I am trying to build mapreduce (http://sourceforge.net/projects/mapreduce-cell). I am cross-compiling from Fedora 7 on an x86 host for the Cell simulator (SDK 3.0). I had to change a couple of the makefiles to point to the right location of make.footer. After that, I ran make, and got this from make.footer:649
Error: No spuxlc found.
I ran
locate spuxlc
which returned nothing. Does spuxlc come with SDK 3.0? How do I install spuxlc?
Thanks,
Scott
Topic
-
Re: No spuxlc found when building mapReduce
2008-06-03T14:57:12Z
This is the accepted answer.
This is the accepted answer.
XLC is now a licensed product and no longer provided for free in the SDK.
See the page XL C/C++ for Multicore Acceleration for Linux for more information about XLC and how to obtain a trial version. Until then you can use GCC (unless that code requires XLC).
--
IBM SDK Service Administrator -
Re: No spuxlc found when building mapReduce
2008-06-05T18:08:14Z
This is the accepted answer.
This is the accepted answer.
Thanks - I changed the spu compiler to gcc in the top level makefile. There were still a bunch of errors during the build. They all seemed to be coming from compilation of the benchmarks (spu code), where some spu intrinsics were not getting recognized. I added "#include <spu_intrinsics.h>" to the *_spu.c source for each to the problematic benchmarks and the build completed.- CellServ
- 2008-06-03T14:57:12Z
XLC is now a licensed product and no longer provided for free in the SDK.
See the page XL C/C++ for Multicore Acceleration for Linux for more information about XLC and how to obtain a trial version. Until then you can use GCC (unless that code requires XLC).
--
IBM SDK Service Administrator -
Re: No spuxlc found when building mapReduce
2010-06-22T15:49:10Z
This is the accepted answer.
This is the accepted answer.
Hi there,- SystemAdmin
- 2008-06-05T18:08:14Z
Thanks - I changed the spu compiler to gcc in the top level makefile. There were still a bunch of errors during the build. They all seemed to be coming from compilation of the benchmarks (spu code), where some spu intrinsics were not getting recognized. I added "#include <spu_intrinsics.h>" to the *_spu.c source for each to the problematic benchmarks and the build completed.
I just built the MapReduce over Cell! I'm doing a Msc thesis on MapReduce over Cell. How was the utility for you of this system? Could you please tell me if it was easy for you to apply new problems?
thank you very much!!!
best regards
