Category: Machine learning

News from the MobileXPRT 3 team

on September 27, 2018

A few months ago, we shared some of our thoughts during the early planning stages of MobileXPRT 3 development. Since then, we’ve started building the new benchmark with Android Studio SDK 27. We’re now at a place where we can share more details about what to expect in MobileXPRT 3. In a nutshell, one of the five workloads in the previous version, MobileXPRT 2015, is getting a major overhaul, the remaining four workloads are getting updated test content, and we’re adding one completely new workload.

One of the first challenges we tackled was to completely rebuild the Create Slideshow workload. In MobileXPRT 2015, the workload uses FFmpeg to convert photos into video. FFmpeg utilizes a C++ executable, and it needs to be compiled differently for different architectures such as x86, x64, arm32, arm64, etc. With each new Android version, the task of maintaining FFmpeg compatibility with numerous architectures and Android versions becomes more complex. MobileXPRT 2015 still works well on most Android devices, but we wanted a more future-proof solution. In MobileXPRT 3, the Create Slideshow workload will use the Android MediaCodec API instead of FFmpeg. This change enables the workload to run successfully on devices that could not complete the workload in MobileXPRT 2015.

We are updating the test content of the following workloads: Apply Photo Effects, Create Photo Collages, Encrypt Personal Content, and Detect Faces to Organize Photos. We will replace items such as photos and videos with more contemporary file resolutions and sizes where applicable.

In the mobile device market, artificial intelligence and machine learning capabilities are rapidly moving from the level of novelty to being integrated into many daily tasks, so we wanted to include an AI or ML element in MobileXPRT 3. Our new workload uses Google’s Mobile Vision API to perform optical character recognition (OCR) tasks involving scanning receipts for personal records or an expense report. The scenario is similar to the OCR receipt-scanning task in WebXPRT 3, though the two workloads are based on different text-recognition technologies.

Finally, we’re updating the MobileXPRT UI to improve the look of the benchmark and make it easier to use. We’ll share a sneak peek of the new UI here in the blog around the time of the community preview. If you have any questions about MobileXPRT 2015 or MobileXPRT 3, please let us know!

Justin

Posted in AI, Android, Benchmarking, BenchmarkXPRT development community, Collaborative benchmark development, Community Preview, Machine learning, Mobile devices, MobileXPRT, MobileXPRT 2015, MobileXPRT 3, Performance benchmarking |

AI and the next MobileXPRT

By Justin Greene

on August 16, 2018

As we mentioned a few weeks ago, we’re in the early planning stages for the next version of MobileXPRT—MobileXPRT 3. We’re always looking for ways to make XPRT benchmark workloads more relevant to everyday users, and a new version of MobileXPRT provides a great opportunity to incorporate emerging tech such as AI into our apps. AI is everywhere and is beginning to play a huge role in our everyday lives through smarter-than-ever phones, virtual assistants, and smart homes. The challenge for us is to identify representative mobile AI workloads that have the necessary characteristics to work well in a benchmark setting. For MobileXPRT, we’re researching AI workloads that have the following characteristics:

They work offline, not in the cloud.
They don’t require additional training prior to use.
They support common use cases such as image processing, optical character recognition (OCR), etc.

We’re researching the possibility of using Google’s Mobile Vision library, but there may be other options or concerns that we’re not aware of. If you have tips for places we should look, or ideas for workloads or APIs we haven’t mentioned, please let us know. We’ll keep the community informed as we narrow down our options.

Justin

Posted in AI, Android, Benchmark metrics, Benchmarking, BenchmarkXPRT, BenchmarkXPRT development community, Collaborative benchmark development, Future of performance evaluation, Google, Machine learning, Mobile devices, MobileXPRT, Performance benchmarking | Tagged AI, benchmark, Google, image processing, mobile, MobileXPRT, optical character recognition |

MWCS18 and AIXPRT: a new video

By Justin Greene

on July 12, 2018

A few weeks ago, Bill shared his first impressions from this year’s Mobile World Congress Shanghai (MWCS). “5G +” was the major theme, and there was a heavy emphasis on 5G + AI. This week, we published a video about Bill’s MWCS experience and the role that the XPRTs can play in evaluating emerging technologies such as 5G, AI, and VR. Check it out!

You can read more about AIXPRT development here. We’re still accepting responses to the AIXPRT Request for Comments, so if you would like to share your ideas on developing an AI/machine learning benchmark, please feel free to contact us.

Justin

Posted in 5G, AI, AIXPRT, AR, Augmented reality, Future of performance evaluation, Machine learning, Mobile devices, Mobile World Congress, Mobile World Congress, Performance benchmarking, Virtual reality | Tagged 5G, AI, AIXPRT, MWCS18, Shanghai, VR |

WebXPRT passes another milestone!

By Justin Greene

on June 14, 2018

We’re excited to see that users have successfully completed over 250,000 WebXPRT runs! From the original WebXPRT 2013 to the most recent version, WebXPRT 3, this tool has been popular with manufacturers, developers, consumers, and media outlets around the world because it’s easy to run, it runs quickly and on a wide variety of platforms, and it evaluates device performance using real-world tasks.

If you’ve run WebXPRT in any of the more than 458 cities and 64 countries from which we’ve received complete test data—including newcomers Lithuania, Luxembourg, Sweden, and Uruguay—we’re grateful for your help in reaching this milestone. Here’s to another quarter-million runs!

If you haven’t yet transitioned your browser testing to WebXPRT 3, now is a great time to give it a try! WebXPRT 3 includes updated photo workloads with new images and a deep learning task used for image classification. It also uses an optical character recognition task in the Encrypt Notes and OCR scan workload and combines part of the DNA Sequence Analysis scenario with a writing sample/spell check scenario to simulate online homework in the new Online Homework workload. Users carry out tasks like these on their browsers daily, making these workloads very effective for assessing how well a device will perform in the real world.

Happy testing to everyone, and if you have any questions about WebXPRT 3 or the XPRTs in general, feel free to ask!

Justin

Posted in Benchmark metrics, Benchmarking, Browser-based benchmarks, Cross-platform benchmarks, Machine learning, Performance benchmarking, WebXPRT, WebXPRT 2013, WebXPRT 2015, WebXPRT 3 | Tagged benchmarks, browser, deep learning, WebXPRT |

AIXPRT: We want your feedback!

By Justin Greene

on April 12, 2018

Today, we’re publishing the AIXPRT Request for Comments (RFC) document. The RFC explains the need for a new artificial intelligence (AI)/machine learning benchmark, shows how the BenchmarkXPRT Development Community plans to address that need, and provides preliminary design specifications for the benchmark.

We’re seeking feedback and suggestions from anyone interested in shaping the future of machine learning benchmarking, including those not currently part of the Development Community. Usually, only members of the BenchmarkXPRT Development Community have access to our RFCs and the opportunity to provide feedback. However, because we’re seeking input from non-members who have expertise in this field, we will be posting this RFC in the New events & happenings section of the main BenchmarkXPRT.com page and making it available at AIXPRT.com.

We welcome input on all aspects of the benchmark, including scope, workloads, metrics and scores, UI design, and reporting requirements. We will accept feedback through May 13, 2018, after which BenchmarkXPRT Development Community administrators will collect and evaluate the feedback and publish the final design specification.

Please share the RFC with anyone interested in machine learning benchmarking and please send us your feedback before May 13.

Justin

Posted in AI, Benchmark metrics, Benchmarking, BenchmarkXPRT, BenchmarkXPRT development community, Collaborative benchmark development, Future of performance evaluation, Machine learning, Performance benchmarking, What makes a good benchmark? |

Machine learning in 2018

By Bill Catchings

on December 21, 2017

We are almost to the end of 2017 and, as you have probably guessed, we will not have a more detailed proposal of our machine learning benchmark ready by the end of the year.

The key aspects of the benchmark proposal we wrote about a few months ago haven’t changed, but we are running behind schedule. We are still hoping to have the proposal ready in Q1 2018 and the tool based on that proposal later in the year. We will keep you posted.

In the meantime, we hope you enjoy as much as we did the recent CGP Grey tech video explanation of machine learning. There are actually two videos—the first one gives a general overview and then the second one does a better job of looking at the current state of machine learning. It talks mainly about the training aspects of machine learning rather than the inference aspects that we are looking into with AIXPRT/MLXPRT.

From all of us in the BenchmarkXPRT Development Community, we hope you and yours have a wonderful holiday and a great start to 2018!

Bill

Posted in BenchmarkXPRT development community, Collaborative benchmark development, Future of performance evaluation, Machine learning, Performance benchmarking | Tagged AIXPRT, MLXPRT |

Category: Machine learning

News from the MobileXPRT 3 team

AI and the next MobileXPRT

MWCS18 and AIXPRT: a new video

WebXPRT passes another milestone!

AIXPRT: We want your feedback!

Machine learning in 2018

Check out the other XPRTs: