Khronos Group member Movidius continues to scale its R&D team, appointed new members to its Technical Advisory Board, collaborated with new customers and partners, such as Google for its Project Tango device, and launched the next generation of its vision processor for mobile and connected devices. Movidius is one of many Khronos Group members working on the OpenVX API.
NVIDIA has released driver 350.12, which is both a Game Ready release and their first R349 WHQL release. NVIDIA has also posted a bit more information (PDF) on their OpenCL 1.2 driver. The OpenCL 1.2 drivers are based on the published Khronos Specification and are expected to pass the Khronos Conformance Testing. OpenCL 1.2 functionality will be available on Kepler and Maxwell GPUs.
The CAD Society has awarded the Lifetime Achievement Award to Jon Peddie of JPR and GraphicSpeak based on his research and publications around 3D graphics technology. The award is given “in recognition of his outstanding lifelong pioneering, consulting, and analysis service dedicated to advancing the underappreciated pixel.” Jon Peddie has been a long time friend and partner of The Khronos Group, and we'd like to congratulate him on this great achievement.
Xiaomi introduced its flagship tablet Mi Pad last week. Xiaomi Mi Pad houses the new generation Nvidia Tegra K1 processor. Inside, it packs ARM Cortex A15 quad-core chipset which can clock 2.2GHz CPU cycles and is backed by GeForce Kepler (192 cores) graphics engine. It is 1.5 times more energy efficient than other mobile GPUs and supports OpenGL 4.4, OpenGL ES 3.1 and tessellation, says the company. Mi Pad runs on the tweaked Android OS (4.4.4 KitKat) with MIUI v6.0. Unlike other UIs we see in Samsung, HTC, Sony and LG devices, Xiaomi's MIUI has a feature-rich interface and is refreshing. It is more like a hybrid between Android and iOS, yet simple and visually appealing.
Microsoft achieves compelling Performance-per-Watt in cloud data center acceleration with OpenCL powered FPGAs
Altera Corporation announced Microsoft is using Altera Arria 10 FPGAs to achieve compelling performance-per-Watt in data center acceleration based on CNN (convolutional neural network) algorithms. This performance is achieved using the open software development language known as OpenCL, or VHDL to code the Arria 10 FPGA and its IEEE754 hard floating point DSP (digital signal processing) blocks.
OpenCL support at Intel is now going mainstream with a full integration into Intel’s portfolio of software development suites. The Intel SDK for OpenCL Applications features are now integrated into various development tools under the new name of OpenCL Code Builder. The different solutions are tailored to the target development environment: for mobile and PC development, the OpenCL Code Builder is now integrated into Intel Integrated Native Developer Experience (Intel INDE); For media server usages, the code builder is now integrated into Intel Media Server Studio. For other usages, a standalone Linux* version of the OpenCL Code Builder is also available. Two major new features are now supported: Commercial OpenCL Linux driver for Intel Processor Graphics is available with Intel Media Server Studio; OpenCL 2.0 support for 5th Generation Intel Core Processors with fine-grained SVM on Intel HD and Iris Graphics is available in Intel INDE.
AMD's first high-performance system-on-a-chip (SoC) and next-generation Mobile APU codenamed "Carrizo" showcased at CES 2015. Expected in market by mid-year, "Carrizo" will arrive with support for next-generation APIs such as DirectX 12 and OpenCL 2.0.
Rightware launched Basemark ES 3.0, a new benchmarking software that enables professional and objective performance comparisons of mobile devices that feature the new OpenGL ES 3.0 graphics API. This API can be found in the latest popular Android and iOS smartphones and tablets and it enables unprecedented graphical fidelity in graphically rich apps such as games and navigation.
There were a few significant news announcements made at SC14 this week around OpenCL. Here are a few of the more significant ones:
- Xilinx Announces SDAccel Dev Environment for C, C++ & OpenCL
- CL2QCD – Lattice QCD based on OpenCL
- AMD got first on the Green500. 2 TFLOPS DGEMM. FirePro S9150, ofcourse powered by OpenCL
- OpenCL at Seoul National University: SnuCL cluster manager.
- SNUCL - An OpenCL Framework for Heterogeneous Clusters
AMD FirePro S9150 GPUs are delivering more than 3 petaFLOPS of AMD FirePro S9150 server GPU performance for the GSI Helmholtzzentrum für Schwerionenforschung GmbH in support of physics research related to heavy ion studies. Each of the 160 ASUS ESC4000 G2S nodes of the new L-CSC cluster include four AMD FirePro S9150 server GPUs, for a total GPU peak of 3.25 petaFLOPS single precision and 1.62 petaFLOPS double precision performance. The cluster is enabling Lattice Quantum Chromo Dynamics (Lattice QCD) computational research using one of the fastest OpenCL implementations for research applications in the world.