Apparently the issue is that it is an NVIDIA 40 series GPU. NVIDIA deliberately cripples the ability of consumer GPUs to perform Float64 operations in order to encourage people to purchase the commercial versions. The time ratio from Float32:Float64 is 1:64, where on commercial devices it is 1:2. Therefore, the poor performance of the GPU is caused by this 32x slowdown. I don't think there is a way to fix this, but I am happy to be proven wrong if someone knows a way.