Help Center/ ECS Instance Performance Evaluation/ Instance Performance Evaluation Guide/ Compute Performance Appraisals for Mainstream Instances
Updated on 2022-02-22 GMT+08:00
View PDF
Share

Compute Performance Appraisals for Mainstream Instances

Scenarios

The cloud platform provides multiple ECS types for different compute and storage capabilities. Each ECS type provides various flavors with different vCPU and memory configurations for you to select. This section uses general computing-plus C6s, general computing-plus C6, and memory-optimized M6 ECSs as examples to describe how to use the benchmark test program from Standard Performance Evaluation Corporation, Integer (SPECInt) to evaluate the compute performance of theses ECSs running Linux.

Description

The evaluation results are obtained by running SPEC CPU®2017 on the tested ECSs. The test is performed 10 times at the performance base. Then, the average score and standard deviation are obtained based on the 10 test results. Take SPECrate®2017 Integer as an example. During the test, the number of vCPUs of the tested ECS is the same as that on the copies. Concurrently run the copies at the performance base obtained using SPECrate®2017 Integer. A higher score indicates a higher throughput per unit time.

The following table lists the test environment.

Table 1 Test environment

Parameter

Description

OS

Linux

Vendor

HUAWEI CLOUD

Compiler

GCC v7.3.0

SPEC benchmark suite

speccpu_2017 (version: 1.0.x)

SPEC CPU®2017 test program

SPECrate®2017 Integer

Image

CentOS 7.4

Evaluation Result

  • General computing-plus C6s
    Table 2 Evaluation result of general computing-plus C6s ECSs running Linux

    Flavor

    vCPUs

    Memory (GB)

    Tests

    Average Rate

    Standard Deviation

    c6s.large.2

    2

    4

    10

    4.430

    0.011

    c6s.xlarge.2

    4

    8

    10

    8.507

    0.027

    c6s.2xlarge.2

    8

    16

    10

    16.499

    0.047

    c6s.3xlarge.2

    12

    24

    10

    24.294

    0.062

    c6s.4xlarge.2

    16

    32

    10

    32.015

    0.109

    c6s.6xlarge.2

    24

    48

    10

    46.841

    0.089

    c6s.8xlarge.2

    32

    64

    10

    60.650

    0.137

    c6s.12xlarge.2

    48

    96

    10

    93.291

    0.190

    c6s.16xlarge.2

    64

    128

    10

    120.540

    0.379
  • General computing-plus C6
    Table 3 Evaluation result of general computing-plus C6 ECSs running Linux

    Flavor

    vCPUs

    Memory (GB)

    Tests

    Average Rate

    Standard Deviation

    c6.large.2

    2

    4

    10

    7.315

    0.01

    c6.xlarge.2

    4

    8

    10

    14.09

    0.02

    c6.2xlarge.2

    8

    16

    10

    26.394

    0.11

    c6.3xlarge.2

    12

    24

    10

    38.848

    0.05

    c6.4xlarge.2

    16

    32

    10

    50.417

    0.058

    c6.6xlarge.2

    24

    48

    10

    57.536

    0.164

    c6.8xlarge.2

    32

    64

    10

    66.247

    0.212

    c6.12xlarge.2

    48

    96

    10

    109.438

    0.284

    c6.16xlarge.2

    64

    128

    10

    124.416

    0.403

    c6.22xlarge.2

    88

    176

    10

    149.042

    0.211
  • Memory-optimized M6
    Table 4 Evaluation result of memory-optimized M6 ECSs running Linux

    Flavor

    vCPUs

    Memory (GB)

    Tests

    Average Rate

    Standard Deviation

    m6.large.2

    2

    4

    10

    7.315

    0.01

    m6.xlarge.2

    4

    8

    10

    14.09

    0.02

    m6.2xlarge.2

    8

    16

    10

    26.394

    0.11

    m6.3xlarge.2

    12

    24

    10

    38.848

    0.05

    m6.4xlarge.2

    16

    32

    10

    50.417

    0.058

    m6.6xlarge.2

    24

    48

    10

    57.536

    0.164

    m6.8xlarge.2

    32

    64

    10

    66.247

    0.212

    m6.16xlarge.2

    64

    128

    10

    124.416

    0.403