diff options
author | Albert Cheng <acheng@hdfgroup.org> | 2000-06-19 22:12:39 (GMT) |
---|---|---|
committer | Albert Cheng <acheng@hdfgroup.org> | 2000-06-19 22:12:39 (GMT) |
commit | 0f36ffd051a23323b1e29f70ec7e4d1e845583bb (patch) | |
tree | a34a63a69a1ab7c39f042a12410a9a9781ba0506 /INSTALL_TFLOPS | |
parent | 6ea3b11837c0b138507043bb101700aa13826812 (diff) | |
download | hdf5-0f36ffd051a23323b1e29f70ec7e4d1e845583bb.zip hdf5-0f36ffd051a23323b1e29f70ec7e4d1e845583bb.tar.gz hdf5-0f36ffd051a23323b1e29f70ec7e4d1e845583bb.tar.bz2 |
[svn-r2395] Updated the instructions.
Diffstat (limited to 'INSTALL_TFLOPS')
-rw-r--r-- | INSTALL_TFLOPS | 80 |
1 files changed, 57 insertions, 23 deletions
diff --git a/INSTALL_TFLOPS b/INSTALL_TFLOPS index 72cdec8..e842d8f 100644 --- a/INSTALL_TFLOPS +++ b/INSTALL_TFLOPS @@ -24,6 +24,23 @@ FROM SASN100, 1) cd <hdf5> 2) ./configure tflop +2.1) Due to a bug, you must first edit the files, src/Makefile and + test/Makefile with the following shell commands before doing make. + +# Patch up various Makefile's. +# patch up src/Makefile +ed - src/Makefile <<'EOF' +/^LT_RUN=.*/s//LT_RUN=$(RUNTEST)/ +w +q +EOF +# patch up test/Makefile +ed - test/Makefile <<'EOF' +/^RUNTEST=$(LT_RUN)/s/^/#/ +w +q +EOF + 3) make H5detect @@ -44,17 +61,16 @@ When everything is finished compiling and linking, you can run the tests by FROM JANUS, -7.1) Due to a bug, you must first remove the following line from - the file test/Makefile before the next step. - RUNTEST=$(LT_RUN) -7.2) make check -7.3) Sometimes the "make check" fails in the sub-directories of test - or tools with a message as "print not found". This is due to the - "make" of Janus thinking some binary code needs to be recompiled. - The easiest way to fix it is - FROM SASN100 - cd <hdf5>/test # or cd <hdf5>/tools - make clean; make # re-make all binary +7) make check + Note: Sometimes the "make check" fails in the sub-directories of test + or tools with a message as "print not found". This is due to the + "make" of Janus thinking some binary code needs to be recompiled. + I suspect it may be due to file timestamp problems between SASN100 + and JANUS. The easiest way to fix it is + + FROM SASN100 + cd <hdf5>/test # or cd <hdf5>/tools + make clean; make # re-make all binary Once satisfied with the test results, you can install @@ -79,7 +95,26 @@ FROM SASN100, 1) cd <hdf5> -2) sh bin/config_para_tflops.sh /* this is different from the sequential version */ +2) CC=/usr/community/mpich/mpich-1.1.2/bin/mpicc ./configure --enable-parallel + /* If you want to use your own MPI and/or MPIO libraries, */ + /* you may check the file bin/config_para_tflops.sh for details. */ +2.1) Due to a bug, you must first edit the files, src/Makefile and + test/Makefile with the following shell commands before doing make. + +# Patch up various Makefile's. +# patch up src/Makefile +ed - src/Makefile <<'EOF' +/^LT_RUN=.*/s//LT_RUN=$(RUNTEST)/ +w +q +EOF +# patch up test/Makefile +ed - test/Makefile <<'EOF' +/^RUNTEST=$(LT_RUN)/s/^/#/ +w +q +EOF + 3) make H5detect @@ -99,17 +134,16 @@ FROM SASN100, When everything is finished compiling and linking, FROM JANUS, -7.1) Due to a bug, you must first remove the following line from - the file test/Makefile before the next step. - RUNTEST=$(LT_RUN) -7.2) make check -7.3) Sometimes the "make check" fails in the sub-directories of test - or tools with a message as "print not found". This is due to the - "make" of Janus thinking some binary code needs to be recompiled. - The easiest way to fix it is - FROM SASN100 - cd <hdf5>/test # or cd <hdf5>/tools - make clean; make # re-make all binary +7) make check + Note: Sometimes the "make check" fails in the sub-directories of test + or tools with a message as "print not found". This is due to the + "make" of Janus thinking some binary code needs to be recompiled. + I suspect it may be due to file timestamp problems between SASN100 + and JANUS. The easiest way to fix it is + + FROM SASN100 + cd <hdf5>/test # or cd <hdf5>/tools + make clean; make # re-make all binary Once satisfied with the parallel test results, as long as you |