These numbers aren't very comparable. Counting war deaths is a straightforward enterprise compared to trying to estimate smoking deaths which is based on limited statistical samples and questionable statistical models.
Evidence seems to show that smoking is bad for your health - hardly a new idea. However, the smoking death statistics are mostly an estimate as smoking isn't listed as a cause of death. Originally, the CDC didn't release their data set or model as to how they calculated the famous or infamous 400,000 number. Hopefully that has changed.
A big problem for these estimations is that smoking strongly correlates with socio-economic class in the US which also correlates with a bunch of seemingly poor health choices. Separating out the true cause isn't something that you can necessarily do with statistical analysis.