Last month, we announced that we’re working on
a new AIXPRT learning tool. Because we want tech journalists, OEM lab
engineers, and everyone who is interested in AIXPRT to be able to find the
answers they need in as little time as possible, we’re designing this tool to serve
as an information hub for common AIXPRT topics and questions.
We’re still finalizing
aspects of the tool’s content and design, so some details may change, but we
can now share a sneak peak of the main landing page. In the screenshot below,
you can see that the tool will feature four primary areas of content:
The FAQ section will provide quick answers to the questions we
receive most from testers and the tech press.
The AIXPRT basics section will describe specific topics such as the
benchmark’s toolkits, networks, workloads, and hardware and software
The testing and results section will cover the testing process,
the metrics the benchmark produces, and how to publish results.
The AI/ML primer will provide brief, easy-to-understand definitions
of key AI and ML terms and concepts for those who want to learn more about the
We’re excited about the new AIXPRT learning tool, and will share more information here in the blog as we get closer to a release date. If you have any questions about the tool, please let us know!
As we discussed last week, we’ve learned that MobileXPRT 3 is crashing after installation
on some Android 11 phones. We now know what is causing this behavior: changes
to the storage strategy and permissions requirements in Android 11. The new file
storage strategy is called scoped storage. By default, scoped storage restricts an app’s storage access
to app-specific directories and media, and prohibits general access to external
or public directories. This default setting requires an opt-in permissions
prompt that MobileXPRT 3 does not currently have. It also prevents third-party
apps such as email clients or file managers from being accessing MobileXPRT 3
To fix this, we are
planning to rebuild MobileXPRT 3 to (1) point all of the benchmark’s file
references to its private directory, and (2) allow users to zip results files
and attach them to results submission emails. While we do not expect that either
of these changes will affect performance results, we’ll perform testing to
confirm this before we publish the new build.
We don’t expect these changes to take too long, and will keep you updated here in the blog. If you have any questions about the update process or MobileXPRT 3, please let us know!
We recently received a tech support inquiry about problems with new MobileXPRT 3 installations on some Android 11 phones. The tester installed MobileXPRT 3 on a selection of phones running Android 11, and the app crashed immediately upon opening. We were able to reproduce the issue on multiple phones in our lab, and currently know that the issue may happen on the Google Pixel 3, Google Pixel 4a 5G, Google Pixel 4XL, Google Pixel 5, and the OnePlus 8T (running Android 11 with an Oxygen OS skin).
MobileXPRT 3 continues
to run without issues on Android 9 and 10 phones. When we updated an Android 10
phone with an existing MobileXPRT 3 installation to Android 11, we found that
the benchmark ran successfully. This suggests a lack of fundamental
incompatibilities between MobileXPRT 3 and current versions of Android 11. Because
some of our lab techs experienced crashes immediately after the app asked for
permissions, we think it’s possible that new permissions-setting requirements
in Android 11 are causing the problem.
working to isolate the problem and identify a course of action. We’ll share
more information here in the blog as soon as possible. If you’ve encountered
this problem in your testing, we apologize for the inconvenience, and we’re
thankful for your patience as we work towards a solution.
If you have any information you’d like to share about running MobileXPRT 3 on Android 11, please let us know!
In addition to
providing practical information about the installation package and minimum
system requirements for the data analytics workload, the paper will describe
test configuration variables, structural components, task workflows, and test
metrics. It will also include guidance on interpreting test results and submitting
them for publication.
As we’ve noted, CloudXPRT is one of the more complex tools in the XPRT family,
with no shortage of topics to explore. Possible future topics include the
impact of adjusting specific test configuration options, recommendations for
results reporting, and methods for results analysis. If there are specific
topics that you’d like us to address in future white papers, please feel free
to send us your ideas!
We hope that the
upcoming Overview of the CloudXPRT Data Analytics Workload paper
will serve as a go-to resource for CloudXPRT testers, and will answer any
questions you have about the workload. Once it goes live, we’ll provide links
in the Helpful Info box on CloudXPRT.com and the CloudXPRT section of our XPRT white papers page.
This week, we’re sharing news on two topics that we’ve discussed
here in the blog over the past several months: CloudXPRT v1.01 and a potential
AIXPRT OpenVINO update.
Last week, we announced that we were very close to releasing an
updated CloudXPRT build (v1.01) with two minor bug fixes, an improved post-test
results processing script, and an adjustment to one of our test configuration
recommendations. Our testing and prep is complete, and the new version is live
in the CloudXPRT GitHub repository and on our site!
None of the v1.01
changes affect performance or test results, so scores from the new build are
comparable to those from previous CloudXPRT builds. If you’d like to know more
about the changes, take a look at last week’s blog post.
The AIXPRT OpenVINO
In late July, we discussed our plans to update the AIXPRT OpenVINO packages
with OpenVINO 2020.3 Long-Term Support (LTS). While there are no
known problems with the existing AIXPRT OpenVINO package, the LTS version
targets environments that benefit from maximum stability and don’t require a
constant stream of new tools and feature changes, so we thought it would be
well suited for a benchmark like AIXPRT.
We initially believed that
the update process would be relatively simple, and we’d be able to release a
new AIXPRT OpenVINO package in September. However, we’ve discovered that the
process is involved enough to require substantial low-level recoding. At this
time, it’s difficult to estimate when the updated build will be ready for
release. For any testers looking forward to the update, we apologize for the
If you have any questions or comments about
these or any other XPRT-related topics, please let us know!
We want to let CloudXPRT testers know that we’re close to
releasing an updated version (build 1.01) with two minor bug fixes, an improved
post-test results processing script, and an adjustment to one of our test
configuration recommendations. None of these changes will affect performance or
test results, so scores from previous CloudXPRT builds will be comparable to
those from the new build.
The most significant changes in CloudXPRT build 1.01 are as
In previous builds, some testers encountered warnings during setup to update the version of Kubernetes Operations (kops) when testing on public-cloud platforms (the CloudXPRT 1.00 recommendation is kops version 1.16.0). We are adjusing the kops installation instructions in the setup instructions for the web microservices and data analytics workloads to prevent these warnings.
In previous builds, post-test cleanup instructions for public-cloud testing environments do not always delete all of the resources that CloudXPRT creates during setup. We are updating instructions to ensure a more thorough cleanup process. This change applies to test instructions for the web microservices and data analytics workloads.
We are reformatting the optional results graphs the web microservices postprocess program creates to make them easier to interpret.
In previous builds, the recommended time interval for the web-microservices workload is 120 seconds if the hpamode option is enabled and 60 seconds if it is disabled. Because we’ve found that the 60-second difference has no significant impact on test results, we are changing the recommendation to 60 seconds for both hpamode settings.
We hope these changes
will improve the CloudXPRT setup and testing experience. We haven’t set the
release date for the updated build yet, but when we do, we’ll announce it here
in the blog. If you have any questions about CloudXPRT, or would like to report
bugs or other issues, please feel free to contact us!