Fifty integrated test cases failed on Intel env

[hongriTianqi]

Describe Current Status and Possible Solution

The following 50 test cases failed on intel machine.

1: 107_PW_outWfcR
1: 107_PW_W90
1: 109_PW_CR_fix_abc
1: 111_PW_elec_add
1: 116_PW_scan_Si2
1: 116_PW_scan_Si2_nspin2
1: 127_PW_15_PK_AF
1: 150_PW_15_CR_VDW3
1: 184_PW_BNDKPAR_SDFT_ALL
1: 184_PW_BNDKPAR_SDFT_MALL
1: 201_NO_KP_DJ_CF_CS_GaAs
1: 201_NO_KP_DJ_Si
1: 208_NO_KP_CS_CR
1: 213_NO_mulliken
1: 215_NO_sol_H2
1: 216_NO_scan_Si2
1: 250_NO_KP_CR_VDW2
1: 250_NO_KP_CR_VDW3
1: 250_NO_KP_CR_VDW3ABC
1: 250_NO_KP_CR_VDW3BJ
1: 281_NO_KP_HSE
1: 282_NO_KP_HSE_complex
1: 283_NO_KP_HF
1: 284_NO_KP_PBE0
1: 285_NO_KP_RE_HSE
1: 286_NO_KP_CR_HSE
1: 283_NO_restart
1: 307_NO_GO_OH
1: 381_NO_GO_S1_HSE
1: 382_NO_GO_S2_HSE
1: 383_NO_GO_SO_HSE
1: 384_NO_GO_S1_HSE_loop0_PU
1: 385_NO_GO_RE_S1_HSE
1: 386_NO_GO_MD_S1_HSE
1: 601_NO_TDDFT_CO_occ
1: 801_PW_LT_sc
1: 802_PW_LT_fcc
1: 803_PW_LT_bcc
1: 804_PW_LT_hexagonal
1: 805_PW_LT_trigonal
1: 806_PW_LT_st
1: 807_PW_LT_bct
1: 808_PW_LT_so
1: 809_PW_LT_baco
1: 810_PW_LT_fco
1: 811_PW_LT_bco
1: 812_PW_LT_sm
1: 813_PW_LT_bacm
1: 814_PW_LT_triclinic
1: 824_NO_LT_fco

Additional Context

No response

[hongriTianqi]

Details can be checked in the following txt file. We need to further check each case to spot out the issues respectively.
unit_test_intel.txt

[jinzx10]

@hongriTianqi I had a detailed study on the first case "107_PW_outWfcR" and found that it might be mainly a format / post-processing issue.

An integrated test invoked by Autotest.sh (1) calls abacus to run some job first, (2) uses the bash script tools/catch_properties.sh to extract & calculate some values, and (3) compare the values with those in result.ref.

In "107_PW_outWfcR", I find that abacus finish the job normally; the problem is in step-2 where tools/catch_properties.sh read OUT.Autotest/running_scf.log to get the total number of fft grid (line 256):

The variable "allgrid" is zero after line 256, which makes further calculation in line 259 yield inf. This is definitely not what we want, but it does faithfully do what command asks for. The following figure shows the place in running_scf.log from which the number is calculated:

As we can see, by using "=" and "," as delimiters, field 2 is "[ 24" instead of "24", thereby causing $2*$3*$4 to be 0.

I notice that format in running_scf.log has a recent change in #2605. Prior to this commit, the same information was displayed as

In this situation, $2*$3*$4 would yield the right number.

This test case can be fixed by either changing the output format of "fft grid for wave functions" back before or change the awk command in post-processing. However, I'm not able to make the desicion because each one has the potential to cause the failure of other tests. The decision should be left to people who can assess the situation and consequence of such modification. @dyzheng @hongriTianqi

[hongriTianqi]

• Understand the problem or question described by the user.
• Check if the issue is a known problem or has been addressed in the documentation.
• Test the issue or problem on a similar system or environment, if possible.
• Identify the root cause or provide clarification on the user's question.
• Provide a step-by-step guide, including any necessary resources, to resolve the issue or answer the question.
• If the issue is related to documentation, update the documentation to prevent future confusion (optional).
• If the issue is related to code, consider implementing a fix or improvement (optional).
• Review and incorporate any relevant feedback from users or developers.
• Ensure the user's issue is resolved or their question is answered and close the ticket.