Pour le coup du . dans le path, franchement je vais pas regarder tous les scripts qui se lancent au démarrage pour chercher le coupable. Voila, j'assume la "faille de sécurité". On verra quand j'aurai plus que ça à faire.

PpHd (./14) :
Tu as des bench ?

Alors, oui, et pour ma machine en plus. J'ai suivi la page du Dr Nakajima, oups pardon, Tajima (http://www.apphy.fukui-u.ac.jp/~tajima/bench/run_e.html)
Résumé du score, 9.44 pour g77 contre 32.16 pour ifort
Comparaison :

measurement date = 08/12/08
log file = bench_g77.log
machine = 2* (Xeon5430 4*2.66 GHz), Linux 2.6.18-8.el5 x86_64 (g77 3.4.6) 
compile with [ g77 ]

        program weight  ref.time time[sec]  time/ref  ref/time
( 1)     matvec      8    296.00     32.77    0.1107    9.0324
( 2)    matvecz      4    279.00     16.53    0.0593   16.8774
( 3)     leqs4h      3    256.00     17.25    0.0674   14.8423
( 4)     leqs1k      3    279.00     43.79    0.1569    6.3715
( 5)   jacobi11      3    261.00     32.64    0.1251    7.9966
( 6)   jacobi1h      3    124.00     22.13    0.1785    5.6022
( 7)      runge      8    204.00     38.81    0.1902    5.2569
( 8)     intgl4      4    300.00     38.34    0.1278    7.8247
( 9)   mathfnc1      2    123.00      7.88    0.0640   15.6190
(10)   mathfnc2      2    226.00     18.50    0.0819   12.2129
(11)     intosc      2    236.00     10.63    0.0450   22.2055
(12)     intpi3      2    122.00     13.65    0.1119    8.9403
(13)   permute1      2     23.00      4.37    0.1902    5.2583
(14)   permute2      2     54.00      4.55    0.0843   11.8577
(15)   permute3      2    228.00      5.62    0.0247   40.5622

  weighted arithmetic  average of (ref/time) =     11.02
  weighted geometrical average of (ref/time) =      9.44
1/weighted arithmetic  average of (time/ref) =      8.41

               fMark =      9.44

----------------------------------------------------------------------

measurement date = 08/12/08
log file = bench_ifort.log
machine = 2* (Xeon5430 4*2.66 GHz), Linux 2.6.18-8.el5 x86_64 (ifort 10.1.008) 
compile with [ ifort ]

        program weight  ref.time time[sec]  time/ref  ref/time
( 1)     matvec      8    296.00     15.89    0.0537   18.6246
( 2)    matvecz      4    279.00      8.97    0.0321   31.1176
( 3)     leqs4h      3    256.00      5.65    0.0221   45.2777
( 4)     leqs1k      3    279.00     14.27    0.0512   19.5488
( 5)   jacobi11      3    261.00      8.95    0.0343   29.1751
( 6)   jacobi1h      3    124.00      3.03    0.0245   40.8836
( 7)      runge      8    204.00      8.53    0.0418   23.9044
( 8)     intgl4      4    300.00      6.17    0.0206   48.6460
( 9)   mathfnc1      2    123.00      2.74    0.0223   44.8905
(10)   mathfnc2      2    226.00      3.75    0.0166   60.2345
(11)     intosc      2    236.00      4.38    0.0186   53.8444
(12)     intpi3      2    122.00      4.62    0.0379   26.4184
(13)   permute1      2     23.00      0.91    0.0397   25.1917
(14)   permute2      2     54.00      1.02    0.0190   52.7344
(15)   permute3      2    228.00      2.37    0.0104   96.4059

  weighted arithmetic  average of (ref/time) =     35.67
  weighted geometrical average of (ref/time) =     32.16
1/weighted arithmetic  average of (time/ref) =     29.42

               fMark =     32.16

Pour les passionnés, le lien vers mon dossier

Edit: d'autres bench http://www.macresearch.org/polyhedron_2005_fortran_benchmark_quad_xeon_macpro