1. Embedded Computer Architecture 5SIA0 Overview
Henk Corporaal
TUEindhoven

2. ECA summary The miniMIPS processor some of you built
What you’ll understand after taking 5SIA0
Also, the technology behind chip-scale multiprocessors
Photo is of an Intel Conroe (Core 2 Duo Desktop processor). Photo taken from Intel web site.
ECA H.Corporaal

3. Course goals Learn advanced computer architecture concepts like:
ILP, DLP, and Multi-issue architectures
O-O-O execution
Correlating branch prediction;
Advanced memory hierarchy; speedup methods
Energy consumption and Technology issues;
etc.
Learn multi-processor architecture concepts like:
Multi-threading
Topologies
Synchronization
Cache Coherence and Memory Consistency
ECA H.Corporaal

4. Book Introduction Impact of technology Processor microarchitecture
Memory hierarchies
Multiprocessor systems
Interconnection networks
Coherence, synchronization, and memory consistency
Chip multiprocessors
Quantitative evaluations
We’ll add ‘embedded’, e.g. ARM
ECA H.Corporaal

5. Organization Credits: Weekly class meetings
5SIA0: 5 credit points (ECTS)
Weekly class meetings
Tuesdays: (Aud 7)
Thursdays: – (Aud 7)
Very advanced Labs: largely in your own time
Student literature research of TOP recent conferences
last week
Examination in end of January/early February 2017
ECA H.Corporaal

6. Practical Experience 3 lab assignments: Lab assistents:
Design and evaluation of a CGRA (Coarse Grain Reconfigurable Array) processor
Extreme parallel (GPU - SIMD) programming
Evaluation of a Multi-Core processor
Lab assistents:
Mark Wijtvliet
Luc Waeijen
Roel Jordans
Mohammad Tahghighi
ECA H.Corporaal

7. Lab 1: CGRA
Co-optimization of application and architecture for a coarse grained reconfigurable architecture
ECA H.Corporaal

8. Architecture and application
CGRA:
Application: Gaussian blur

9. Your job: Modify the architecture to: Modify the algorithm Enjoy!
Reduce energy
Reduce area
Improve performance
Modify the algorithm
Code optimizations
Modify code to use architecture configuration
Enjoy!

10. Lab 2: GPU - SIMD NVIDIA's PASCAL architecture
One SM: streaming multiprocessor
supports FP16, 32 & 64
ECA H.Corporaal

11. NVIDIA's PASCAL (5-20 Tflop)
ECA H.Corporaal

12. GPU trends (NVIDIA)
see:
ECA H.Corporaal

13. Lab 3: Multicore Assignment!

14. What are the objectives?
To get familiar with multiprocessor architectures and their programming models
To look at different configurations, e.g., the number of processors, block-size and associativity of different levels of caches.
Finally to optimize the Energy-Delay-Area-Product (EDAP) of the system

15. What you will learn
How to use the GEM5 as cycle accurate simulator to run applications
The impacts of different architectural parameters on performance
The size of different levels of caches
Cache Associativity
Applying loop transformation techniques to optimize the memory accesses
Applying the application partitioning technique for task level parallelism
Using McPAT for power and area estimation

16. Schedule 2016-2017 (preliminary)
Date
Topic
Material
Assignments + Remarks
15 Nov
Introduction
Ch 1
17 Nov
Technology Impact
Ch 2
22 Nov
CGRA and Accelerators: Mark Wijtvliet
CGRA assignment
(deadline: Dec 8)
24 Nov
Processor Architectures - 1
Ch 3
29 Nov
Processor Architectures - 2
1 Dec
GPU: Zhenyu Ye / Gert-Jan
GPU lab
(deadline: Jan 9)
6 Dec
Processor Architectures - 3
Ch 3; ARM
8 Dec
Processor Architectures - 4
13 Dec
Memory hierarchy
Ch 4
15 Dec
Loop transformation for Data Reuse
(extended in compiler course)
20 Dec
Multiprocessor systems
Ch 5
22 Dec
GEM5+ Simulation: Mohammad Tahghighi + Luc Waeijen
Ch 9
GEM5 multiproc. assign.
(deadline: Jan 22)
10 Jan
Interconnection networks
Ch 6
12 Jan
Coherence, synchr. and consistency
Ch 7
17 Jan
Deep Learing Neural Networks: Maurice Peemen
ECA H.Corporaal

17. Extra Material
Handouts and slides; see course web site:
Study recent articles from top conferences and journals
Alternative reading book:
Computer Architecture A quantitative approach
5th ed. by Hennessy and Patterson
ch 1-5, app A-C
ECA H.Corporaal

18. Where is computing going?
ECA H.Corporaal 18