Computer Architecture: A quantitative approach - Cap4 - Section 1

1,703 views
1,615 views

Published on

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,703
On SlideShare
0
From Embeds
0
Number of Embeds
156
Actions
Shares
0
Downloads
128
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Computer Architecture: A quantitative approach - Cap4 - Section 1

  1. 1. Multiprocessors and Thread-Level Parallelism Introduction “ The blueprints for our UltraSPARC T2, the core design files and test suites, will be available to the open source community, via its most popular license: the GPL.” Sun President Jonathan Schwartz (2007)‏
  2. 2. Multiprocessing Reliance Factors <ul><li>A growing interest in servers and server performance </li></ul><ul><li>A growth in data-intensive applications </li></ul><ul><li>The insight that increasing performance on the desktop is less important (outside of graphics, at least)‏ </li></ul><ul><li>An improved understanding of how to use multiprocessors effectively, especially in server environments where there is significant natural thread-level parallelism </li></ul><ul><li>The advantages of leveraging a design investment by replication rather than unique design — all multiprocessor designs provide such leverage </li></ul>
  3. 3. Taxonomy of Parallel Architectures <ul><li>Four Categories ( Flynn[1966] ): </li></ul><ul><li>SISD: Single Instruction Stream, Single Data Stream </li></ul><ul><li>SIMD: Single Instruction Stream, Multiple Data Stream </li></ul><ul><li>MISD: Multiple Instruction Stream, Single Data Stream </li></ul><ul><li>MIMD: Multiple Instruction Stream, Multiple Data Stream </li></ul>
  4. 4. Why Multiple Instruction, Multiple Data <ul><li>MIMDs offer flexibility. MIMDs can function as single-user multiprocessors, as multiprocessors running many tasks simultaneously, or as some combination of these functions. </li></ul><ul><li>MIMDs can build on the cost-performance advantages of off-the-shelf processors. Multicore chips leverage the design investment in a single processor core by replicating it. </li></ul>
  5. 5. Centralized Shared-memory Multiprocessor
  6. 6. Distributed-memory Multiprocessor
  7. 7. Communication and Memory Architecture <ul><li>Distributed Shared-Memory (DSM)‏ </li></ul><ul><li>Message-Passing Multiprocessors </li></ul>
  8. 8. Challenges of Parallel Processing <ul><li>Limited parallelism available in programs </li></ul><ul><li>Large latency of remote access in a </li></ul><ul><li>parallel processor </li></ul>
  9. 9. Thank you! Author: Prof. Sergio Takeo, Marcelo Arbore. Bibliography: Patterson, D. A.; Hennessy, J. L. Computer Architecture: A Quantitative Approach, 4 th Ed. Morgan Kaufmann Publishers. “ The blueprints for our UltraSPARC T2, the core design files and test suites, will be available to the open source community, via its most popular license: the GPL.” Sun President Jonathan Schwartz (2007)‏

×