Embedded Supercomputing

Embedded Supercomputing[1] (EmbSup) a relatively new solution which targets fine grain and coarse grain parallelism altogether. This combination thought to be a best way for exploiting fine and coarse grain parallelism by targeting fine grain parallelism towards FPGAs and coarse grained parallelism towards super computers or clusters.

Basically Embedded Supercomputing is a hybrid network of CPU and FPGA hardware, where FPGA acts as external co-processor to CPU. However, this programming model is still evolving and has many challenges.

Programming Model for EmbSup

gollark: Consider a silicon fab, which is used to make computer chips we need. That requires billions of $ in capital and thousands of people and probably millions more in supply chains.

gollark: Also, what do you mean "so what"? Technological progress directly affects standards of living.

gollark: ... that makes no sense that wouldn't even work.

gollark: Dunbar's number is 150 or so - humans can have meaningful social relationships with 150 or so people, apparently. Many systems require larger-scale coordination than this.

gollark: ... so we can have technology?

References

Deconinck, Geert; De Florio, Vincenzo; A. Varvarigou, Theodora; AVerentziotis, Evangelos (March 2002). "The EFTOS Approach to Dependability in Embedded Supercomputing". IEEE Transactions on Reliability. 51: 76–90. doi:10.1109/24.994916.

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.

[1] Deconinck, Geert; De Florio, Vincenzo; A. Varvarigou, Theodora; AVerentziotis, Evangelos (March 2002). "The EFTOS Approach to Dependability in Embedded Supercomputing". IEEE Transactions on Reliability. 51: 76–90. doi:10.1109/24.994916.