[R] R-devel and gcc 3.4

[Martin Maechler]

>>>>> "Robert" == Robert Kruus <robert.kruus at utoronto.ca>
>>>>>     on Fri, 20 Aug 2004 16:33:33 -0400 writes:

    Robert> Here: CFLAGS=-Os -march=athlon-xp -pipe
    Robert> -fomit-frame-pointer -ftracer

is this now gcc (w/o -O) or the intel one?

In any case, your numeric results seem completely off target,
not just slightly inaccurate..

    Robert> Will try some other flags to see if that is an
    Robert> issue.

  comparing 'd-p-q-r-tests.Rout' to './d-p-q-r-tests.Rout.save'
  ...196,227d195< [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 16.72081"
  < [1] "Mean scaled  difference: 60.41034"
  < [1] "Mean scaled  difference: 60.41034"
  < [1] "Mean scaled  difference: 22472.41"
  < [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 450.4906"
  < [1] "Mean scaled  difference: 117.9663"
  < [1] "Mean scaled  difference: 11833.79"
  < [1] "Mean scaled  difference: 117.9663"
  < [1] "Mean scaled  difference: 874.8262"
  < [1] "Mean scaled  difference: 117.9663"
  < [1] "Mean scaled  difference: 31.44371"
  < [1] "Mean scaled  difference: 60.41034"
  < [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 1689.909"
  < [1] "Mean scaled  difference: 450.4906"
  < [1] "Mean scaled  difference: 1689.909"
  < [1] "Mean scaled  difference: 31.44371"
  < [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 7.480203"
  < [1] "Mean scaled  difference: 3245.883"
  < [1] "Mean scaled  difference: 450.4906"
  < [1] "Mean scaled  difference: 60.41034"
  < [1] "Mean scaled  difference: 80399.63"
  < [1] "Mean scaled  difference: 11833.79"
  < [1] "Mean scaled  difference: 230.9933"
  < [1] "Mean scaled  difference: 450.4906"
  < [1] "Mean scaled  difference: 450.4906"
  < [1] "Mean scaled  difference: 874.8262"
  < [1] "Mean scaled  difference: 230.9933"
  240,314d207
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 136.1327"
  < [1] "Mean scaled  difference: 203.5565"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 114.8797"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 114.8797"
  < [1] "Mean scaled  difference: 227.0455"
  < [1] "Mean scaled  difference: 114.8797"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 94.3997"
  < [1] "Mean scaled  difference: 56.3034"
  < [1] "Mean scaled  difference: 350.2388"
  < [1] "Mean scaled  difference: 275.2651"
  < [1] "Mean scaled  difference: 397.2022"
  < [1] "Mean scaled  difference: 483.6942"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 161.1166"
  < [1] "Mean scaled  difference: 126.6956"
  < [1] "Mean scaled  difference: 397.2022"
  < [1] "Mean scaled  difference: 483.6942"
  < [1] "Mean scaled  difference: 161.1166"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 440.0712"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 161.1166"
  < [1] "Mean scaled  difference: 528.008"
  < [1] "Mean scaled  difference: 94.3997"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 483.6942"
  < [1] "Mean scaled  difference: 234.9492"
  < [1] "Mean scaled  difference: 64.668"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 126.6956"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 197.2793"
  < [1] "Mean scaled  difference: 355.1587"
  < [1] "Mean scaled  difference: 94.3997"
  < [1] "Mean scaled  difference: 314.0298"
  < [1] "Mean scaled  difference: 234.9492"
  < [1] "Mean scaled  difference: 314.0298"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 180.5405"
  < [1] "Mean scaled  difference: 74.79829"
  < [1] "Mean scaled  difference: 114.8797"
  < [1] "Mean scaled  difference: 56.3034"
  < [1] "Mean scaled  difference: 250.9591"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 180.5405"
  < [1] "Mean scaled  difference: 74.79829"
  < [1] "Mean scaled  difference: 136.1327"
  < [1] "Mean scaled  difference: 203.5565"
  < [1] "Mean scaled  difference: 203.5565"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 158.0424"
  < [1] "Mean scaled  difference: 94.3997"
  < [1] "Mean scaled  difference: 403.9424"
  < [1] "Mean scaled  difference: 337.4193"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 337.4193"
  < [1] "Mean scaled  difference: 213.9039"
  < [1] "Mean scaled  difference: 403.9424"
  < [1] "Mean scaled  difference: 62.57874"
  < [1] "Mean scaled  difference: 618.1512"
  < [1] "Mean scaled  difference: 770.7226"
  < [1] "Mean scaled  difference: 337.4193"
  < [1] "Mean scaled  difference: 618.1512"
  < [1] "Mean scaled  difference: 213.9039"
  < [1] "Mean scaled  difference: 273.9165"
  < [1] "Mean scaled  difference: 618.1512"
  < [1] "Mean scaled  difference: 213.9039"
  766,768c659,662
  <  [1] -5.576862 -6.208975 -6.008831 -7.604174 -5.757453 -5.719751
  -5.477951<  [8] -6.328233 -5.505609 -6.041780 -6.075742 -5.528914
  -5.503182 -6.471671< [15] -6.899387 -5.477143 -6.626737 -5.646456
  -7.929201 -5.727071---
  >  [1] 0.0037844241 0.0020112978 0.0024569596 0.0004983669 0.0031591482
  >  [6] 0.0032805288 0.0041778794 0.0017851857 0.0040639142 0.0023773231
  > [11] 0.0022979400 0.0039702993 0.0040737886 0.0015466398 0.0010084031
  > [16] 0.0041812581 0.0013244780 0.0035300054 0.0003600738 0.0032566009
  783,785c677,680
  <  [1] -4.528263 -4.154630 -4.154630 -6.125949 -4.143093 -4.113462
  -4.281849<  [8] -4.167814 -4.154630 -4.124981 -4.108528 -4.391428
  -4.785674 -4.306719< [15] -5.049338 -5.874622 -4.333257 -5.359233
  -4.785674 -4.258675---
  >  [1] 0.010799421 0.015691594 0.015691594 0.002185415 0.015873676
  >  0.016351073[7] 0.013817085 0.015486075 0.015691594 0.016163798
  >  0.016431951 0.012383032
  > [13] 0.008348499 0.013477697 0.006413577 0.002809857 0.013124724
  > 0.004704515[19] 0.008348499 0.014141025
  make[2]: *** [d-p-q-r-tests.Rout] Error 1


    Robert> -- robert.kruus at utoronto.ca I am what you will be; I
    Robert> was what you are.


    Robert> On 20 Aug 2004 22:23:30 +0200 it is rumored that
    Robert> Peter Dalgaard <p.dalgaard at biostat.ku.dk> wrote:
 
    PD> "Samuelson, Frank*" <FWS4 at cdrh.fda.gov> writes:
    PD> 
    PD> > I encountered a similar problem with > 1. Intel
    PD> compilers.  > 2. gcc when I turn off optimizations
    PD> (no -Ox)
    PD> 
    PD> Ooops. I didn't notice #2 the first time around. If
    PD> we're failing the checks with no optimization, then
    PD> we surely have set the tolerances too tight.

well, not in this case.
The above numbers are *way* off.

Yes, Robert, do try other compilere switches.