search for: xopenmp

...onv in /usr/local/lib (no sure why I am getting the referencing error). The config.site file is attached. ANY HELP YOU CAN GIVE WOULD BE GREATLY APPRECIATED! :) ./configure --without-readline DYLIB_LDFLAGS=-xarch=v9 MAIN_LDFLAGS=-xarch=v9 Then I run /usr/ccs/bin/make cc -xc99=all -xarch=v9 -xopenmp -L/usr/local/bin -o R.bin Rmain.o libR.a -L../../lib -lRblas -R/usr/local/opt/SUNWspro/lib/v9:/opt/SUNWspro/lib/v9 -L/usr/local/opt/SUNWspro/lib/v9 -L/usr/local/opt/SUNWspro/prod/lib/v9 -L/usr/ccs/lib/sparcv9 -L/lib/sparcv9 -L/usr/lib/sparcv9 -lfui -lfai -lfai2 -lfsumai -lfprodai -lfminlai -lfmaxla...

...bs library so that they're available for later libraries. The configure script is ignoring the -z extraction directives, resulting in an out-of-order sequence of libs, and a subsequent failure to resolve the symbols listed above. Solution: ------------ As a temporary workaround, adding "-xopenmp=stubs" to LDFLAGS seems to work. For a more permanent solution, I think the -z extraction directives need to be preserved.

...ma omp target map(to: a[0:size])   #pragma omp teams distribute   for (int i = 0; i < size; ++i) {     a[i] = a[i]*(i*i)/2;   }   return 0; } PD: The command to compile is this: clang -v -save-temps \ -I/path/to/llvm/release/9.x/include -fopenmp \ -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target \ -march=sm_61 memory_transfer.c -o memory_transfer Thanks. Hamilton. On 23/03/20 11:11 a. m., Doerfert, Johannes wrote: > On 3/22/20 12:33 PM, Hamilton Tobon Mosquera wrote: > > Ahhh I understand. So I believe the project you posted is still > opened. Do you any adv...

...i)/2; > >   } > > > >   return 0; > > } > > > > > > PD: The command to compile is this: > >     clang -v -save-temps \ > >     -I/path/to/llvm/release/9.x/include -fopenmp \ > >     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target \ > >         -march=sm_61 memory_transfer.c -o memory_transfer > > > > Thanks. > > Hamilton. > > > > On 23/03/20 11:11 a. m., Doerfert, Johannes wrote: > >> On 3/22/20 12:33 PM, Hamilton Tobon Mosquera wrote: > >>   >...

...ILITIES: * R CMD check checks for and R CMD build corrects CRLF line endings in shell scripts configure and cleanup (even on Windows). INSTALLATION on a UNIX-ALIKE: * The order of selection of OpenMP flags has been changed: Oracle Developer Studio 12.5 accepts -fopenmp and -xopenmp but only the latter enables OpenMP so it is now tried first. BUG FIXES: * within(List, rm(x1, x2)) works correctly again, including when List[["x2"]] is NULL. * regexec(pattern, text, *) now applies as.character(.) to its first two arguments, as documented....

...ILITIES: * R CMD check checks for and R CMD build corrects CRLF line endings in shell scripts configure and cleanup (even on Windows). INSTALLATION on a UNIX-ALIKE: * The order of selection of OpenMP flags has been changed: Oracle Developer Studio 12.5 accepts -fopenmp and -xopenmp but only the latter enables OpenMP so it is now tried first. BUG FIXES: * within(List, rm(x1, x2)) works correctly again, including when List[["x2"]] is NULL. * regexec(pattern, text, *) now applies as.character(.) to its first two arguments, as documented....

...bute >> for (int i = 0; i < size; ++i) { >> a[i] = a[i]*(i*i)/2; >> } >> return 0; >> >> PD: The command to compile is this: clang -v -save-temps >> -I/path/to/llvm/release/9.x/include -fopenmp >> -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_61 >> memory_transfer.c -o memory_transfer >> >> Thank you all. >> >> Hamilton. >> >> >> On 18/03/20 9:21 p. m., Doerfert, Johannes wrote: >>> Hi Hamilton, >>> >>> I personally don't believe the effort t...

Hi, I am trying to compile R 1.9.1 and 2.0.0 on Solaris 2.8 with Sun Studio Fortran 95 compiler but I have obtained some errors which I enclose. I have compiled R with g77 GNU compiler and it was OK. Does anyone know if R 2.0.0 has been successfully built on Solaris 2.8 with Sun Studio Fortran 95 compiler? or at least, How could I achieve this? The R installation manual does not have any

...gt; } > >>   > > >>   > > >>   > PD: The command to compile is this: > >>   >     clang -v -save-temps \ > >>   >     -I/path/to/llvm/release/9.x/include -fopenmp \ > >>   >     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target \ > >>   >         -march=sm_61 memory_transfer.c -o memory_transfer > >>   > > >>   > Thanks. > >>   > Hamilton. > >>   > > >>   > On 23/03/20 11:11 a. m., Doerfert, Johannes wrote: > >>   >&gt...

Hi Alexey, Is there any option for clang to turn on debug for the host code only but not the device code? I've been using something like -ggdb3 -O0 but this generate debug info for both host and device. I'm trying to work around the aforementioned ptxas bug. Thanks, Char At 2019-02-28 02:09:54, "Alexey Bataev" <a.bataev at outlook.com> wrote: Hi Char, it looks like

...ze]) >   #pragma omp teams distribute >   for (int i = 0; i < size; ++i) { >     a[i] = a[i]*(i*i)/2; >   } >   return 0; > > PD: The command to compile is this: clang -v -save-temps > -I/path/to/llvm/release/9.x/include -fopenmp > -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_61 > memory_transfer.c -o memory_transfer > > Thank you all. > > Hamilton. > > > On 18/03/20 9:21 p. m., Doerfert, Johannes wrote: >> Hi Hamilton, >> >> I personally don't believe the effort to make the IR >> "parallelism-aw...

...   > > >>   >>   > PD: The command to compile is this: > >>   >>   >     clang -v -save-temps \ > >>   >>   >     -I/path/to/llvm/release/9.x/include -fopenmp \ > >>   >>   >     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target \ > >>   >>   >         -march=sm_61 memory_transfer.c -o memory_transfer > >>   >>   > > >>   >>   > Thanks. > >>   >>   > Hamilton. > >>   >>   > > >>   >>   > On 23/03/2...

Hi Hamilton, I personally don't believe the effort to make the IR "parallelism-aware" is worth it right now. What I propose is something else, see for example https://youtu.be/zfiHaPaoQPc and the OpenMPOpt pass. There are multiple bigger tasks towards better offloading, one of which is described here https://github.com/llvm/llvm-project/issues/180 It might be interesting to add the

...this? The R installation manual does not have any reference. What is `Sun Studio'? The R-admin manual does have references to what I think is the same compiler, `Sun ONE Studio 7 Compiler Suite' (aka Forte 7) (possibly not the same version). Forte 9 has also been used successfully, if -xopenmp=stubs was added to LDFLAGS (and that is in the R-admin manual for the imminent R 2.0.1). > I want Sun Fortran in order to use the Sun performance library > libsunperf since I noticed that R on Windows XP is twice faster than R > on Sun Solaris 2.8 without library libsunperf. Yes, well,...