Tera-scale Computing

  Section 6 of 9  

Media Mining—Emerging Tera-scale Computing Applications

CONCLUSION

Media mining can help us retrieve, organize, and manage the exponentially growing media data easily. We explored several usage models in media mining and showed that most applications require tera-scale computing. To efficiently use the processing power provided by multi-core processors, we studied common parallelization schemes and proposed a general parallel framework for these media-mining applications. Furthermore, we conducted a performance analysis of several representative media-mining workloads on an 8-core system and a 32-core CMP simulator. Our analysis shows they have no obvious limits on parallelism. With a proper parallelization scheme, future large-scale CMP systems can deliver real-time performance for these media applications. Taking advantage of next-generation tera-scale computing platforms, new usage models in media mining will be enabled.

  Section 6 of 9  

Back to Top