HDR gaming has been a topic of long discussion. One point in the discussion would HDR have a performance impact. The short answer should be not much, or close to NIL. However, in a more extreme situation like Ultra HD and HDR, there could be an effect. Some benchmark is now available. 

Our German colleagues from Computerbase performed a new test run ay 3840x2160 (Ultra HD) with HDR enabled games. The results are quite surprising and lead to new discussion as NVIDIA is scoring substantially lower overall compared to AMD. Some claim HDR should not but could invoke another pass, a slight increase in bandwidth, or is it tone mapping related causing the substantial difference, as 10% in a heavy on the GPU resolution of Ultra HD, that is a lot of frames not showing. 

Whatever the reason it, at a recent presentation NVIDIA specifically stated that HDR gaming will not have an effect on performance, and that contradicts with these result sets. It does seem to specifically happen in Ultra HD. In some recent testing ourselves, we noticed little to no difference for both brands in 2560x1440. Who knows, a new NVIDIA GeForce Driver will fix whatever is going on. We should receive an Ultra HD HDR enabled monitor ourselves soon for review, and definitely will check out and test this behaviour ourselves. Thanks, El1te for the news-submit.

Update, thanks Nizzen for the hint. The performance loss is likely due to the way NVIDIA converts RGB to reduced chroma YCbCr 4:2:2 in HDR (we discussed that previously here) which was used for HDR testing. If they used 4:4:4 or RGB, the performance would be the same as SDR. Considering GSYNC Ultra HD monitors bog down to 4:2:2, this looks the be the most plausible explanation.

HDR Benchmarks perf impact on AMD (-2%) vs Nvidia (-10%) - 12 Games at 4K HDR