Later this week AMD will hold a birthday celebration, in honor of 30 years of innovation in graphics and gaming at both AMD and its GPU predecessor ATI. The event, hosted by the company’s Chief Gaming Scientist Richard Huddy, looks back at what the company — and its predecessor in the GPU world ATI — have done with gaming, graphics and GPUs evolving from the VGA Wonder and Mach8 to the Radeon 290X and APU.
But the days of the perpetual graphics card arms race, the drive to push performance from these boards, may soon be coming to a close.
A recent report by JPR shows that in the last quarter, both AMD and Nvidia’s shipments of discrete GPUs have decreased dramatically. AMD’s discrete GPU shipments decreased by 10.7% while Nvidia saw a slump of 21% in the same market.
In a different generation, this would have been disastrous for both companies. But the AMD and Nvidia of today are different firms than they once were: both have diversified into new markets or are bundling GPUs into non-traditional environments. Both have their respective growing professional graphics lines. AMD has its profitable semi-custom silicon division, as well as the CPU-GPU hybrid desktop SoC called the APU. Nvidia has its Tegra division, which finally is making the company some money after its focus was shifted.
Is there a demand for a better graphics card?
Both AMD and Nvidia perpetually try and out-do eachother in the GPU board market, but as JPR’s most recent report shows demand is slackening. Gamers no longer require the top cards from either company, AMD knows this as the bulk of the cards released at its #gpu14 event in Hawaii were rebrands of previous offerings.
Most of the blame for this should lie at the feet of developers, who produce games for the hardware of consoles then port to PC from there. Both the Xbox One and Playstation 4 are based on yesteryear’s hardware; the latest PC GPUs from both AMD and Nvidia are orders of magnitude more powerful than the semi-custom Jaguar hardware that powers both consoles. If developers aim for this then scale up, they are missing the best features of these new GPUs.
Because of slack demand for actual high-end gaming cards, AMD and Nvidia are left producing vanity cards such as the Radeon R9 295X2, and the Nvidia Titan Z. Neither card serves a practical use, but neither company really cares as both want to compete for the honor of having the fastest card on the market.
While these tactics may serve to please investors in the short-term, AMD will have to accelerate its transition to be less reliant on add-in boards and more reliant on divisions with proven profitability. This year’s Computex trade show in Taipei demonstrated that AMD is past its crossroads, and has reconfigured itself to meet future market needs. Radeon, as the last few years of development have shown, will certainly live on though in different applications. The GPU is more important than ever in computing; Radeon GPU cores are what give AMD’s APU its competitive advantage.
There’s no longer an active and vibrant market in pushing out perpetually faster consumer graphics cards, what’s currently on the market is “good enough” for the time being. The perpetual increases in clock speed and graphics capability that we saw in the GPU world, particularly in the early-to-mid 2000s, that was driven by ever more demanding games has come to a close for the time being. If VR and 4K gaming become mainstream — both of which have a long way to go to crack the 5% market share barrier — this might create a new renaissance in add-in board GPU capability growth. But until this hypothetical becomes a reality, it’s an era of slow growth ahead.
Radeon will live on for another 30 years, but in a different focus than the product we knew for its first 30 years.