Hello all-

I've been trying to compile ATLAS 3.4.1-1 on 10.1.5 (G3, 600MHz), but have
been having problems during the Level 1 BLAS tuning (I think that's what
it's doing).

After compiling LAPACK, doing the configuration, and starting the compile,
this happens:

> make[8]: `res/L1CacheSize' is up to date.
> ln -s /sw/src/atlas-3.4.1-1/ATLAS/tune/sysinfo/OSX_PPCG4/res/L1CacheSize
> res/L1CacheSize
> 
>     Read in L1 Cache size as = 32KB.
> 
> 
> FINDING ROUGHLY HOW MANY REGISTERS TO USE:
> 
> make mmcase pre=s loopO=JIK ta=T tb=N mb=80 nb=80 kb=80 lda=80 ldb=80 ldc=0
> mu=1 nu=1 ku=80 alpha=1 beta=1 muladd=1 lat=4 cleanup=1
> :
> /usr/bin/cc -c -DL2SIZE=4194304 -I/sw/src/atlas-3.4.1-1/ATLAS/include
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/contrib  -DAdd__ -DStringSunStyle
> -DATL_OS_OSX -DATL_ARCH_PPCG4 -fomit-frame-pointer -O3 -traditional-cpp
> -faltivec ../emit_mm.c
> /usr/bin/cc -DL2SIZE=4194304 -I/sw/src/atlas-3.4.1-1/ATLAS/include
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/contrib  -DAdd__ -DStringSunStyle
> -DATL_OS_OSX -DATL_ARCH_PPCG4 -fomit-frame-pointer -O3 -traditional-cpp
> -faltivec -o xemit_mm emit_mm.o
> cd /sw/src/atlas-3.4.1-1/ATLAS/bin/OSX_PPCG4 ; make xccobj
> /usr/bin/cc -DL2SIZE=4194304 -I/sw/src/atlas-3.4.1-1/ATLAS/include
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4
> -I/sw/src/atlas-3.4.1-1/ATLAS/include/contrib  -DAdd__ -DStringSunStyle
> -DATL_OS_OSX -DATL_ARCH_PPCG4 -fomit-frame-pointer -O3 -traditional-cpp
> -faltivec -o xccobj ../ccobj.c -DTOPDIR=/sw/src/atlas-3.4.1-1/ATLAS
> cp -f /sw/src/atlas-3.4.1-1/ATLAS/bin/OSX_PPCG4/xccobj .
> rm -f smm.c
> ./xemit_mm -p s -ta T -tb N -muladd 1 -l 4 \
>       -L JIK -M 80 -N 80 -K 80 -m 1 -n 1 -k 80 \
>       -lda 80 -ldb 80 -ldc 0 -csA 1 -csB 1 -csC 1\
>       -F0 -1 -FN 0 -FF 0 -b 1 -a 1 -Z 1 -C 1\
>> smm.c
> USAGE: ./xemit_mm -p <pre> -muladd <0/1> -ta <n/t> -tb <n/t> -l <lat> -m <mu>
> -n <nu> -k <ku> -M <MB> -N <NB> -K <KB> -lda <lda> -ldb <ldb> -ldc <ldc> -csA
> <col stride A> -csB <csb> -csC <csC> -a <alpha> -b <beta> -L <IJK/JIK> -f
> <file> -F0 <init fetch> -FN <nfetch> -FF <0/1> -Cleanup <0/1> -Z <0/1> -R
> <0/nb0> 
> make[7]: *** [mmcase] Error 255
> ERROR IN COMMAND:  make mmcase pre=s loopO=JIK ta=T tb=N mb=80 nb=80 kb=80
> lda=80 ldb=80 ldc=0 mu=1 nu=1 ku=80 alpha=1 beta=1 muladd=1 lat=4 cleanup=1
>  PROPOSED FILENAME: res/sJIKmmTN80_80x80x80_80x80x0_1x1x80_a1_b1_1x4_1
> make[6]: *** [res/snreg] Error 255
> make[5]: *** [res/snreg] Error 2
> ../GetSysSum.c:134: failed assertion `system(fnam) == 0'
> make[4]: *** [/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4/atlas_ssysinfo.h]
> Abort trap
> make[4]: *** Deleting file
> `/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4/atlas_ssysinfo.h'
> make[3]: *** [/sw/src/atlas-3.4.1-1/ATLAS/include/OSX_PPCG4/atlas_ssysinfo.h]
> Error 2
> make[2]: *** [IStage1] Error 2
> ERROR 396 DURING CACHESIZE SEARCH!!.  CHECK INSTALL_LOG/Stage1.log FOR
> DETAILS.
> cd ../.. ; make error_report arch=OSX_PPCG4
> make -f Make.top error_report arch=OSX_PPCG4
> uname -a 2>&1 >> bin/OSX_PPCG4/INSTALL_LOG/ERROR.LOG
> /usr/bin/cc -v 2>&1  >> bin/OSX_PPCG4/INSTALL_LOG/ERROR.LOG
> Reading specs from /usr/libexec/gcc/darwin/ppc/2.95.2/specs
> Apple Computer, Inc. version gcc-937.2, based on gcc version 2.95.2 19991024
> (release)
> /usr/bin/cc -V 2>&1  >> bin/OSX_PPCG4/INSTALL_LOG/ERROR.LOG
> cc: argument to `-V' is missing
> make[4]: [error_report] Error 1 (ignored)
> /usr/bin/cc --version 2>&1  >> bin/OSX_PPCG4/INSTALL_LOG/ERROR.LOG
> /sw/bin//tar  cf error_OSX_PPCG4.tar Make.OSX_PPCG4
> bin/OSX_PPCG4/INSTALL_LOG/*
> /sw/bin//gzip  --best error_OSX_PPCG4.tar
> mv error_OSX_PPCG4.tar.gz error_OSX_PPCG4.tgz
> Error report error_<ARCH>.tgz has been created in your top-level ATLAS
> directory.  Be sure to include this file in any help request.
> 
> 
> First, make sure this error is not covered in the ATLAS errata file,
>  http://math-atlas.sourceforge.net/errata.html
> All known errors, and most common mistakes/system problems are documented
> in this file, so it will save everyone hassle in the long run if you take
> the time to scope it out before sending e-mail to the help list.
> 
> If this does not have the solution to your problem, and you can't figure out
> what went wrong, submit a support request to the ATLAS support tracker, as
> discussed here:
>  http://math-atlas.sourceforge.net/faq.html#help
> 
> In the above examples, <ARCH> should be replaced with your architecture
> string, such as "Linux_PIII" or "FreeBSD_21264".
> 
> make[1]: *** [install] Error 255
> make: *** [install] Error 2
> ### execution of  failed, exit code 2
> Failed: compiling atlas-3.4.1-1 failed
> [localhost:~] josh%

I have the .tgz file that it spit out should anyone need it, along with
everything that was written to Terminal.app during this.

A thing I found odd: It identified my system as 'OSX_PPCG4' but I don't have
a G4. I went through the configuration with all of the defaults (but it
never gave me a chance to select UNKNOWN as detailed in 'fink describe
atlas').

Thanks in advance!
-- 

-- sinclair44

"I'm a doctor, not a mechanic."
                -- "The Doomsday Machine", when asked if he had heard of
                   the idea of a doomsday machine.
"I'm a doctor, not an escalator."
                -- "Friday's Child", when asked to help the very pregnant
                   Ellen up a steep incline.
"I'm a doctor, not a bricklayer."
                -- Devil in the Dark", when asked to patch up the Horta.
"I'm a doctor, not an engineer."
                -- "Mirror, Mirror", when asked by Scotty for help in
                   Engineering aboard the ISS Enterprise.
"I'm a doctor, not a coalminer."
                -- "The Empath", on being beneath the surface of Minara 2.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Fink-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-users

Reply via email to