AMD has provided its biggest glimpse yet into its forthcoming new range of PC and server microprocessors that will be based on its next-generation Zen architecture. At the same time, it also provided a first look at its new Vega-based graphics processing unit (GPU) in action that, it hopes, will redress the balance against Nvidia in the market for GPUs.
The company is intending to stoke-up interest in its long-awaited new architecture that, it claims, will be available in the first quarter of 2017, while putting off anyone planning to buy an Intel Skylake-based PC for Christmas.
The company claims that, with the Zen architecture, it will no longer be the poor relation to Intel, performance-wise, and that it can compete with Intel's finest (and most expensive) parts.
In an hour-long event hosted by games journalist Geoff Keighley, the company showed off an eight-core, 16-thread desktop Zen-based processor running at 3.4GHz that, it claimed, can outperform the Intel i7 6900K (current UK list price: £1,125.99) running heavyweight applications like the open-source 3D imaging application Blender.
AMD also show-cased its latest architecture in combination with the also-upcoming Vega GPU by demonstrating Star Wars Battlefront - Rogue One at 4k resolution with decently high frame-rates. It was the first time that the company had shown off the new Vega GPU cores, which are intended to compete with the Nvidia's GeForce 10 series architecture that has turned heads this year.
AMD claims that the Ryzen/Vega combination can outperform the Intel Core i7 6900K paired with an Nvidia Titan X playing Battlefield 1 at 4k resolution.
"The 'Zen' core at the heart of our Ryzen processors is the result of focused execution and thousands of engineering hours designing and delivering a next-level experience for high-end PC and workstation users," said Doctor Lisa Su, presenting what is arguably the first real technical fruits of her tenure as the company's chief executive.
The key to Zen, she added, in addition to the shift to 14nm process manufacturing, was what the company called SenseMI, which has enabled an increase of more than 40 per cent in instructions per clock cycle. It, claims AMD, is comprised of five main components:
- Pure Power. More than 100 embedded sensors with accuracy to the millivolt, milliwatt, and single degree level of temperature enable optimal voltage, clock frequency, and operating mode with minimal energy consumption;
- Precision Boost. Smart logic that monitors integrated sensors and optimises clock speeds, in increments as low as 25MHz, at up to a thousand times a second;
- Extended Frequency Range. When the system senses added cooling capability, this raises the Precision Boost frequency to enhance performance;
- Neural Net Prediction. An AI-based neural network that learns to predict what future pathway an application will take based on past runs;
- Smart Prefetch. Learning algorithms that track software behaviour to anticipate the needs of an application and prepare the data in advance.
AMD has been drip-feeding information about its next-gen architecture for the past year or so, in a bid to persuade the market that it isn't being totally left behind by Intel.
In February, for example, it revealed plans for 32-core CPUs built using 14nm FinFET technology, while in August it revealed more on the high-performance desktop microprocessors it is planning to ship next year.
Acton's warnings come as Facebook is embroiled in one of the biggest data scandals in history
The unmanned tanks could eventually be kitted with AI systems
Dubbed I-MacEtch, it will help meet demand for more powerful nano-tech
GPU firm's research unit for self-driving cars is growing