On Thu, Nov 6, 2008 at 2:01 AM, Ei-ji Nakama <[EMAIL PROTECTED]> wrote:

> Hi
>
> If you want to use MKL10(no 8 or 9), you should obey instructions of
> R-admin.
> MKL10 needs openmp.  This needs gcc4.2 or later.
> OpenMP for Intel was necessary with MKL10.0.1.x (There was not it in a
> manual of MKL).
> # Please try an additional effect of "-liomp5" in the %*%
>
> # fp-model of Linux of IA32 is 387(80bit) about the precision,
> # but SSE2(64bit) is a default in Intel64.
> # It is caused by this that results are different.
>
> There was not the problem in R-2.8.0 either
>  MKL_LIB_PATH=/opt/intel/mkl/10.0.5.025/lib/em64t
>  MKL="   -L${MKL_LIB_PATH}                               \
>             -Wl,--start-group                               \
>                     ${MKL_LIB_PATH}/libmkl_gf_lp64.a        \
>                     ${MKL_LIB_PATH}/libmkl_gnu_thread.a     \
>                     ${MKL_LIB_PATH}/libmkl_core.a           \
>             -Wl,--end-group                                 \
>             -liomp5 -lguide -lpthread -lgomp"
>  ./configure --with-lapack="$MKL" --with-blas="$MKL"


Thanks Ei-Ji,

I tried that earlier, and it doesn't encounter the zdotu error (maybe it was
fixed in 10.0.5?) but I got a different error:

gcc -std=gnu99 -I../../src/extra/zlib -I../../src/extra/bzip2
-I../../src/extra/pcre  -I. -I../../src/include -I../../src/include
-I/usr/local/include -DHAVE_CONFIG_H   -fpic  -g -O2 -c Rmain.c -o Rmain.o
gcc -std=gnu99 -Wl,--export-dynamic -L/usr/local/lib64 -o R.bin Rmain.o
-L../../lib -lR
/opt/intel/fce/10.1.018/lib/libiomp5.so: undefined reference to
`pthread_atfork'
collect2: ld returned 1 exit status
make[3]: *** [R.bin] Error 1
make[3]: Leaving directory `/working_copies/R/src/main'
make[2]: *** [R] Error 2

However, I saw the same error with 10.0.2 and the broken thermometer;
setting the environment variables listed previously:

CFLAGS=-pthread -O3
FC=gfortran -pthread
FFLAGS=-pthread
CXXFLAGS=-O3 -pthread
FCLAGS=-pthread
LDFLAGS=-lpthread

makes that error not happen, so it looks like we have a real solution!

Anand

        [[alternative HTML version deleted]]

______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

Reply via email to