This week Nvidia Corp. unveiled two new GPUs, the GTX 1050 and GTX 1050 Ti. Unlike the TSMC 16 nanometer FinFET manufacturing process utilized with the other Pascal GPUs released since May, the 1050 series is based on a 14nm part from an unspecified provider.
Both the 2GB GTX 1050 ($109) and 4GB GTX 1050 Ti ($139) are inexpensive GPUs with modest 75W power requirements targeting entry-level gaming consumers who wish to upgrade graphics on an existing PC with lower-performance GPUs integrated into the motherboard. At 75W, these GPUs do not require separate power cords to the computer’s power supply, or larger power supplies. As such they are easy upgrades that only need to be plugged into an open motherboard slot. At 1.8 TFLOPS (trillion floating-point operations per second) for the GTX 1050, and 2.1 TFLOPS for the GTX 1050 Ti, both cards out-perform previous Nvidia GPUs such as the GTX 750 and GTX 950 in the same price categories while adding modern support for DirectX 12. The closest competitors are the 2GB RX 460 ($109) and 4GB RX 460 ($119) GPUs from Advanced Micro Devices, Inc. – both rated at 2.2 TFLOPS.
Impact: Budget-priced yet powerful graphic cards are an important segment within gaming hardware. Consumers often don’t jump to multi-thousand dollar, high-performance desktops in a single bound. They usually take small steps – upgrading an existing PC to play competitive titles such League of Legends or Overwatch that do not require top-level systems. Unfortunately, it is difficult to gauge how these new cards will scale in performance since Nvidia did not release reference models. The first versions anyone will see are coming direct from manufacturing vendors on October 25th, and these GP 107 GPUs are not from the same TSMC fabricator as the previous GP 106 Pascal GPUs. Most bets so far are the 14nm GP 107 is fabricated by Samsung. Solid test results will tell how these new GPUs will perform, but is hard to say at this juncture whether they will out-perform AMD’s RX 460 line. What is firm is that Nvidia now has products from top-tier high-performance down to budget-level – matching AMD’s thunder at the low-end. AMD’s strategy this year was to target the mid-range to low-end while leaving the high-end to Nvidia. Now Nvidia has GPUs in every price category while AMD will not have a high-end product until next year. The startling performance spikes that came with jumping to 14nm and 16nm manufacturing is what made AMD’s strategy possible since this year’s mid-range GPU handily beats last-year’s high-end GPU. But with Nvidia now in a wide range of price/performance brackets, the pressure will be on both firms to see which can scale their hardware more quickly in the months ahead to higher performance thresholds.
