Templates do help with performance. Look up the performance characteristics of qsort (C) vs std::sort (templatized C++ sort).
I tested both extensively lately and the performance was exactly the same (under Intel Compiler and Visual Studio) so probably those compilers found a way to optimize it. They both are much slower than hand coded version anyway so it doesn't really matter (link to implementation which beats standard qsort/std::sort performance by 2x/3x times (at least on my data): http://www.ucw.cz/libucw/doc/sort.html).
Yes, Go is slower but it's not slower because of fixing those errors, it's slower because it introduces different features suitable for its niche (GC, reflection etc) Rust takes similar approach to avoid C++ design mistakes and doesn't sacrifice performance in doing so.
Those mistakes (mainly the whole "object model" thing) is what he is saying in the article. Pointing out performance of Go isn't really an argument against his point.