Whenever someone plays games on their gaming PC through Parsec, we collect diagnostic information to help identify issues with the connection. On an aggregate level, it can be quite interesting to see what the most common hardware platforms being used in the Parsec Community. For instance, we know that Nvidia GPUs are the most common in our community. Nvidia’s dominance is to be expected — according to Jon Peddie Research, Nvidia owned 72.8% of the consumer GPU market at the end of 2017. Among the Parsec Community, 78% of hosting computers use an Nvidia GPU. Interestingly, many of our users are hosting on Intel HD Graphics (even though we don’t officially support this). They’re connecting to a PC that doesn’t have a discrete GPU and are relying on the GPU that is embedded directly on the the Intel CPU.
Nvidia most likely over-indexes in the Parsec Community because a significant percentage of people using Parsec play on cloud gaming PCs. Cloud gaming PCs overwhelmingly skew to Nvidia due to the public cloud companies focusing on machine learning. To accommodate the needs of machine learning engineers, the public cloud companies have to use Nvidia GPUs due to their exclusive support of the CUDA platform. CUDA is a platform and API that allows machine learning engineers get the most from the Nvidia GPUs. The open standard, OpenCL, hasn’t caught on yet.
The Most Popular GPU Is The Nvidia 1060
The most popular GPU in the Parsec Community is the GeForce 1060. About 75% of the Parsec players are using the 6GB version versus 25% who are using the 3GB version. Following the 1060, the next most popular are the higher-end versions of the GeForce 900 (970) and 10 series (1070). For AMD, their most popular GPU in our community is from the previous generation — the R9 200. The high-end GPU from AMD, the RX 480, follows that.