NVIDIA GeForce RTX 2080 Ti To GTX 980 Ti TensorFlow Benchmarks With ResNet-50, AlexNet, GoogLeNet, Inception, VGG-16
The TensorFlow performance with the GeForce RTX 2080 Ti is very strong compared to the GeForce GTX 1080 Ti especially with FP16 precision. Here is a look at the overall GPU temperature over the course of all the TensorFlow benchmarks executed:
The RTX 2080 Ti had an average core temperature of 60 degrees with a peak of 73 degrees, measurably lower than both the GTX 980 Ti and GTX 1080 Ti Founder's Edition on the same system.
For the overall AC system power consumption across all of the TensorFlow benchmarks, the RTX 2080 Ti with the Core i7 8086K system had an average AC power draw of 224 Watts and a peak of 343 Watts... The average power consumption is right on par with the GTX 1080 Ti at 220 Watts while the RTX Turing card had a peak power draw 20 Watts higher. The close average power draw is likely due to the big efficiency improvement at FP16.
Wrapping things up are the performance-per-dollar metrics: