EViews.com

Hello, maybe stupid question, but I want to be clear in this:

I made VAR model in EViews, where there are some statistically insignificant variables. I ran forecast and I wanted to compute Root Mean Square Error. I tought that RMSE is computed according to model with only statistically significant variables but EViews gives me results that seem to be computed from model with all variables (regardless there are also insignificant variables). Such RMSE can´t be true then.
Can you help me? Thank you.

That's the way how RMSE (or any other evaluation metric) is computed. There is nothing inherently wrong here. Whether or not to keep insignificant variables in the model is up to you (based on the research question, hypotheses and other statistical criteria of course).