Hi, I was having problems getting past the './configure' step on my linux system -- checking for -lcln to be exact. I eventually learned to look at the 'config.log' file and realized the problem. While compiling the test program for the "-lcln" check my system ran out of virtual memory (see config.log snippet and /proc/meminfo data below). However, if I try to compile the test program on my own - outside the 'configure' script - then I have no problems. Any ideas about how to get around this? I could just kluge the 'configure' script since I know '-lcln' works. I'm wondering though if my 233 MHz,32MB RAM linux box isn't going to be able to handle ginac. Any ideas? In the mean time, I'm trying to get ginac installed our Sun Enterprise servers at work but I'm having trouble there too (but I'll tinker with that some more before bugging all of you!) Thanks ~ pat <notz@purdue.edu> FROM 'config.log' . . . configure:2558: checking for cln.h configure:2568: c++ -E conftest.C >/dev/null 2>conftest.out configure:2603: checking for doublefactorial in -lcln configure:2639: c++ -o conftest -g -O2 conftest.C -lcln 1>&5 /usr/local/include/cl_SV_rational.h: In method `cl_SV<cl_RA,cl_SV_R>::~cl_SV()': In file included from /usr/local/include/cln.h:119, from configure:2633: /usr/local/include/cl_SV_rational.h:18: virtual memory exhausted configure: failed program was: #line 2632 "configure" #include "confdefs.h" #include <cln.h> int main() { doublefactorial(2); ; return 0; } configure:2680: checking for doxygen configure:2716: checking for latex . . . localhost# cat /proc/meminfo total: used: free: shared: buffers: cached: Mem: 31563776 22753280 8810496 18296832 1040384 10530816 Swap: 53637120 25653248 27983872 MemTotal: 30824 kB MemFree: 8604 kB MemShared: 17868 kB Buffers: 1016 kB Cached: 10284 kB SwapTotal: 52380 kB SwapFree: 27328 kB - To UNSUBSCRIBE, email to ginac-list@ginac.de with a subject of "unsubscribe".
On Wed, 23 Feb 2000, Patrick K Notz wrote:
know '-lcln' works. I'm wondering though if my 233 MHz,32MB RAM linux box isn't going to be able to handle ginac. Any ideas?
I have a 133 MHz Pentium with 32 MB at home and it is possible to compile GiNaC on this machine.
configure:2639: c++ -o conftest -g -O2 conftest.C -lcln 1>&5 ^^^ Try to compile GiNaC without optimization. The optimization step eats a lot of memory (same in the check directory).
total: used: free: shared: buffers: cached: Mem: 31563776 22753280 8810496 18296832 1040384 10530816 Swap: 53637120 25653248 27983872
Or add some more swap space. You can add swap space during run time with dd if=/dev/zero of=swapfile bs=1024 count=65536 mkswap swapfile swapon swapfile for an additional 64 MB swap file. IIRC the -lcln step in configure needs about 70MB, so your 32 MB physical + 53 MB swap is not enough if you have some other processes like X around. Best regards Alex -- Alexander Frink E-Mail: Alexander.Frink@Uni-Mainz.DE Institut fuer Physik Phone: +49-6131-3923391 Johannes-Gutenberg-Universitaet D-55099 Mainz, Germany - To UNSUBSCRIBE, email to ginac-list@ginac.de with a subject of "unsubscribe".
On Thu, 24 Feb 2000, Alexander Frink wrote:
On Wed, 23 Feb 2000, Patrick K Notz wrote:
know '-lcln' works. I'm wondering though if my 233 MHz,32MB RAM linux box isn't going to be able to handle ginac. Any ideas?
I have a 133 MHz Pentium with 32 MB at home and it is possible to compile GiNaC on this machine.
Just to clarify something: Patrick mentions that he could compile the test program by hand without such memory consumption. The waste of memory is most certainly due to optimization of headers in <cln/cln.h>. If you compile it by hand you probably didn't optimize. configure wants to set optimization to -O2 if no CXXFLAGS were set so it compile the test program with that level, too. This should account for the difference. I have just fixed it in current CVS. Configure runs smoother now because the test program only includes <cln/cl_integer.h> (assuming it *must* be where <cln/cln.h> was found, which is a safe assumtion). But in any case, this does not spare you a costly compilation of ginac/numeric.cpp. Regards -rbk. PS: I just observed a current prerelease of gcc consume 220MB, so the inliner seems to be effective with heavy use of templates only (as advertiesed). We'd all better buy some memory soon... -- Richard Kreckel <Richard.Kreckel@Uni-Mainz.DE> <http://wwwthep.physik.uni-mainz.de/~kreckel/> - To UNSUBSCRIBE, email to ginac-list@ginac.de with a subject of "unsubscribe".
participants (3)
-
Alexander Frink
-
Patrick K Notz
-
Richard B. Kreckel