vortal
15-03-2007, 21:12
Para aqueles que planeiam usar a sua PS3 para programar o Cell e fazer aplicações gráficas interessantes, esta promete ser uma evolução prometedora. Parece ser possível obter aceleração gráfica usando os SPEs.
In a grassroots fashion, the hack-a-thon gave birth to a unique international community whose members come from academia, DOE labs, and several industries. The overwhelming response from folks was that they completed in one week what would have taken them 3 months to do on their own.
There was expert instruction and impromptu tutelage from industry experts at IBM, Mercury, and RapidMind.
No fewer than 4 development toolkits were invoked: RapidMind, Indiana University's CorePy, Mercury's MCF, IBM's SDK.
Local work began with (and in some cases extended prior efforts) several bioinformatics algorithms (BLAST, HMMER, ClustalW, Smith-Waterman), while other consortium members across the globe began remote development on molecular dynamics code (Gromacs).
Mesa was re-engineered to utilize Cell BE SPEs, resulting in astonishing 80x acceleration.
Povray and VTK porting were a success.
A special purpose kernel built with kprobes and SystemTap support was successfully built by the LANL folks.
A Cell BE compiler was created for Windows.
The configuration of several developmental tools, such as Eclipse, remote desktop access, and cross compiling environments were completed.
Professors and graduate students from CSU and UC, Boulder participated.
Subversion code repositories have been established for participants' projects.
All project progress and direction continue to be documented on the Consortium wiki.
http://www.hpc-consortium.net/events/hack-a-thon/200701/summary.shtml
In a grassroots fashion, the hack-a-thon gave birth to a unique international community whose members come from academia, DOE labs, and several industries. The overwhelming response from folks was that they completed in one week what would have taken them 3 months to do on their own.
There was expert instruction and impromptu tutelage from industry experts at IBM, Mercury, and RapidMind.
No fewer than 4 development toolkits were invoked: RapidMind, Indiana University's CorePy, Mercury's MCF, IBM's SDK.
Local work began with (and in some cases extended prior efforts) several bioinformatics algorithms (BLAST, HMMER, ClustalW, Smith-Waterman), while other consortium members across the globe began remote development on molecular dynamics code (Gromacs).
Mesa was re-engineered to utilize Cell BE SPEs, resulting in astonishing 80x acceleration.
Povray and VTK porting were a success.
A special purpose kernel built with kprobes and SystemTap support was successfully built by the LANL folks.
A Cell BE compiler was created for Windows.
The configuration of several developmental tools, such as Eclipse, remote desktop access, and cross compiling environments were completed.
Professors and graduate students from CSU and UC, Boulder participated.
Subversion code repositories have been established for participants' projects.
All project progress and direction continue to be documented on the Consortium wiki.
http://www.hpc-consortium.net/events/hack-a-thon/200701/summary.shtml