James Misaka

Also: > Before GCC14, implicit function declaration is treated as warning but since GCC14 this is treated as error by default, which will cause compilation to stop. For these implicit...

@YuugataShinonome Thanks a lot! I will have a mind in it when doing update for next version of toolchain ,and you can also open a commit before me, we're welcome...

@YuugataShinonome Toolchain 2024.2 is ongoing I've seen that cp2k toolchain use another modification ``` CFLAGS=${cflags} FFLAGS=${fflags} cmake -DCMAKE_FIND_ROOT_PATH="$ROOTDIR" \ -DCMAKE_INSTALL_PREFIX="${pkg_install_dir}" \ -DCMAKE_INSTALL_LIBDIR="lib" \ **-DCMAKE_VERBOSE_MAKEFILE=ON \** -DBUILD_SHARED_LIBS=NO \ -DCMAKE_BUILD_TYPE=Release .. \...

@iduygnay Can you try `exx_separate_loop=1` @PeizeLin Any comments？

@iduygnay I notice that you're using ABACUS 3.7.0, can you update to the newest ABACUS with LibRI 0.2.0 ?

@iduygnay What's your dependencies of ABACUS and QE? And What's parallel setting do you use (the parallel number of MPI and OpenMP) in HSE of ABACUS and QE? Also, What's...

@iduygnay - mixing_method and mixing_beta is recommended to use the default (broyden, and sigma is relate to your bandgap of system) - more OpenMP threads is recommend for HSE in...

@sgyang345 Can you share your INPUT file ?

We've done some tests, this OOM problem seeems to be related to `mixing_restart` in DFT+U method.

@xintianwangwendy Hello! You can try executure `OMP_NUM_THREADS=64 abacus` without mpirun if you are only using one process with 64 threads for acceleration since your PBE step seems abnormally time-consuming