Updated on 2024-06-21 GMT+08:00

AI-accelerated ECSs

AI-accelerated ECSs, powered by Ascend processors and software stacks, are dedicated for accelerating AI applications.

AI inference-accelerated ECSs use Ascend 310 processors for AI inference acceleration.

AI-accelerated ECS Types

Table 1 AI-accelerated ECS features

Series

Compute

Disk Type

Network

Ai1s

  • vCPU to memory ratio: 1:4
  • Number of vCPUs: 2 to 32
  • 2nd Generation Intel® Xeon® Scalable Processor
  • Basic/Turbo frequency: 2.6 GHz/3.5 GHz
  • Ultra-high I/O
  • General Purpose SSD
  • High I/O
  • Extreme SSD
  • Ultra-high PPS throughput
  • An ECS with higher specifications has better network performance.
  • Maximum PPS: 2,000,000
  • Maximum intranet bandwidth: 25 Gbit/s

Ai1

  • vCPU to memory ratio: 1:4
  • Number of vCPUs: 2 to 32
  • 2nd Generation Intel® Xeon® Scalable Processor
  • Basic/Turbo frequency: 2.6 GHz/3.5 GHz

Public Images Supported by AI-accelerated ECSs

Table 2 Public images

Type

Series

Public Images

Enhanced AI inference-accelerated (type I)

Ai1s

Ubuntu Server 18.04 64bit

CentOS 7.6 64bit

AI inference-accelerated (type I)

Ai1

Ubuntu Server 16.04 64bit

CentOS 7.4 64bit

Enhanced AI Inference-accelerated Ai1s (Type I)

Overview

Ai1s ECSs use Ascend 310 processors for AI acceleration. Ascend 310 processors feature low power consumption, high computing capabilities, and significantly improved energy efficiency ratio (EER). This facilitates the wide application of AI inference. Ai1s ECSs deliver the computing acceleration capabilities of the Ascend 310 processors on the cloud platform.

Ai1s ECSs are based on Atlas 300I accelerator cards. For details, go to Ascend Community.

AI-accelerated ECSs are ideal for computer vision, smart campus, smart city, smart transportation, smart retail, Internet-based real-time communication, and video encoding and decoding scenarios.

Specifications

Table 3 Ai1s ECS specifications

Flavor

vCPUs

Memory

(GiB)

Max./Assured Bandwidth

Max. PPS

(10,000)

Ascend 310 Processors

Ascend RAM

(GiB)

Max. NIC Queues

Max. NICs

Virtualization

ai1s.large.4

2

8

4/1.3

20

1

8

2

2

KVM

ai1s.xlarge.4

4

16

6/2

35

2

16

2

3

KVM

ai1s.2xlarge.4

8

32

10/4

50

4

32

4

4

KVM

ai1s.4xlarge.4

16

64

15/8

100

8

64

8

8

KVM

ai1s.8xlarge.4

32

128

25/15

200

16

128

8

8

KVM

Features

Ai1s ECSs have the following features:

  • vCPU to memory ratio: 1:4
  • 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)
  • Ascend 310 processors, four of which in an Atlas300I accelerator card
  • 16 TeraOPS of integer-precision computing (INT8) on one processor
  • 8 GiB of GPU memory with a memory bandwidth of 50 GiB/s on one processor
  • 5-channel HD video decoder (H.264/H.265) based on built-in hardware video codec engine

Notes

  1. Ai1s ECSs support the following public images:
    • Ubuntu Server 18.04 64bit
    • CentOS 7.6 64bit
  2. Ai1s ECSs do not support modification of specifications.
  3. Ai1s ECSs support automatic recovery when the hosts accommodating such ECSs become faulty.

AI Inference-accelerated Ai1 (Type I)

Overview

Ai1 ECSs use Ascend 310 processors for AI acceleration. Ascend 310 processors feature low power consumption, high computing capabilities, and significantly improved energy efficiency ratio (EER), facilitating the wide application of AI inference. Ai1 ECSs deliver the computing acceleration capabilities of the Ascend 310 processors on the cloud platform.

Ai1 ECSs are based on Atlas 300I accelerator cards. For details, go to Ascend Community.

Ai1 ECSs are ideal for computer vision, speech recognition, and natural language processing to support smart retail, smart campus, robot cloud brain, and safe city scenarios.

Specifications

Table 4 Ai1 ECS specifications

Flavor

vCPUs

Memory

(GiB)

Max./Assured Bandwidth

Max. PPS

(10,000)

Ascend 310 Processors

Ascend RAM

(GiB)

Max. NIC Queues

Max. NICs

Virtualization

ai1.large.4

2

8

4/1.3

20

1

8

2

2

KVM

ai1.xlarge.4

4

16

6/2

35

2

16

2

3

KVM

ai1.2xlarge.4

8

32

10/4

50

4

32

4

4

KVM

ai1.4xlarge.4

16

64

15/8

100

8

64

8

8

KVM

ai1.8xlarge.4

32

128

25/15

200

16

128

8

8

KVM

Features

Ai1 ECSs have the following features:

  • 1:4 ratio of vCPUs to memory
  • 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)
  • Ascend 310 processors, four of which in an Atlas300I accelerator card
  • 8 TeraFLOPS of half-precision computing (FP16) on one processor
  • 16 TeraOPS of integer-precision computing (INT8) on one processor
  • 8 GiB of GPU memory with a memory bandwidth of 50 GiB/s on one processor
  • 16-channel HD video decoder (H.264/H.265) based on built-in hardware video codec engine

Notes

  1. Ai1 ECSs support the following OSs:
    • Ubuntu Server 16.04 64bit
    • CentOS 7.4 64bit
  2. Ai1 ECSs do not support modification of specifications.
  3. Ai1 ECSs support automatic recovery when the hosts accommodating such ECSs become faulty.

Using an AI-accelerated ECS

Perform the following steps:

  1. Create an ECS. For details, see Step 1: Configure Basic Settings.
    • In the Specifications field, select AI-accelerated specifications.
    • In the Image field, select Public image or Private image.
      • Public image: The CANN 3.1.0 development kit has been included and environment variables have been configured in public images by default. You need to verify the environment availability.
      • Private image: You need to install the driver, firmware, and development kit, and configure environment variables by yourself. For details, see the CANN Software Installation Guide of the corresponding version in Ascend Documentation.
  2. Remotely log in to the ECS.

    If your Ai1 ECS runs Linux, use an SSH password to log in to the ECS.

  3. Verify the environment availability.

    Use a sample for compilation and running. For details, see "Sample Overview" in the Model Development Learning Map of the corresponding CANN edition in Ascend Documentation.

    The sample shows how to classify images (decode, resize, and infer images) based on the Caffe ResNet-50 network.