Ziilabs MandelMark Benchmark Shows ZMS-40 GPU Compute Performance is 18 Times Faster than Nvidia Tegra 3

Ziilabs released comparative GPU compute benchmark results for its ZMS-40 stem cells processor and made available its MandelMark floating point benchmark for Android 4.0 as an apk file. Since Ziilabs is using its own benchmark tool, which they released as a binary but did not open source, the results should be taken with a grain of salt. GPU Compute (aka GPGPU – General Purpose computing on GPU) is a technique to use the GPU to perform complex calculations usually performed by the CPU. MandelMark benchmark visualizes the solutions to the Mandelbrot to measure the floating point performance of mobile processors. Ziilabs compared their ZMS-40 Quad Core Cortex A9 processor with 96 stemcell cores to the following competitors: IPAD 3 – Apple A5X processor with a quad-core PowerVR SGX543MP4 GPU Galaxy Tab 7+ – Samsung Exynos 4210 processor with a quad-core ARM Mali-400 GPU Galaxy Nexus – TI OMAP 4460 processor […]

Binary Blobs Attack – ELC 2012

Zach Pfeffer, Android Platform Lead at Linaro, talks about (GPU) Binary Blobs and the problems they cause at Embedded Linux Conference 2012. Abstract: Binary Blobs Attack!!! Most SoC vendors distribute binary blobs with Linux kernel shims. These binary blobs enable graphics engines, DSPs and other cores on ARM and other architecture SoCs. These binary blobs tend to be tied to specific kernel versions which limits unreadability and hackability and complicates device manufactures, which slows down innovation and the introduction of new and unique computing devices. Here’s what I noted from this presentation: Most of the issues at Linaro come from (GPU) Binary blobs, they delay projects and cause projects to cost more as FAE and more engineers must be involved in solving problems. Blobs are here to stay mostly due to legal reasons (patents). Binary Blob are usually OS independent, e.g. the same binary is used in Windows and Linux, […]

ARM Mali-200 and Mali-400 GPU Open Source Driver Released

There has been a lot of controversy around GPU drivers and open source, as GPU drivers usually come with a blob (a binary file). If you have been lurking in Raspberry Pi forums you’ll know what I mean. But this will change thanks to Lima. No, not the capital of Peru but the open source graphics driver for ARM Mali GPUs (Mali-200 and Mali-400) also called Lima whose goal is stated as follows: The aim of this driver is to finally bring all the advantages of open source software to ARM SoC graphics drivers. Currently, the sole availability of binary drivers is increasing development and maintenance overhead, while also reducing portability, compatibility and limiting choice. Anyone who has dealt with GPU support on ARM, be it for a Linux with a GNU stack, or for an Android, knows the pain of dealing with these binaries. Lima is going to solve […]

20 Times More Graphics Performance with Imagination G6200 and G6400 PowerVR Series 6 GPU

Imagination Technologies announced the first two IP cores, namely PowerVR G6200 and G6400 GPU IP cores, part of PowerVR Series6 GPU cores. PowerVR Series6 relies on PowerVR Rogue architecture based on a scalable number of compute clusters and designed to target the requirements of new high-end graphics applications for smartphones, tablets, PC, console, automotive, DTV and more. The G6200 and G6400 have 2 and 4 compute clusters respectively. The company claims that PowerVR Series6 GPUs can deliver 20x or more of the performance of current generation GPU cores targeting comparable markets thanks to the new Rogue architecture that is around 5x more efficient than previous generations. The PowerVR Series6 GPU cores offer computing performance exceeding 100GFLOPS (gigaFLOPS) and reaching the TFLOPS (teraFLOPS) range. The PowerVR Series6 family introduces new technologies and features such as: An advanced scalable compute cluster architecture. High efficiency compression technology including lossless image and parameter compression […]

ARM Announces New Mali-T658 GPU

About one year ago, ARM announced the Mali-T604 GPU which has yet to be used in current products. Yet today, ARM announced an even more powerful GPU called Mali-T658 in Tokyo at Japan ARM Technical Symposium. Mali T658 is based on Midgard GPU architecture (as is Mali T604) which allows great GPGPU capabilities thanks to three types of pipeline (‘tri-pipe’) optimized repetitively for arithmetic, load/store and texturing. The GPU will also support standard graphics APIs such as Khronos OpenGL ES, OpenVG andMicrosoft DirectX® 11) as well as Compute APIs such as Khronos OpenCL (Full Profile), Google RenderScript compute and Microsoft DirectCompute. Performance-wise, Mali-T658 has twice as many shader cores and double the arithmetic pipelines per core which means the GPU can offer up to 10 times the performance of the Mali-400 MP GPU. On the compute side, Mali-T658 provides 4 times the processing power of Mali-T604. Mali-T658 will work with […]

Embedded Linux Optimization Techniques – ELCE 2011

Benjamin Zores, Alcatel-Lucent, describes different optimization techniques (focusing on hardware choice and software architecture) that can be used to improve the performance of embedded linux at Embedded Linux Conference Europe 2011. Abstract: This presentation provides a series of techniques that can be used for Linux embedded systems fine-grain tuning and performances optimization. Embedded systems are, by definition, always limited in terms of resources while people keep on trying to use desktop-oriented software on top of it. This talk presents a series of tips that can be used to actually measure, find and isolate bottlenecks in your system, whether it is by complete system profiling or software architecture optimization. Focus is also made on the traditional caveats that need to be avoided for your system not to be slow by design. You can also download the presentation slides. Jean-Luc Aufranc (CNXSoft)Jean-Luc started CNX Software in 2010 as a part-time endeavor, before […]