Rumors suggest that Samsung's new mobile processor will have decidedly current-gen graphical capabilities. The Exynos 5 Octa was claimed at CES to have twice the 3D gaming performance of any other mobile processor, but did not reveal its GPU configuration, leading to speculation it will have performance between that of Apple's A5X and A6X.
The world's fastest open-science supercomputer has finished its NVIDIA Kepler-based upgrade. The newly-rebranded Titan at the US Department of Energy's Oak Ridge National Laboratory in Tennessee now contains 18,688 Tesla K20 GPU accelerators that helps it achieve a peak performance of more than 20 petaflops.
NVIDIA's 2012 Kepler-based GPU lineup may have been revealed ahead of an official announcement. A list of alleged specs posted by EXPreview suggest the card range will be built from three different GPUs, including the GK110 produced from a 550mm˛ die. The flagship GTX690 is said to integrate two GK110 chips, for a total of 3.5GB of memory, with a 1.5GHz shader clock and a 4.5GHz memory clock.
Aiming to clear up confusion and alleged misinformation surrounding Android's graphics handling, Google engineer Dianne Hackborn has provided a deeper look into the mobile platform's hardware acceleration and rendering methods. "Full" hardware acceleration is said to have been added with Android 3.0 and higher, however the software has implemented some hardware accelerated drawing since before version 1.0.
ARM has introduced a new graphics core, known as the Mali-T658, that promise to bring a significant leap in performance over its predecessor. The GPU is designed for mobile applications, working in conjunction with the company's Cortex-A15 and Cortex-A7 processors. The system is also compatible with a range of APIs such as OpenGL ES, OpenCL, OpenVG and DirectX 11.
Last week rumors flew about NVIDIA delaying the launch of its next generation Kepler GPUs from late 2011 until some time in 2012. Then, Chris Malachowsky, senior vice president of research and co-founder of Nvidia, reaffirmed that the Keplers would indeed be shipping by year's end. NVIDIA has now backtracked and claims that, while it's on target for receiving the silicon to build the GPUs, actual production of commercial units won't start until 2012.
AMD on Wednesday formally released ATI Radeon HD 5870 Eyefinity 6 Edition graphics card, offering 2GB of GDDR5 memory and delivering rendering in excess of 1 billion pixels per second in current generation gaming titles--or up to 12 times more resolution than 1080p HD. The company claims that with up to 2GB of GDDR5 memory users can experience high frame rates and image quality settings in the latest DirectX 11 games in 3- and 6-display gaming setups (using one of five included adapters: 2 mini-DisplayPort-to-DisplayPort adapters, 2 passive mini-DisplayPort-to-DVI dongles, and 1 passive mini-DisplayPort-to-HDMI dongle).
Intel is delaying the launch of its much-anticipated Larrabee standalone graphics chip. While only admitting that its Larrabee discrete graphics chip project was behind schedule and that it would initially only appear as a software development platform "next year," one analyst suggests that the company may be missing its performance targets for Intel's first major-market graphics chip.
AMD on Tuesday launched its Radeon HD 4770 graphics processor as its latest performance product in the 4000 series for the desktop PC market. Featuring over 826 million transistors and the latest 40nm manufacturing process, the GPU offers performance starting at $99 (after mail-in rebate). The new card features third-generation hardware support for the latest DirectX 10.1 games (e.g., Battleforge, HAWX, S.T.A.L.K.E.R Clear Sky and Stormrise) as well as support for 128-bit GDDR5 memory (twice as fast as GDDR3 at the same clock speeds for better performance), and ATI CrossFireX, which enables a second ATI Radeon HD 4770 card to be added for a "near-linear" increase in performance.
AMD on Tuesday debuted its ATI Radeon HD 4800 X2 series of graphics chips, including its flagship ATI Radeon HD 4870 X2 dual-CPU graphics cards; the combination, it claims, delivers 2.4 teraFLOPS of processing power and its value-oriented sibling ATI Radeon HD 4850 X2. The new cards combine two ATI Radeon HD 4800 series GPUs using a faster PCIe 2.0-based interconnect and 2GB of frame buffer memory. Launched in New York City this week, the GPUs are based on the advanced 55nm design and offer DirectX 10.1 support for advanced gaming and ATI CrossFireX for maximum performance: ATI says that the efficient use of all four GPUs in a dual ATI Radeon HD 4870 X2 CrossFireX configuration will bring more than three times the performance (using 3DMark Vantage v1.01 benchmarks) of a single ATI Radeon HD 4870 card in many games.
HP on Monday addressed which of its laptops were impacted by the recent Nvidia defect, citing 24 different models that range as far back as November 2007, when HP began covering failed GPUs under warranty. The issue affects HP's Pavilion dv2000, dv6000, and dv9000 series laptops, and the Compaq Presario V3000 and V6000 series as well. HP is offering an updated BIOS for the affected machines, and is offering customers a complete set of diagnostic instructions, for determining whether a model is affected by the problem.
After a recent bout of bad press – alleged mass product defects, competitive price cuts, and legal problems – two of Nvidia's largest partners have supposedly jumped ship to pursue alternate ventures with an unnamed company. The news follows the loss of Gainward, a company now fabricating ATI-based boards. While details of the supposed deal are scarce, it is known that the two companies, XFX and EVGA, are not defecting to rival GPU manufacturer ATI.
Nvidia says that a heat problem in with one of its mobile chips will hurt its earnings for the second quarter, sending its shares tumbling down almost 25 percent to below $14 in after-hours trading. According to the Wall Street Journal, Nvidia said that the overheating problem -- which appears to be focused on older chips for notebook computers -- caused higher-than-normal failure rates in certain GPU models as well as related chip sets; the company's said that it could see a revenue shortfall of nearly 24 percent in the second quarter (ending July 27th), but also cited problems such as general market conditions, price cuts to match rival AMD, and delayed product ramps. In June, the company debuted its Nvidia GeForce 9M series, which it said was built on a new, multi-core graphics engine that is up to 40 percent faster than its earlier 8M chipsets; the GeForce 9M series chips were set to appear this summer, and while a few manufacturers had already begun shipping designs with the newer GPUs, the company used most of the quarter to push out its older chips.
NVIDIA plans on refreshing its GPU line-up next month with two new graphics card that will feature its next-generation CUDA-enabled graphics core, codenamed D10U. The company is expected to deliver both the GeForce GTX 260 (D10U-20) and GeForce GTX 280 (D10U-30) on June 18th as part of its summer refresh, according to DailyTech. The report claims that the GTX 260 will be a "significantly" scaled down version of the GTX 280, which will enable all 240 unified stream processors designed into the processor. These second-generation unified shaders perform 50 percent better than the shaders than its previous-generation offering, the company claims in its documentation; however, the new NVIDIA cards will only support GDDR3 memory and DirectX 10, while AMD focuses on faster GDDR5-based cards due early this summer.
