Login

Important information

This site uses cookies to store information on your computer. By continuing to use our site, you consent to our cookies.

ARM websites use two types of cookie: (1) those that enable the site to function and perform as required; and (2) analytical cookies which anonymously track visitors only while using the site. If you are not happy with this use of these cookies please review our Privacy Policy to learn how they can be disabled. By disabling cookies some features of the site will not work.

ARM Community: Mali GPUs at MWC: Market Leading Graphics and GPU Compute - ARM Community

Jump to content

Mali GPUs at MWC: Market Leading Graphics and GPU Compute

At Mobile World Congress last week, the ARM® Mali™ team showed a selection of mobile features accelerated by GPU Compute functionality. There has been a lot of discussion round GPU Compute and last week ARM focused on showing the tangible benefits that GPU Compute can bring to real devices.

Attached Image


First part of turning GPU Compute into a reality is having real products. One of the first mobile devices able to showcase GPU Compute is the Google Nexus 10. This is the first device available that supports Google’s Renderscript GPU Compute. Renderscript is Google’s implementation of GPU Compute and is supported in Android. According to Google, Renderscript provides a platform-independent computation engine that operates at the native level. So it can be used to accelerate apps that require higher computational requirements. Google have a series of interesting blogs (here and here) that will provide you with more background. The Nexus 10 has an Exynos 5 Dual under the cover which makes use of the combination of the ARM Cortex™-A15 and the Mali-T604, the first GPU available in mobile devices with Full Profile GPU Compute functionality.

Ecosystem momentum is key. A wide range of Mali partners are now developing and seeing the advantages of GPU Compute in real applications. MulticoreWare have been demonstrating the performance benefits that GPU Compute can bring to image processing on a mobile device, so that you can do fast photo editing without having to power up your laptop. Watch the demo video below, filmed at MWC, to see GPU Compute in action. The demo is running Renderscript Compute on the Nexus 10 and shows how image processing can be significantly accelerated by handing the task over to the GPU freeing up the CPU, while also improving the energy efficiency of the device. The results typically show a performance increase from between x3 to x11, with a few showing speed ups of over x30 dependent on the filter being applied.




To help developers get hands-on with the Exynos 5 Dual, Samsung have brought out a $250 development board called the Arndale board which opens up to the wider developer community the ability to develop GPU Compute. Look out for my colleague Roberto’s blog which goes into more detail about other partners and videos of the activities they are doing.

GPU Compute can be implemented in different ways – on the ARM booth we had on show a range of OpenCL based demos. ARM Mali GPUs are the first GPUs targeting mobile devices to have conformance for Full Profile OpenCL. The OpenCL demos were running on the Arndale development board and showcased physics demos and the performance improvement possible by moving physics tasks from the CPU to the GPU, as well as the performance possible for face detection applications when the GPU is utilized.

Attached Image


The ability to benchmark and evaluate the performance enabled by GPU Compute is vital so that its impact can be accurately be evaluated. Kishonti have run the Renderscript Benchmark for the first time on the Nexus 10 and they have been impressed at how well the Nexus 10 has performed. Today, only Mali-T604 based devices are capable of running Renderscript Compute. The benchmark shows a series of filters being applied to images using various CPU cores and then showing the performance uplift that can be obtained by moving to the GPU to run the task. Watch a range of the Mali GPU demos including the Kishonti benchmark HERE.

Look out over the next few months for more Mali partners promoting what GPU Compute is enabling in real devices. We will also have a series of videos that will be available on ARMFlix next week showing these demos in action. Stay tuned!

Trina Watt, Director of Channel Marketing, Media Processing Division, ARM. I like to think of myself as a “geek in marketers clothing”. Gadgets and technology have been a passion for me as long I can remember – from dismantling my first radio when I was about 8 to now running around regularly with 3 phones, a tablet and laptop to feed my tech thirst. I started in the tech industry nearly 20 years ago in Motorola and I have never ventured far from it. I am currently focused on promoting the visually exciting Mali graphics processors. I get to work with a wide range of partners who are creating the innovative devices of the future. For a geek it doesn’t get much better than that!
All company and product names appearing in the ARM Blogs are trademarks and/or registered trademarks of ARM Limited per ARM’s official trademark list. All other product or service names mentioned herein are the trademarks of their respective owners.

0 Comments On This Entry

Please log in above to add a comment or register for an account

 
Maximise
Minimise
» 

My Blog Links

» 

Recent Entries

» 

ARM Onsite

» 

Search My Blog