1 .. This work is licensed under a Creative Commons Attribution 4.0 International
3 .. http://creativecommons.org/licenses/by/4.0
4 .. (c) OPNFV, Huawei Technologies Co.,Ltd and others.
6 *************************************
7 Yardstick Test Case Description TC069
8 *************************************
10 .. _RAMspeed: http://alasir.com/software/ramspeed/
12 +-----------------------------------------------------------------------------+
15 +--------------+--------------------------------------------------------------+
16 |test case id | OPNFV_YARDSTICK_TC069_Memory Bandwidth |
18 +--------------+--------------------------------------------------------------+
19 |metric | Megabyte per second (MBps) |
21 +--------------+--------------------------------------------------------------+
22 |test purpose | To evaluate the IaaS compute performance with regards to |
23 | | memory bandwidth. |
24 | | Measure the maximum possible cache and memory performance |
25 | | while reading and writing certain blocks of data (starting |
26 | | from 1Kb and further in power of 2) continuously through ALU |
27 | | and FPU respectively. |
28 | | Measure different aspects of memory performance via synthetic|
29 | | simulations. Each simulation consists of four performances |
30 | | (Copy, Scale, Add, Triad). |
31 | | Test results, graphs and similar shall be stored for |
32 | | comparison reasons and product evolution understanding |
33 | | between different OPNFV versions and/or configurations. |
35 +--------------+--------------------------------------------------------------+
36 |configuration | File: opnfv_yardstick_tc069.yaml |
38 | | * SLA (optional): 7000 (MBps) min_bandwidth: The minimum |
39 | | amount of memory bandwidth that is accepted. |
40 | | * type_id: 1 - runs a specified benchmark (by an ID number): |
41 | | 1 -- INTmark [writing] 4 -- FLOATmark [writing] |
42 | | 2 -- INTmark [reading] 5 -- FLOATmark [reading] |
43 | | 3 -- INTmem 6 -- FLOATmem |
44 | | * block_size: 64 Megabytes - the maximum block size per array|
45 | | * load: 32 Gigabytes - the amount of data load per pass |
46 | | * iterations: 5 - test is run 5 times iteratively. |
47 | | * interval: 1 - there is 1 second delay between each |
50 +--------------+--------------------------------------------------------------+
51 |test tool | RAMspeed |
53 | | RAMspeed is a free open source command line utility to |
54 | | measure cache and memory performance of computer systems. |
55 | | RAMspeed is not always part of a Linux distribution, hence it|
56 | | needs to be installed in the test image. |
58 +--------------+--------------------------------------------------------------+
59 |references | RAMspeed_ |
63 +--------------+--------------------------------------------------------------+
64 |applicability | Test can be configured with different: |
66 | | * benchmark operations (such as INTmark [writing], |
67 | | INTmark [reading], FLOATmark [writing], |
68 | | FLOATmark [reading], INTmem, FLOATmem); |
69 | | * block size per array; |
70 | | * load per pass; |
71 | | * number of batch run iterations; |
72 | | * iterations and intervals. |
74 | | There are default values for each above-mentioned option. |
76 +--------------+--------------------------------------------------------------+
77 |pre-test | The test case image needs to be installed into Glance |
78 |conditions | with RAmspeed included in the image. |
80 | | No POD specific requirements have been identified. |
82 +--------------+--------------------------------------------------------------+
83 |test sequence | description and expected result |
85 +--------------+--------------------------------------------------------------+
86 |step 1 | The host is installed as client. RAMspeed is invoked and logs|
87 | | are produced and stored. |
89 | | Result: logs are stored. |
91 +--------------+--------------------------------------------------------------+
92 |test verdict | Test fails if the measured memory bandwidth is below the SLA |
93 | | value or if there is a test case execution problem. |
95 +--------------+--------------------------------------------------------------+