Interrupt in `Charge_Mixing::mix_rho`

[PeizeLin]

Describe the bug

128 He atoms PBE scf nspin=1 Interrupt in Charge_Mixing::mix_rho at the first step.

1 MPI * 56 OpenMP Remaining memory is 194 GB

ABACUS version: 2024.11.13-dcff74dbeb

He_PBE_128_k1.zip

Expected behavior

No response

To Reproduce

No response

Environment

• CentOS Linux release 7.9.2009 (Core)
• icpc version 2021.5.0 (gcc version 10.2.0 compatibility)
• mpiicpc for the Intel(R) MPI Library 2021.5 for Linux*

Additional Context

No response

Task list for Issue attackers (only for developers)

• [ ] Verify the issue is not a duplicate.
• [ ] Describe the bug.
• [ ] Steps to reproduce.
• [ ] Expected behavior.
• [ ] Error message.
• [ ] Environment details.
• [ ] Additional context.
• [ ] Assign a priority level (low, medium, high, urgent).
• [ ] Assign the issue to a team member.
• [ ] Label the issue with relevant tags.
• [ ] Identify possible related issues.
• [ ] Create a unit test or automated test to reproduce the bug (if applicable).
• [ ] Fix the bug.
• [ ] Test the fix.
• [ ] Update documentation (if necessary).
• [ ] Close the issue and inform the reporter (if applicable).

[mohanchen]

Could you check old versions of ABACUS, such as 3.4, 3.5, 3.6, 3.7?

[WHUweiqingzhou]

@PeizeLin PR #5508 fixed a serious memory leak in FFT, could you try again?

[WHUweiqingzhou]

BTW, I tried to reproduce your issue, but failed. Could you debug your calculation with GDB?

                              ABACUS v3.8.3

               Atomic-orbital Based Ab-initio Computation at UStc                    

                     Website: http://abacus.ustc.edu.cn/                             
               Documentation: https://abacus.deepmodeling.com/                       
                  Repository: https://github.com/abacusmodeling/abacus-develop       
                              https://github.com/deepmodeling/abacus-develop         
                      Commit: 264915205 (Wed Nov 20 08:53:18 2024 +0800)

 Wed Nov 20 13:40:33 2024
 MAKE THE DIR         : OUT.ABACUS/
 RUNNING WITH DEVICE  : CPU / Intel(R) Xeon(R) Platinum 8269CY CPU @ 2.50GHz
 UNIFORM GRID DIM        : 375 * 375 * 2160
 UNIFORM GRID DIM(BIG)   : 75 * 75 * 540
 DONE(3.84129    SEC) : SETUP UNITCELL
 DONE(5.239      SEC) : INIT K-POINTS
 ---------------------------------------------------------
 Self-consistent calculations for electrons
 ---------------------------------------------------------
 SPIN    KPOINTS         PROCESSORS  THREADS     NBASE       
 1       1               32          32          640         
 ---------------------------------------------------------
 Use Systematically Improvable Atomic bases
 ---------------------------------------------------------
 ELEMENT ORBITALS        NBASE       NATOM       XC          
 He      2s1p-6au        5           128         
 ---------------------------------------------------------
 Initial plane wave basis and FFT box
 ---------------------------------------------------------
 DONE(7.58819    SEC) : INIT PLANEWAVE
 -------------------------------------------
 SELF-CONSISTENT : 
 -------------------------------------------
 START CHARGE      : atomic
 DONE(136.831    SEC) : INIT SCF
 ITER       ETOT/eV          EDIFF/eV         DRHO     TIME/s
 GE1     -9.61443788e+03   0.00000000e+00   4.3042e-02  19.98
 GE2     -9.61463134e+03  -1.93460842e-01   6.3655e-03  20.05
 GE3     -9.61464143e+03  -1.00895202e-02   1.9757e-03  20.15
 GE4     -9.61464588e+03  -4.45173080e-03   1.0249e-03  20.21
 GE5     -9.61464973e+03  -3.85078496e-03   6.7370e-04  20.28
 GE6     -9.61465098e+03  -1.25104934e-03   4.6435e-04  20.36
 GE7     -9.61465159e+03  -6.07939182e-04   3.4548e-04  20.44
 GE8     -9.61465184e+03  -2.47790231e-04   2.3762e-04  20.49
 GE9     -9.61465198e+03  -1.42980558e-04   1.4803e-04  20.46
 GE10    -9.61465203e+03  -4.75686398e-05   8.3014e-05  20.43
 GE11    -9.61465207e+03  -3.84768551e-05   5.5287e-05  20.42
 GE12    -9.61465207e+03  -5.74759178e-06   2.7248e-05  20.42
 GE13    -9.61465207e+03  -7.90944256e-07   1.5309e-05  20.41

[A-006]

Due to my mistake, the memory in FFT was not properly released. I have corrected this part of the code in the new PR. Can it run now? Let me check.I don't have this issue on my machine. Could you please check if you can reproduce it?"

[dyzheng]

I have found the bug of this Issue: in file mixing_data.cpp and line https://github.com/deepmodeling/abacus-develop/blob/develop/source/module_base/module_mixing/mixing_data.cpp#L28

    if (ndim * length > 0)
    {
        this->data = malloc(ndim * length * type_size);
    }

In this case of Issue, the variables would have values of

ndim = 8
length = 303750000

and the INT value of "ndim * length" is 2.43e9 > 2.147e9 (INT limit)

The solution is use size_t rather than int in this code.

[WHUweiqingzhou]

Fixed by #5545