Skip to content

R-0.64.0 installation on SGI IRIX 6.5 machine

4 messages · Jayshree Sarma, Peter Dalgaard

Jayshree Sarma
# 
Hi,

I am trying to install R-0.64.0 version on SGI machine running
IRIX 6.5.

Machine details:
319 <mind.krasnow.gmu.edu:R-0.63.3> uname -a
IRIX64 mind 6.5 07271714 IP27

Configure script runs fine and the output at the end of configure is:

R is now configured for mips-sgi-irix6.5

  Source directory:         /usr/local/R-0.64.0
  Installation directory:   /usr/local
  C compiler:               gcc -g
  FORTRAN compiler:         /usr/local/bin/g77  
  Gnome support enabled:    no


I am using gcc (egcs1.1.1) and g77 combination.
I use the GNU make(version 3.77) rather than the sgi make.
I went through your help mailing list and found that
I have to edit the Mathlib.h file in src/include directory
to undefine certain definitions which conflict with sgi's
definition.

When I run "gmake test", it fails; the details are given below.

....................
gmake[5]: Entering directory
`/usr/local/R-0.64.0/src/library/profile'
Building system startup profile
gmake[5]: Leaving directory `/usr/local/R-0.64.0/src/library/profile'
gmake[4]: Leaving directory `/usr/local/R-0.64.0/src/library'
gmake[3]: Leaving directory `/usr/local/R-0.64.0/src'
You should `make docs' now ...
gmake[2]: Leaving directory `/usr/local/R-0.64.0'
LC_ALL=C ../bin/R --vanilla < simple-true.R > simple-true.Rout
Comparing `simple-true.Rout' to
`/usr/local/R-0.64.0/tests/simple-true.Rout.save' ... OK
LC_ALL=C ../bin/R --vanilla < simple-tests.R > simple-tests.Rout
Comparing `simple-tests.Rout' to
`/usr/local/R-0.64.0/tests/simple-tests.Rout.save' ...17c17
< [1] FALSE
---
gmake[1]: *** [simple-tests.Rout] Error 1
gmake[1]: Leaving directory `/usr/local/R-0.64.0/tests'
gmake: [test-All] Error 2 (ignored)
***********************************

The test that fails from the simple-test.Rout file is
10^(-.Machine$double.max.exp*log10(2)))/Meps < 1e3

According to the file the test should return true but get a false on this
machine.

I ran R and printed the list of objects:

.....
.Machine 
List of 14
 $ double.eps           : num 2.22e-16
 $ double.neg.eps       : num 1.11e-16
 $ double.xmin          : num 2.23e-308
 $ double.xmax          : num 1.80e+308
 $ double.base          : int 2
 $ double.digits        : int 53
 $ double.rounding      : int 2
 $ double.guard         : int 0
 $ double.ulp.digits    : int -52
 $ double.neg.ulp.digits: int -53
 $ double.exponent      : int 11
 $ double.min.exp       : int -1022
 $ double.max.exp       : int 1024
 $ integer.max          : int 2147483647
......

All these outputs match with the numbers I found on the help mailing list.

Can someone tell why the test returns false? Should I be using some other
combination (sgi cc instead of gcc) to make this test work?

Thanks,

Jayshree Sarma
ps: reply to jsarma at cs.gmu.edu - I do not subscribe to the r-help list.


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard
# 
Jayshree Sarma <jsarma at cs.gmu.edu> writes:
The interesting bit of information missing from your post is what the 
actual value of the left hand side of that comparison is. On
Linux/i386 I get 560. If you get 1001 (say), I wouldn't worry too
much, but if it's some wild number, then you may be in deeper trouble.
(e.g. if log10(2) is not calculating the right value).
Jayshree Sarma
#
On 29 Apr 1999, Peter Dalgaard BSA wrote:

            

      
      
      
    

        
Given below is the output from R.  I do not get 560 but a rather large
number.  I ran the individual pieces too so you can compare the output
with the output from your machine and maybe tell me which is one that is
wrong. Is the code within abs()  correct?

***************output from R*************
Type    "q()" to quit R.

            

      
      
      
    

        
[1] 0.30103

            

      
      
      
    

        
[1] -1024

            

      
      
      
    

        
[1] 0

            

      
      
      
    

        
[1] 1.797693e+308

            

      
      
      
    

        
[1] 0

            

      
      
      
    

        
[1] 1

            

      
      
      
    

        
[1] 2.220446e-16

            

      
      
      
    

        
[1] 4.5036e+15

            

      
      
      
    

        
[1] FALSE

******************************end **************

Thanks,

Jayshree Sarma

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Peter Dalgaard
    

    

      
      #
    

  


  

      
Jayshree Sarma <jsarma at cs.gmu.edu> writes:

            

      
      
      
    

        
This is where the trouble starts. Other systems get a small number:
PD> 10^(-.Machine$double.max.exp*log10(2))
[1] 5.562685e-309

This underflow is probably no big deal, just an indication that your
system might not do perfect IEEE-standard arithmetic. Martin Maechler
may have more information on this.

            

      
      
      
    

        
Normally 1

            

      
      
      
    

        
Normally  (near-) zero.