A bandwidth-efficient architecture for media processing

Media applications are characterized by large amounts of available parallelism, little data reuse, and a high computation to memory access ratio. While these characteristics are poorly matched to conventional microprocessor architectures, they are a good fit for modern VLSI technology with its high...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Proceedings of the annual International Symposium on Microarchitecture s. 3 - 13
Hlavní autoři: Rixner, S., Dally, W.J., Kapasi, U.J., Khailany, B., Lopez-Lagunas, A., Mattson, P.R., Owens, J.D.
Médium: Konferenční příspěvek Journal Article
Jazyk:angličtina
Vydáno: IEEE 1998
Témata:
Arithmetic
Bandwidth
Computer architecture
Concurrent computing
Kernel
Microprocessors
Parallel processing
Parallel programming
Streaming media
Very large scale integration
ISBN:9780818686092, 081868609X
ISSN:1072-4451
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Media applications are characterized by large amounts of available parallelism, little data reuse, and a high computation to memory access ratio. While these characteristics are poorly matched to conventional microprocessor architectures, they are a good fit for modern VLSI technology with its high arithmetic capacity but limited global bandwidth. The stream programming model, in which an application is coded as streams of data records passing through computation kernels, exposes both parallelism and locality in media applications that can be exploited by VLSI architectures. The Imagine architecture supports the stream programming model by providing a bandwidth hierarchy tailored to the demands of media applications. Compared to a conventional scalar processor. Imagine reduces the global register and memory bandwidth required by typical applications by factors of 13 and 21 respectively. This bandwidth efficiency enables a single chip Imagine processor to achieve a peak performance of 16.2GFLOPS (single-precision floating point) and sustained performance of up to 8.5GFLOPS on media processing kernels.
Bibliografie:SourceType-Scholarly Journals-2
ObjectType-Feature-2
ObjectType-Conference Paper-1
content type line 23
SourceType-Conference Papers & Proceedings-1
ObjectType-Article-3
ISBN:9780818686092
081868609X
ISSN:1072-4451
DOI:10.1109/MICRO.1998.742118