Thanks, @bgoodri! I got this to work with a slightly different Makevars copied from @Gregory’s post
CXX14FLAGS = -DSTAN_THREADS -pthread
CXX14FLAGS += -O3 -march=native -mtune=native
CXX14FLAGS += -fPIC
However, the multithreaded version took 318.84 seconds while the single-threaded took 247.153 seconds. Any advice? For example, should I run fewer chains and more threads?? Or should I set cores=1 and threads=12?