Fresh compilation of McLachlan on Ranger failed with out of memory message

Create issue
Issue #613 closed
Bruno Mundim created an issue

A fresh compilation of McLachlan on Ranger failed with the following messages:

An internal threshold was exceeded for routine _Z17ML_BSSN_RHS1_BodyPK4_cGHiiPKdS3_S3_PKiS5_iPrKPd and optimization level may be reduced. An internal threshold was exceeded for routine _Z17ML_BSSN_RHS2_BodyPK4_cGHiiPKdS3_S3_PKiS5_iPrKPd and optimization level may be reduced. Fatal compilation error: Out of memory asking for 36864.

compilation aborted for ~/Cactus/configs/cigr_mix/build/ML_BSSN/ML_BSSN_RHS2.cc (code 1) gmake[3]: [ML_BSSN_RHS2.cc.o] Error 1 gmake[2]: [make.checked] Error 2 gmake[1]: [~/Cactus/configs/cigr_mix/lib/libthorn_ML_BSSN.a] Error 2 gmake: [cigr_mix] Error 2

I am using simfactory ranger optionlist: ranger-intel11.cfg. My bet goes for vectorization error. Any ideas?

Thanks.

Keyword: McLachlan

Comments (4)

  1. Ian Hinder
    • removed comment

    Are you using the most up-to-date version of the optionlist, and simfactory2? You should have

    VECTORISE_INLINE = no

    If you have this setting, then it must be something else. Do you have the latest version of the Vectors thorn? I can't see anything in the optionlist which would cause this. I compile with Intel 11.1 on Datura regularly and don't have this problem.

  2. Erik Schnetter
    • removed comment

    I built the ET on Ranger and submitted results on October 7. I updated the option list for Ranger after this.

  3. Bruno Mundim reporter
    • changed status to resolved
    • removed comment

    Ok, I was using on ranger a config file from the outdated location of simfactory2. So my config file had

    VECTORISE = yes VECTORISE_ALIGNED_ARRAYS = no

    but not

    VECTORISE_INLINE = no

    I am closing this ticket then. Thanks.

  4. Log in to comment