2019 will see the introduction of three new GPUs, the first change in the GPU landscape in over 18 years. Nvidia lead the change with their Turing architecture introduced in late 2018. We’ve written about it extensively, but its noteworthy aspects relative to graphics are its hardware ray tracing engine, and the use of AI to do anti-aliasing. Samsung lifted … Read more
Back when I was a young engineer at Inmos (remember them?) I attended a pep talk given by the flamboyant Texan who had the job of running our new wafer fab in Duffryn, South Wales. I don’t recall his name or much of the talk, but I do remember the white suit, the ten-gallon hat, and the ostrich-skin cowboy boots. … Read more
Arm chose this year’s IEEE Hot Chips Symposium to show off their first-generation Machine Learning Processor claiming a design that delivers ‘massive’ efficiency improvements over GPU, CPU, or DSP implementations as well as the ability to scale from IoT to server applications. Being an English design, it probably serves tea in a porcelain cup, as well. Leaving aside the issue … Read more
Intel and Samsung are gearing up to introduce new GPUs to the market. GPU design has been stable for over 10 years, but change is coming. In the meantime, the GPU market is showing itself to be large, resilient, and very elastic.
The last major architectural improvement to the basic GPU design happened when the major GPU suppliers went to unified shaders with the introduction of Direct3D 10, Shader Model 4.0, in 2008. That was ten years ago. The unified shading architecture was introduced with the Nvidia GeForce 8 series, ATI Radeon HD 2000, S3 Chrome 400, Intel GMA X3000 series, Xbox … Read more
Based on HPE's Apollo 70 the Astra supercomputer will be comprised of 2,592 dual-socket nodes, containing 145,000 cores—the largest such system the company has delivered. Each node will have two 28-core Cavium ThunderX2 processors running at 2.0 GHz. The nodes will draw 1.2 MW of power. The Astra supercomputer is the first deployment of the of the Department of Energy’s … Read more
Basemark just released its Basemark GPU, a new graphics performance evaluation tool for systems with Vulkan 1.0, OpenGL 4.5, or OpenGL ES 3.1 graphics APIs. This tool, says the company, enables the industry to objectively and reliably quantify and compare graphics performance of next-generation mobile, automotive, and desktop processors. “We have poured all of our soul and expertise in making … Read more
More of a computing device than a graphics processor now? Khronos has just released version 1.1 of its high-performance GPU API, Vulkan. The heir apparent to OpenGL, and OpenGL ES, this thin and light API exposes the many features and benefits of a modern GPU like no other API and takes it into the GPU compute domain with a new … Read more
Too little too late, or the foreshadowing of history? Arm took advantage of the post-CES news lull to announce, “Project Trillium,” a suite of software tools for neural network inferencing at the edge, along with new hardware IP for deep neural network (DNN) acceleration that is scheduled to be available for licensing in late 2018. The key software element of … Read more
Finding intelligence out at the edge, or not Arm has a new group, formed quietly back in July and announced recently, that it calls its Machine Learning Group, which is a sub-group within the Client business line. Headed up by Jem Davies, most recently in charge of the media IP group, the remit and scope of this new group remained … Read more
Tiny almost invisible things that will change your life. The dominant theme at the recent Arm Techcon 2017 conference in Santa Clara was security and the messaging from execs and invited speakers alike stayed on point. This led to slim pickings for pixel junkies at the keynotes and press events, but vision and machine learning are topics that no company … Read more
