This question is part of the Hill Climbing Round of the Forecasting AI Progress Tournament. You can view all other questions in this round here.
Most modern neural network implementations are based on graphics processing units [1]. This is because GPUs have a much larger number of cores than CPUs, and are thus better suited to executing highly parallelizable programs. Hence, migrating data into graphical form and then using the GPU to scan and analyze it can create a large training speedup for training ML models [2].
Videocard benchmarks is a benchmarking project by Passmark, a software company. It aggregates performance and price data on GPUs. One of the measures of performance it uses is 3DMark, a computer benchmarking tool.
What will the average top price performance (in G3D Mark /$) of the best available GPU be, on Jan 14, 2022?
This question resolves as the mean price performance (in G3D Mark / Price) of the top two currently available GPUs by price performance, on January 14th, 2022, 11:59PM GMT, according to Videocard benchmarks.
GPU prices used to calculate price performance by Passmark will be converted to Q2 2020 USD. Prices are adjusted to (average) 2019 prices using Consumer Price Index: Total All Items for the United States.
The question resolves ambiguously if Videocard benchmarks ceases to report G3D Mark, or its methodology changes substantially.