If true, the upcoming the Red Team’s upcoming performance GPU won’t be seen until Q2, missing the RyZen launch window as well as Intel Kaby Lake.
If true, the upcoming the Red Team’s upcoming performance GPU won’t be seen until Q2, missing the RyZen launch window as well as Intel Kaby Lake.
The post-CES rumour mill continues to churn this week, and perennial rumour-mongers WCCFTech are at it again with a new report regarding the eagerly anticipated AMD Vega GPU line. Citing undisclosed sources, the website stated that Vega wouldn’t be available before May 2017 and could be held until Computex at the tail end of that month.
AMD Vega’s launch timing could be critical to its success or failure in capturing market share. Aggressive pricing on 400-series Polaris cards have seen AMD regain position which it previously lost due to a run of lacklustre GPU releases, but Radeon still doesn’t have an up-to-date GPU which can go toe-to-toe with NVIDIA’s best. Currently NVIDIA has free reign at the high end of the market – ageing and power-hungry Radeon Fury designs notwithstanding – potentially making them the only choice when it comes to new system builds at the high performance end of the spectrum.
AMD Vega GPU package
The problem is acute due to two other factors: Intel’s launch of the Kaby Lake platform, and the expect launch of AMD RyZen in March. Many customers assembling brand new systems for the occasion will have to wait if they want a high-end AMD card or, more likely, opt for the NVIDIA alternative (likely a GeForce GTX 1080).
Whilst Computex is a traditional home for new product launches and indeed say the debut of Polaris last year, it may be a day late and a dollar short for AMD. As promising as Vega looks, money that would have been spent could have long since gone elsewhere.
—
AMD Vega was teased at CES last week with a prominent showcase of DOOM running at 4K at Ultra quality settings, and boasting considerable technological advancements over Polaris.
The most notable improvements are the introduction of the Next-Generation Compute Unit (a large development of the Compute Unit model used in AMD GPUs since the 7000-series) and HBM2 memory (rather than first-gen HBM as seen on Fury and R9 Nano products), however other important architectural changes should make Vega much more attractive to general enterprise customers and the ever-more significant deep learning industry.
Specific information on consumer models of Vega remain rather thin, but it has been revealed that top-end Vega 10 model will achieve 12.5 TFLOPS, an almost 50% higher performance figure than the R9 Fury ‘Fiji’ GPUs, and be available with capacious HBM2 or GDDR memory configurations; it was this GPU running DOOM in 4K on the show floor. Nothing concrete has been revealed on the lower-end Vega 11, although there are indications that it will effectively replace Polaris 10 models and not be shipped with HBM2 memory due to costs.
AMD have launched their official site for Vega, where you also have the chance of winning your own. It can be found at http://ve.ga.
