Hi all,
I'm encountering the same building error as described by Hamideh Hosseini on Fri, 16 Jul 2021 07:11:58 -0700 (sorry, can't seem to find a way to reply to that thread).
The error is
[test-vectors-4-SSE.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hhossei/Cactus/configs/sim/lib/libthorn_Vectors.a] Error 2 & Creating /home/hhossei/Cactus/configs/sim/lib/libthorn_Fortran.a make: *** [sim] Error 2
I'm working on a cluster with Linux Centos7. As suggested by Roland Haas on the previous thread, I attach the files
make.log configs/sim/config-info configs/sim/config-data/config.log and configs/sim/config-data/make.config.defn
Do you have any advice?
Thank you! Tomas
Hello Tomas,
thank you for including the files.
Looking at make.config.defn you are using a very old compiler (gcc 4.8).
The best solution would be to user a newer compiler. On clusters you will typically have to load an environment module (you can try "module avail" to see which ones are available).
The "vectors" thorn where you see errors uses some C++ templating that may not be supported by that compiler (gcc 4.8 was released in 2014, https://gcc.gnu.org/releases.html).
If there is no newer gcc compiler you may also be able to use the Intel compiler (if it is installed on the cluster).
If there is a helpdesk then you may also be able to ask them about how to use a newer compiler.
Yours, Roland
Hi all,
I'm encountering the same building error as described by Hamideh Hosseini on Fri, 16 Jul 2021 07:11:58 -0700 (sorry, can't seem to find a way to reply to that thread).
The error is
[test-vectors-4-SSE.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hhossei/Cactus/configs/sim/lib/libthorn_Vectors.a] Error 2 & Creating /home/hhossei/Cactus/configs/sim/lib/libthorn_Fortran.a make: *** [sim] Error 2
I'm working on a cluster with Linux Centos7. As suggested by Roland Haas on the previous thread, I attach the files
make.log configs/sim/config-info configs/sim/config-data/config.log and configs/sim/config-data/make.config.defn
Do you have any advice?
Thank you! Tomas
thank you Roland for your super quick response!
If I do 'module avail', I see that the default gcc compiler is gcc/11.1 (the earliest version being gcc/8.2), do you know why the build could take another version?
thanks! Tomas
On Fri, 23 Jul 2021 at 17:18, Roland Haas rhaas@illinois.edu wrote:
Hello Tomas,
thank you for including the files.
Looking at make.config.defn you are using a very old compiler (gcc 4.8).
The best solution would be to user a newer compiler. On clusters you will typically have to load an environment module (you can try "module avail" to see which ones are available).
The "vectors" thorn where you see errors uses some C++ templating that may not be supported by that compiler (gcc 4.8 was released in 2014, https://gcc.gnu.org/releases.html).
If there is no newer gcc compiler you may also be able to use the Intel compiler (if it is installed on the cluster).
If there is a helpdesk then you may also be able to ask them about how to use a newer compiler.
Yours, Roland
Hi all,
I'm encountering the same building error as described by Hamideh Hosseini on Fri, 16 Jul 2021 07:11:58 -0700 (sorry, can't seem to find a way to reply to that thread).
The error is
[test-vectors-4-SSE.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hhossei/Cactus/configs/sim/lib/libthorn_Vectors.a] Error 2 & Creating /home/hhossei/Cactus/configs/sim/lib/libthorn_Fortran.a make: *** [sim] Error 2
I'm working on a cluster with Linux Centos7. As suggested by Roland Haas
on
the previous thread, I attach the files
make.log configs/sim/config-info configs/sim/config-data/config.log and configs/sim/config-data/make.config.defn
Do you have any advice?
Thank you! Tomas
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .
Hello Tomas,
thank you Roland for your super quick response!
If I do 'module avail', I see that the default gcc compiler is gcc/11.1 (the earliest version being gcc/8.2), do you know why the build could take another version?
The one marked "default" is the default one when using modules, ie if you do (schematically):
module load gcc
it will load gcc/11.1.
However you must use "module load gcc", if you do nothing at all, then your compile will use the system installed gcc compiler (from the RedHat Linux base system) in /usr/bin.
Yours, Roland
I see, thank you, Roland!
I loaded the gcc module and I'm not getting that error anymore. However, there is now another error related to dwarf
/usr/bin/ld: Dwarf Error: found dwarf version '7424', this reader only handles version 2, 3 and 4 information.
/home/lv65/lv65221/ETK/Cactus/configs/sim/scratch/external/MPI/lib/libopen-pal.a(btl_openib_async.o): In function `btl_openib_async_device':
btl_openib_async.c:(.text+0x3a9): undefined reference to `ibv_get_async_event'
btl_openib_async.c:(.text+0x43d): undefined reference to `ibv_ack_async_event'
btl_openib_async.c:(.text+0x79d): undefined reference to `ibv_get_device_name'
collect2: error: ld returned 1 exit status
make[1]: *** [/home/lv65/lv65221/ETK/Cactus/exe/cactus_sim] Error 1
make: *** [sim] Error 2
Googling a bit, it seems that this can be solved specifying some compilation flags correctly, but I don't have enough experience to do it.
Have you seen this behaviour before?
thanks! Tomas
On Fri, 23 Jul 2021 at 17:36, Roland Haas rhaas@illinois.edu wrote:
Hello Tomas,
thank you Roland for your super quick response!
If I do 'module avail', I see that the default gcc compiler is gcc/11.1 (the earliest version being gcc/8.2), do you know why the build could
take
another version?
The one marked "default" is the default one when using modules, ie if you do (schematically):
module load gcc
it will load gcc/11.1.
However you must use "module load gcc", if you do nothing at all, then your compile will use the system installed gcc compiler (from the RedHat Linux base system) in /usr/bin.
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .
sorry, here are my logs
On Fri, 23 Jul 2021 at 22:40, Tomas Andrade tandrade@icc.ub.edu wrote:
I see, thank you, Roland!
I loaded the gcc module and I'm not getting that error anymore. However, there is now another error related to dwarf
/usr/bin/ld: Dwarf Error: found dwarf version '7424', this reader only handles version 2, 3 and 4 information.
/home/lv65/lv65221/ETK/Cactus/configs/sim/scratch/external/MPI/lib/libopen-pal.a(btl_openib_async.o): In function `btl_openib_async_device':
btl_openib_async.c:(.text+0x3a9): undefined reference to `ibv_get_async_event'
btl_openib_async.c:(.text+0x43d): undefined reference to `ibv_ack_async_event'
btl_openib_async.c:(.text+0x79d): undefined reference to `ibv_get_device_name'
collect2: error: ld returned 1 exit status
make[1]: *** [/home/lv65/lv65221/ETK/Cactus/exe/cactus_sim] Error 1
make: *** [sim] Error 2
Googling a bit, it seems that this can be solved specifying some compilation flags correctly, but I don't have enough experience to do it.
Have you seen this behaviour before?
thanks! Tomas
On Fri, 23 Jul 2021 at 17:36, Roland Haas rhaas@illinois.edu wrote:
Hello Tomas,
thank you Roland for your super quick response!
If I do 'module avail', I see that the default gcc compiler is gcc/11.1 (the earliest version being gcc/8.2), do you know why the build could
take
another version?
The one marked "default" is the default one when using modules, ie if you do (schematically):
module load gcc
it will load gcc/11.1.
However you must use "module load gcc", if you do nothing at all, then your compile will use the system installed gcc compiler (from the RedHat Linux base system) in /usr/bin.
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .
Hello Tomas,
I have not looked at he logs, but errors from ancient linkers (such as your systems /usr/bin/ld) with modern gcc compilers usually mean that the "binutils" package is too old.
If you can I would check if the output of "module avail" has any likely looking module (eg binutils or ld or so) and load that one.
Alternatively, you can use a (slightly) older gcc. Actually anything starting from gcc 4.9 should do (just not 4.8). I know for a fact that the toolkit compiles with eg gcc 6.3 which may be old enough for your linker.
Yours, Roland
sorry, here are my logs
On Fri, 23 Jul 2021 at 22:40, Tomas Andrade tandrade@icc.ub.edu wrote:
I see, thank you, Roland!
I loaded the gcc module and I'm not getting that error anymore. However, there is now another error related to dwarf
/usr/bin/ld: Dwarf Error: found dwarf version '7424', this reader only handles version 2, 3 and 4 information.
/home/lv65/lv65221/ETK/Cactus/configs/sim/scratch/external/MPI/lib/libopen-pal.a(btl_openib_async.o): In function `btl_openib_async_device':
btl_openib_async.c:(.text+0x3a9): undefined reference to `ibv_get_async_event'
btl_openib_async.c:(.text+0x43d): undefined reference to `ibv_ack_async_event'
btl_openib_async.c:(.text+0x79d): undefined reference to `ibv_get_device_name'
collect2: error: ld returned 1 exit status
make[1]: *** [/home/lv65/lv65221/ETK/Cactus/exe/cactus_sim] Error 1
make: *** [sim] Error 2
Googling a bit, it seems that this can be solved specifying some compilation flags correctly, but I don't have enough experience to do it.
Have you seen this behaviour before?
thanks! Tomas
On Fri, 23 Jul 2021 at 17:36, Roland Haas rhaas@illinois.edu wrote:
Hello Tomas,
thank you Roland for your super quick response!
If I do 'module avail', I see that the default gcc compiler is gcc/11.1 (the earliest version being gcc/8.2), do you know why the build could
take
another version?
The one marked "default" is the default one when using modules, ie if you do (schematically):
module load gcc
it will load gcc/11.1.
However you must use "module load gcc", if you do nothing at all, then your compile will use the system installed gcc compiler (from the RedHat Linux base system) in /usr/bin.
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from https://urldefense.com/v3/__http://pgp.mit.edu__;!!DZ3fjg!pTF0AgYZuq90rTZuCY... .
users@lists.einsteintoolkit.org