AMFI integrals not correct when using either Intel and/or MKL

I regularly use spin-orbit mean field integrals from Dalton. However, I have run into the issue mentioned above. I do not get the same integrals when using Intel and/or MKL compared to using the GNU compilers. For a test case, I tried running the test rsp_mnf from the list of Dalton tests. While the true result for the calculated spin-orbit coupling constant is 191.03250 cm-1, using Intel I get -106.11167 cm-1. I can confirm that the correct result is calculated when I use GCC 9.1.0 and the builtin Lapack/BLAS libraries. I cannot tell if this is due to the Intel compiler and Intel MKL or if it is only an issue with the Intel compilers.O2.out