Message boards :
Number crunching :
platform 'powerpc-pc-linux-gnu' application for my PS3 ?
Message board moderation
Author | Message |
---|---|
Send message Joined: 2 Sep 04 Posts: 12 Credit: 1,363 RAC: 0 |
Hi all, finally I have succeeded in installing Fedora 9 ppc on my PS3, and installed BOINC with yum ... but all I get from the projects in which I am participating is that they don't have a application for my platform 'powerpc-pc-linux-gnu' :( Can someone help me to get a platform 'powerpc-pc-linux-gnu' application for this project ? I know that such a application would only use the PPC core (with Altivec) of my CELL CPU, but better 1 of 7 cores than 0 of 7 ! If someone can tell me how to compile a platform 'powerpc-pc-linux-gnu' for my PS3 myself, I could try this too, as in the long term I intend to optimize the sources for my PS3. MfG, MEX 0xFF |
Send message Joined: 18 Sep 04 Posts: 163 Credit: 1,682,370 RAC: 0 |
Hi all, Hi MEX, First, there is no PPC science-application for LHC and there probably will never be one (see other threads). Second, BOINC is available for PS3 and PPC in general, see e.g ps3grid and SETI (3rd party) HTH Michael Team Linux Users Everywhere |
Send message Joined: 2 Sep 04 Posts: 12 Credit: 1,363 RAC: 0 |
First, there is no PPC science-application for LHC and there probably will never be one (see other threads).Why ? Does LHC@home need good floating point performace ? MfG, MEX 0xFF |
Send message Joined: 7 Aug 05 Posts: 60 Credit: 76,331 RAC: 99 |
First, there is no PPC science-application for LHC and there probably will never be one (see other threads).Why ? Same as on the other projects, too. The application must complete rewritten to get the full performance of the PS3 SPEs. Rewriting the source is a lot of work... Also other problem, LHC uses a special Fortran compiler which is only available at Windows and Linux. Other compilers produce different resutls, but LHC needs identical results. |
Send message Joined: 14 Jul 05 Posts: 275 Credit: 49,291 RAC: 0 |
Does LHC@home need good floating point performace ? The LHC@home app doesn\'t even support Mac, because there isn\'t any good Fortran compiler for OS X that can cope with sixtrack. The PS3 does have good floating point performance. But it\'s way more than just a PPC. It\'s quite a different problem to get an application running fast on a PS3. The only projects that will support PS3 are those that are written from the beginning with Cell in mind. |
Send message Joined: 2 Sep 04 Posts: 12 Credit: 1,363 RAC: 0 |
Rewriting the source isn\'t what i am asking for !Same as on the other projects, too. The application must complete rewritten to get the full performance of the PS3 SPEs. Rewriting the source is a lot of work...First, there is no PPC science-application for LHC and there probably will never be one (see other threads).Does LHC@home need good floating point performace ? And as I have installed Fedora 9 on my PS3, it should be possible to install a Cross-compiler on your computer for your Fortran to produce PPC ( without SPE ! ) code for all PPC Linux platforms. MfG, MEX 0xFF |
Send message Joined: 7 Aug 05 Posts: 60 Credit: 76,331 RAC: 99 |
Rewriting the source isn\\\'t what i am asking for !Same as on the other projects, too. The application must complete rewritten to get the full performance of the PS3 SPEs. Rewriting the source is a lot of work...First, there is no PPC science-application for LHC and there probably will never be one (see other threads).Does LHC@home need good floating point performace ? No. It it not posible to install a cross compiler. There is no crosscompiler for the compiler which CERN has used available for PPC or the PS3. The compiler which has CERN used is only available for Windows and Linux on x86. Without coderewriting the PS3 performs comparable as a G4/G5 Mac. |
Send message Joined: 7 Aug 05 Posts: 60 Credit: 76,331 RAC: 99 |
Rewriting the source isn\\\\\\\'t what i am asking for !Same as on the other projects, too. The application must complete rewritten to get the full performance of the PS3 SPEs. Rewriting the source is a lot of work...First, there is no PPC science-application for LHC and there probably will never be one (see other threads).Does LHC@home need good floating point performace ? |
©2024 CERN