csee4824_f11_lec17

csee4824_f11_lec17 - CSEE W4824 Computer Architecture Fall...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon
1 CSEE W4824 – Computer Architecture Fall 2011 Luca Carloni Department of Computer Science Columbia University in the City of New York http://www.cs.columbia.edu/~cs4824/ Lecture 17 The Rise of Multi-Core Architectures CSEE4824 – Fall 2011 - Lecture 17 Page 2 Luca Carloni – Columbia University Announcements • Midterm Results 29.8 20.0 13.0 6.3 Std. Dev. 200 75 75 50 Highest Grade 200 75 75 50 Base Points 141.2 38.4 58.9 43.9 Average Total Problem 3 Problem 2 Problem 1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
2 CSEE4824 – Fall 2011 - Lecture 17 Page 3 Luca Carloni – Columbia University Announcements: The Road to the Final 360 Total 29.8 141.2 200 Midterm 22.2 62.2 80 HW3 10.3 34.4 45 HW2 4.5 31.8 35 HW1 Std. Dev. Average Base Points more than 60% of the course points are still unassigned! 45 HW4 65 HW5 350 Final 640 Total 180 Project Std. Dev. Average Base Points CSEE4824 – Fall 2011 - Lecture 17 Page 4 Luca Carloni – Columbia University Announcements: Pre-Taped Class Taught by Dr. Nicola Concer Lecture #19 Regular Class Wednesday 11/16 Pre-taped this Fri 11/11 at 12:30pm in Mudd 535 Lecture #18 (Pre-taped) Monday 11/14 • Reason: Instructor is traveling to attend two research workshops • The Pre-taped lecture will be shown as videos from the class PC during regular class time in Mudd 833 • Instructor’s office hours are canceled for the week of November 11 • Homework #4 Deadline Extension: until Wed 11/16
Background image of page 2
3 CSEE4824 – Fall 2011 - Lecture 17 Page 5 Luca Carloni – Columbia University Review: Multithreading cycles Single-Threaded Single-Threaded Blocked Multithreaded Interleaved Multithreaded Simultaneous Multithreaded • Simultaneous multithreading exploits both ILP and TLP, thus potentially recovering both horizontal and vertical waste CSEE4824 – Fall 2011 - Lecture 17 Page 6 Luca Carloni – Columbia University Pentium® 4 with Hyperthreading: NetBurst Pipeline Executing The Two Threads • Static Partitioning of key resources between threads – tread-off fairness vs. additional thread-level parallelism – reduce design complexity while ensuring fairness – prevent slow threads from blocking (or slowing down) faster thread • all pipe queues are partitioned • front-end stall (e.g. I-cache miss) doesn’t block pipeline back-end • back-end stall (e.g. D-cache miss) doesn’t block pipeline front-end Source: Koufaty2003]
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4 CSEE4824 – Fall 2011 - Lecture 17 Page 7 Luca Carloni – Columbia University Pentium® 4 with Hyperthreading: Performance Improvements • Two types of speed-up – multithreaded applications • each logical processor runs SW threads from same application – multiple single-thread application (multitasking) • each logical processor runs SW threads from different applications [Source: Koufaty2003] CSEE4824 – Fall 2011 - Lecture 17 Page 8 Luca Carloni – Columbia University Review: Measure: AMD Opteron vs. Pentium 4 • A 2.6 GHz AMD Opteron has lower CPI by a factor of 1.27 versus a 3.3Ghz Pentium 4 – this should be expected given the much deeper pipeline of Pentium 4 • Comparing the two processors at their highest available clock rate in 2005 – on average a 2.8GHz Opteron is slightly faster than a 3.8GHz Pentium4
Background image of page 4
Image of page 5
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 24

csee4824_f11_lec17 - CSEE W4824 Computer Architecture Fall...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online