[thestephen]

Further down in the article:

In our original demo, we ran a V100 GPU against 8 CPU nodes. The new T4 GPUs cut our cost in half, which meant we reduced the Apache Spark cluster to 4 CPU nodes in order to maintain price parity. Even with the reduced GPU memory, the whole workload ran significantly faster.

[einpoklum]

Actually, what you should normalize is operating costs, so probably power consumption is the better figure to normalize by.

... on the other hand, that's more difficult to measure, and TDP figures are not what really happens.