work in workstations and desktop scenarios. For more details about GRID vGPUs, see GRID VIRTUAL GPU User Guide.

Supported Common Software

Notes

• After a G5 ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a G5 ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• When the Windows OS running on a G5 ECS is started, the GRID driver is loaded by default, and vGPUs are used for video output by default. In such a case, the remote login function provided on the management console is not supported. To access such an ECS, use RDP, such as Windows MSTSC. Then, install a third-party VDI tool on the ECS for remote login, such as VNC.
• For G5 ECSs, you need to configure the GRID license after the ECS is created.
• G5 ECSs created using a public image have had the GRID driver of a specific version installed by default. However, you need to purchase and configure a GRID license by yourself. Ensure that the GRID driver version meets service requirements.

  For details about how to configure a GRID license, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• If a G5 ECS is created using a private image, make sure that the GRID driver was installed during the private image creation. If not, install the driver for graphics acceleration after the ECS is created.

  For details, see "Installing a GRID Driver on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.

• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

Computing-accelerated P3v

Overview

P3v ECSs use NVIDIA A800 GPUs and provide flexibility and ultra-high-performance computing. P3v ECSs have strengths in AI-based deep learning, scientific computing, Computational Fluid Dynamics (CFD), computing finance, seismic analysis, molecular modeling, and genomics. Theoretically, the FP32 is 19.5 TFLOPS, the TF32 tensor core is 156 TFLOPS, and the BFLOAT16 tensor core is 312 TFLOPS.

Specifications

P3v ECS Features

• CPU: 3rd Generation Intel® Xeon® Scalable 6348 processors (2.6 GHz of base frequency and 3.5 GHz of turbo frequency)
• Up to eight NVIDIA A800 GPUs on an ECS
• NVIDIA CUDA parallel computing and common deep learning frameworks, such as TensorFlow, Caffe, PyTorch, and MXNet
• 19.5 TFLOPS of single-precision computing and 9.7 TFLOPS of double-precision computing
• NVIDIA Tensor cores with 156 TFLOPS of single- and double-precision computing for deep learning
• Up to 40 Gbit/s of network bandwidth on a single ECS
• 80 GB HBM2 GPU memory per graphics card, and multiple GPU cards interconnected based on NVLink for up to 2,039 Gbit/s
• Comprehensive basic capabilities
  • User-defined network with flexible subnet division and network access policy configuration
  • Mass storage, elastic expansion, and backup and restoration
  • Elastic scaling
• Flexibility

  Similar to other types of ECSs, P3v ECSs can be provisioned in a few minutes.

• Excellent supercomputing ecosystem

  The supercomputing ecosystem allows you to build up a flexible, high-performance, cost-effective computing platform. A large number of HPC applications and deep-learning frameworks can run on P3v ECSs.

Supported Common Software

P3v ECSs are used in computing acceleration scenarios, such as deep learning training, inference, scientific computing, molecular modeling, and seismic analysis. If the software is required to support GPU CUDA, use P3v ECSs. P2vs ECSs support the following commonly used software:

• Common deep learning frameworks, such as TensorFlow, Spark, PyTorch, MXNet, and Caffe
• CUDA GPU rendering supported by RedShift for Autodesk 3ds Max and V-Ray for 3ds Max
• Agisoft PhotoScan
• MapD
• More than 2,000 GPU-accelerated applications such as Amber, NAMD, and VASP

Notes

• After a P3v ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a P3v ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• If a P3v ECS is created using a private image, make sure that the Tesla driver was installed during the private image creation. If not, install the driver for computing acceleration after the ECS is created. For details, see "Installing a Tesla Driver and CUDA Toolkit on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.
• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

Computing-accelerated P2s

Overview

P2s ECSs use NVIDIA Tesla V100 GPUs to provide flexibility, high-performance computing, and cost-effectiveness. P2s ECSs provide outstanding general computing capabilities and have strengths in AI-based deep learning, scientific computing, Computational Fluid Dynamics (CFD), computing finance, seismic analysis, molecular modeling, and genomics.

Specifications

Supported Common Software

Inference-accelerated Pi3

Overview

Pi3 ECSs use NVIDIA A30 GPUs dedicated for real-time AI inference. 24 GiB of GPU memory plus up to 933 GB/s of bandwidth allows Pi3 ECSs to be used in AI training scenarios. Its theoretical AI training throughput is three times that of NVIDA V100 graphics card and six times that of T4 graphics cards on previous-generation Pi2 ECSs. With NVIDIA A30 GPUs, Pi3 ECSs provide 330 peak INT 8 TOPS (with sparsity enabled). Theoretically, the TF32 is 10.3 TFLOPS and the TF32 tensor core is 82 TFLOPS | 165 TFLOPS (sparsity enabled).

Specifications

Pi3 ECS Features

• CPU: 3rd Generation Intel® Xeon® Scalable 6348 processors (2.6 GHz of base frequency and 3.5 GHz of turbo frequency)
• Up to two NVIDIA A30 GPUs on an ECS, and multiple GPU cards interconnected based on NVLink
• Up to 10.3 TFLOPS of single-precision computing on a single GPU
• Up to 330 TOPS of INT8 computing on a single GPU
• 24 GiB of HBM2 GPU memory with a bandwidth of 933 GiB/s on a single GPU
• One OFA, one NVJPEG, and four NVDECs embedded

Supported Common Software

Pi3 ECSs are used in GPU-based inference computing scenarios, such as image recognition, speech recognition, and natural language processing. The Pi3 ECSs can also be used for light-load training.

Pi3 ECSs support the following commonly used software:

• Deep learning frameworks, such as TensorFlow, Caffe, PyTorch, MXNet, and Spark
• More than 2,000 GPU-accelerated software applications such as AMBER, NAMD, and OPENFOAM

Notes

• After a Pi3 ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.

Resources will be released after a Pi3 ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• Pi3 ECSs support automatic recovery when the hosts accommodating such ECSs become faulty.
• When creating a Pi3 ECS, make sure that the Tesla driver was installed during the private image creation. If not, install the driver for computing acceleration after the ECS is created. For details, see "Installing a Tesla Driver and CUDA Toolkit on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.
• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

Inference-accelerated Pi2

Overview

Pi2 ECSs use NVIDIA Tesla T4 GPUs dedicated for real-time AI inference. These ECSs use the T4 INT8 calculator for up to 130 TOPS of INT8 computing. The Pi2 ECSs can also be used for light-load training.

Specifications

Pi2 ECS Features

• CPU: 2nd Generation Intel® Xeon® Scalable 6278 processors (2.6 GHz of base frequency and 3.5 GHz of turbo frequency), or Intel® Xeon® Scalable 6151 processors (3.0 GHz of base frequency and 3.4 GHz of turbo frequency)
• Up to four NVIDIA Tesla T4 GPUs on an ECS
• GPU hardware passthrough
• Up to 8.1 TFLOPS of single-precision computing on a single GPU
• Up to 130 TOPS of INT8 computing on a single GPU
• 16 GiB of GDDR6 GPU memory with a bandwidth of 320 GiB/s on a single GPU
• One NVENC engine and two NVDEC engines embedded

Supported Common Software

Pi2 ECSs are used in GPU-based inference computing scenarios, such as image recognition, speech recognition, and natural language processing. The Pi2 ECSs can also be used for light-load training.

Pi2 ECSs support the following commonly used software:

• Deep learning frameworks, such as TensorFlow, Caffe, PyTorch, and MXNet

Notes

• After a Pi2 ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a Pi2 ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• Pi2 ECSs support automatic recovery when the hosts accommodating such ECSs become faulty.
• By default, Pi2 ECSs created using a public image have the Tesla driver installed.
• If a Pi2 ECS is created using a private image, make sure that the Tesla driver was installed during the private image creation. If not, install the driver for computing acceleration after the ECS is created. For details, see "Installing a Tesla Driver and CUDA Toolkit on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.
• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.

Inference-accelerated Pi2nl

Overview

Pi2nl ECSs use NVIDIA Tesla T4 GPUs dedicated for real-time AI inference. These ECSs use the T4 INT8 calculator for up to 130 TOPS of INT8 computing. The Pi2nl ECSs can also be used for light-workload training.

Pi2nl ECS Features

• CPU: 2nd Generation Intel® Xeon® Scalable 6278 processors (2.6 GHz of base frequency and 3.5 GHz of turbo frequency), or Intel® Xeon® Scalable 6151 processors (3.0 GHz of base frequency and 3.4 GHz of turbo frequency)
• Up to four NVIDIA Tesla T4 GPUs on an ECS
• GPU hardware passthrough
• Up to 8.1 TFLOPS of single-precision computing on a single GPU
• Up to 130 TOPS of INT8 computing on a single GPU
• 16 GiB of GDDR6 GPU memory with a bandwidth of 320 GiB/s on a single GPU
• One NVENC engine and two NVDEC engines embedded

Supported Common Software

Pi2nl ECSs are used in GPU-based inference computing scenarios, such as image recognition, speech recognition, and natural language processing. The Pi2nl ECSs can also be used for light-load training.

Pi2 ECSs support the following commonly used software:

• Deep learning frameworks, such as TensorFlow, Caffe, PyTorch, and MXNet

Notes

• After a Pi2nl ECS is stopped, basic resources (including vCPUs, memory, image, and GPUs) are not billed, but its system disk is billed based on the disk capacity. If other products, such as EVS disks, EIP, and bandwidth are associated with the ECS, these products are billed separately.
  

  Resources will be released after a Pi2nl ECS is stopped. If resources are insufficient at the next start, the start may fail. If you want to use such an ECS for a long period of time, do not stop the ECS.

• Pi2nl ECSs support automatic recovery when the hosts accommodating such ECSs become faulty.
• By default, Pi2nl ECSs created using a public image have the Tesla driver installed.
• If a Pi2nl ECS is created using a private image, make sure that the Tesla driver was installed during the private image creation. If the Tesla driver has not been installed, install the driver for computing acceleration after the ECS is created. For details, see "Installing a Tesla Driver and CUDA Toolkit on a GPU-accelerated ECS" in the Elastic Cloud Server User Guide.
• GPU-accelerated ECSs differ greatly in general-purpose and heterogeneous computing power. Their specifications can only be changed to other specifications of the same instance type.
• GPU-accelerated ECSs do not support live migration.