Error in calculating sigma

[sdwang]

Dear developers,
I used version-4.3.2 to calculate a 2D system. The calculation of chi is OK and after that the sigma stopped with
"
P0006: NULL parallel world for CPU#1
P0006: [ERROR] STOP signal received while in :[07] Bare local and non-local Exchange-Correlation
P0006: [ERROR]USER parallel structure does not fit the current run parameters. 72 CPU for 48 elements (ROLE is 'b'"

in the log file. The same system has been tested with 32 cores and the results are OK but now for 72 cores, it does not work.
Below is the input of the 2D system:
X_all_q_CPU= "1 1 1 72 1" # [PARALLEL] CPUs for each role
X_all_q_ROLEs= "q g k c v" # [PARALLEL] CPUs roles (q,g,k,c,v)
X_all_q_nCPU_LinAlg_INV= 1 # [PARALLEL] CPUs for Linear Algebra
SE_CPU= "1 1 72" # [PARALLEL] CPUs for each role
SE_ROLEs= "q qp b" # [PARALLEL] CPUs roles (q,qp,b)

Thanks!

SD

[Daniele Varsano]

Dear Shudong,
it seems you are using more CPU than available bands in your calculation.
Can you post your entire input and report?

Best,
Daniele

[sdwang]

Dear Daniele,
Thanks for your reply.
Attach are the input, log and report.
If as you said I used more cores than the occupied bands, how should I give the paralell option for my system?
Thanks!

Best

Shudong

[Daniele Varsano]

Dear Shudong,
the problem is in the calculation of the exchange part of the self energy where only occupied states are considered (44 states) so you cannot assign 72 cpus on bands.
By setting:

Code: Select all

SE_CPU= "1 2 36"                     # [PARALLEL] CPUs for each role
SE_ROLEs= "q qp b"      

or

Code: Select all

SE_CPU= "1 4 16"                     # [PARALLEL] CPUs for each role
SE_ROLEs= "q qp b"      

it should work without any loss of performance.
Best,
Daniele

[sdwang]

Thanks Daniele. I will try it.
Ciao

Shudong