WCCCE Conference Presentation

1. Teaching Computer Architecture
Labs using a MCU Platform
Rafael Roman Otero and Alex Aravind
University of Northern British Columbia
(UNBC)

2. Focus of this paper

3. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science

4. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
Organization?

5. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
Organization?

6. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
Computer Organization and Architecture
(CPSC 231 at UNBC)
Organization?

7. Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
Focus of this paper
• Instruction Set Architecture
• Addressing modes
• Interrupts
• Pipelining
Computer Organization and Architecture
(CPSC 231 at UNBC)

8. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
• Instruction Set Architecture
• Addressing modes
• Interrupts
• Pipelining
Computer Organization and Architecture
(CPSC 231 at UNBC)
Lecture
(Conceptual Explanation and Internals)

9. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
(Concepts Demonstrated with Assembly Exercises)
• Instruction Set Architecture
• Addressing modes
• Interrupts
• Pipelining
Computer Organization and Architecture
(CPSC 231 at UNBC)
Lecture + Labs
(Conceptual Explanation and Internals)

10. Focus of this paper
Teaching Computer Architecture Labs to Undergraduate Students in
Computer Science
(Concepts Demonstrated with Assembly Exercises)
Computer Organization and Architecture
(CPSC 231 at UNBC)
Lecture + Labs
(Conceptual Explanation and Internals)
• Instruction Set Architecture
• Addressing modes
• Interrupts
• Pipelining

11. The problem

12. The problem
Architecture Complexity
Low Student Engagement

13. The problem
Architecture Complexity (simulators)
Low Student Engagement

14. The problem
Architecture Complexity (simulators)
Low Student Engagement (a “fun” factor)

15. The problem
Architecture Complexity (simulators)
Low Student Engagement (a “fun” factor)
“Courses in computer architecture and organization […] involve intricate details
of the implementation and operation of the computer. […] This paper proposes a
simulator called EASE.
--Skillen et al

16. The problem
Architecture Complexity (simulators)
Low Student Engagement (a “fun” factor)
“The description [...] of internal computer operation is not an easy task. The
processes involved are sophisticated. […] visualization is one tool to deal with
the emerging complexity“
-- Yehezkel et al
(They propose EasyCPU)

17. The problem
Architecture Complexity (simulators)
Low Student Engagement (a “fun” factor)
“motivating students in an undergraduate assembly language course is a tremendous
challenge […] Our findings […] indicate that the […] Raspberry Pi […] contributed
positively to their learning outcomes
-- Kawash et al
“The description [...] of internal computer operation is not an easy task. The
processes involved are sophisticated. […] visualization is one tool to deal with
the emerging complexity“
-- Yehezkel et al

18. The problem
Architecture Complexity (simulators)
Low Student Engagement (a “fun” factor)
“SPIMbot [ a virtual robot simulator ] was written to provide an engaging environment
to motivate students to learn assembly language concepts”
--Zilles et al
“The description [...] of internal computer operation is not an easy task. The
processes involved are sophisticated. […] visualization is one tool to deal with
the emerging complexity“
-- Yehezkel et al

19. The problem
“motivating students in an undergraduate assembly language course is a tremendous challenge […] Our
findings […] indicate that the […] Raspberry Pi […] contributed positively to their learning outcomes
-- Kawash et al
“SPIMbot [ a virtual robot simulator ] was written to provide an engaging environment to motivate
students to learn assembly language concepts”
--Zilles et al
“The description [...] of internal computer operation is not an easy task. The processes involved
are sophisticated. […] visualization is one tool to deal with the emerging complexity“
-- Yehezkel et al
“Courses in computer architecture and organization […] involve intricate details of the
implementation and operation of the computer. […] This paper proposes a simulator called EASE.
--Skillen

20. The problem
“motivating students in an undergraduate assembly language course is a tremendous challenge […] Our
findings […] indicate that the […] Raspberry Pi […] contributed positively to their learning outcomes
-- Kawash et al
“SPIMbot [ a virtual robot simulator ] was written to provide an engaging environment to motivate
students to learn assembly language concepts”
--Zilles et al
“The description [...] of internal computer operation is not an easy task. The processes involved
are sophisticated. […] visualization is one tool to deal with the emerging complexity“
-- Yehezkel et al
“Courses in computer architecture and organization […] involve intricate details of the
implementation and operation of the computer. […] This paper proposes a simulator called EASE.
--Skillen

21. Simulators

22. Simulators
bochs was used to teach CPSC 231 in the past at UNBC

23. Simulators
bochs was used to teach CPSC 231 in the past at UNBC
bochs is an emulator, not a simulator—bear with me!

24. Simulators
bochs was used to teach CPSC 231 in the past at UNBC
bochs is an emulator, not a simulator—bear with me!

25. Simulators
Positive:
Negative:
• .

26. Simulators
Positive:
• Students find it interesting
Negative:
• .

27. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
Negative:
• .

28. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• .

29. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• .

30. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• No debugger (too difficult to set up)
• .

31. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• No debugger (too difficult to set up)
• Almost a blackbox
.

32. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• No debugger (too difficult to set up)
• Almost a blackbox
• Only means of visualization was a rudimentary print function
.

33. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• No debugger (too difficult to set up)
• Almost a blackbox
• Only means of visualization was a rudimentary print function
• Only means of visualization was a buggy print function

34. Simulators
Positive:
• Students find it interesting
• It’s a real architecture
• Students get to say that they’ve had some experience with x86
Negative:
• x86 bare-metal is hard!
• No debugger (too difficult to set up)
• Almost a blackbox
• Only means of visualization was a rudimentary print function
• Only means of visualization was a buggy print function
Simulators makes sense.
(It’s easy to see why they’re so widely accepted as “the way”
of teaching architecture)

35. Simulators
“Hardware simulators are used to eliminate the burden of working on
a bare machine […] can provide sophisticated development, user
interface, testing, and debugging environments unavailable when
working on bare machine”
---Goldweber et al

36. Simulators
“Hardware simulators are used to eliminate the burden of working on
a bare machine […] can provide sophisticated development, user
interface, testing, and debugging environments unavailable when
working on bare machine”
---Goldweber et al

37. Simulators
“Hardware simulators are used to eliminate the burden of working on
a bare machine […] can provide sophisticated development, user
interface, testing, and debugging environments unavailable when
working on bare machine”
---Goldweber et al
Simulators make sense!

38. Simulators
There exists an alternative solution
(similar benefits, but not a simulator)

39. Solution
The reign of the simulator is over
All hail the MCU platform!

40. Solution
The reign of the simulator is over
All hail the MCU platform!
SAM4S MCU (32-bit ARM
Cortex-M4 core):

41. Solution
The reign of the simulator is over
All hail the MCU platform!
SAM4S MCU (32-bit ARM
Cortex-M4 core):
• Memory Protection
• Cache
• Atomic Loads/Restore
• Software Ints
• User mode and Kernel
mode
• Privileged/Unprivileged
Execution

42. MCU Advantages

43. MCU Advantages
Pros:
• Off-the-shelf debugging!

44. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)

45. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)
• Pause and Resume the CPU
• Inspect and modify: memory, registers, IO interfaces.

46. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)
• Pause and Resume the CPU
• Inspect and modify: memory, registers, IO interfaces.
• Simpler architecture

47. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)
• Pause and Resume the CPU
• Inspect and modify: memory, registers, IO interfaces.
• Simpler architecture
• e.g. Switching to user mode in x86 vs in ARM Cortex-M4

48. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)
• Pause and Resume the CPU
• Inspect and modify: memory, registers, IO interfaces.
• Simpler architecture
• e.g. Switching to user mode in x86 vs in ARM Cortex-M4
• Reduces effort and time to, say, switch to user mode, or set up the MPU

49. MCU Advantages
Pros:
• Off-the-shelf debugging!
• The debugger is visual and interactive (like a simulator)
• Pause and Resume the CPU
• Inspect and modify: memory, registers, IO interfaces.
• Simpler architecture
• e.g. Switching to user mode in x86 vs in ARM Cortex-M4
• Reduces effort and time to, say, switch to user mode, or set up the MPU
• Goes well with undergrad introductory courses (exposure, not expertise)

50. MCU Advantages
Pros:
• It’s good for student engagement

51. MCU Advantages
Pros:
• It’s good for student engagement
• Students get to play with sensors and actuators
• Their code gets to execute on real hardware
“fun” factor

52. Solution Part two

53. Solution Part two
Guzdial argues that the amount of instruction matters when teaching
computer science to beginners:

54. Solution Part two
Guzdial argues that the amount of instruction matters when teaching
computer science to beginners:
“putting introductory students in the position of discovering
information for themselves is a bad idea”

55. Solution Part two
Guzdial argues that the amount of instruction matters when teaching
computer science to beginners:
“putting introductory students in the position of discovering
information for themselves is a bad idea”
“insofar as there is any evidence from controlled studies, it almost
uniformly supports direct, strong instructional guidance”

56. Solution Part two
An in-detail tutorial book

57. Solution Part two
An in-detail tutorial book
• Demonstrate Architecture
concepts

58. Solution Part two
An in-detail tutorial book
• Demonstrate Architecture
concepts
• Cover all technical details
necessary for lab
completion

59. Solution Part two
An in-detail tutorial book
• Demonstrate Architecture
concepts
• Cover all technical details
necessary for lab
completion
1. Introduction
2. ISA
3. Memory
4. IO
5. Stack
6. Interrupts

60. Solution Part two
An in-detail tutorial book

61. Solution Part two
An in-detail tutorial book

62. Solution Part two
An in-detail tutorial book

63. Solution Part two
An in-detail tutorial book

64. Solution Part two
An in-detail tutorial book

65. Solution Part two
An in-detail tutorial book

66. Solution Part two
An in-detail tutorial book

67. Solution Part two
An in-detail tutorial book

68. Solution Part two
An in-detail tutorial book

69. Solution Part two
An in-detail tutorial book

70. Solution Part two
An in-detail tutorial book
If the guide offers too many
details or too many steps, what
is left for the student to do?
More advanced things.

71. Solution Part two
An in-detail tutorial book
If the guide offers too many
details or too many steps, what
is left for the student to do?
More advanced things.
E.g.: Loader in Lab 3
(in previous years it was difficult enough to be a final project)

72. Evaluation
2017 Winter Semester CPSC 231
Final Projects

73. Evaluation
2017 Winter Semester CPSC 231
Final Projects

74. Evaluation
2017 Winter Semester CPSC 231
Maze solver
Final Projects

75. Evaluation
2017 Winter Semester CPSC 231
Working knowledge to pursue something of their own

76. Evaluation
2017 Winter Semester CPSC 231
Working knowledge to pursue something of their own
2015

77. Evaluation
2017 Winter Semester CPSC 231
Working knowledge to pursue something of their own
2015

78. Evaluation
2017 Winter Semester CPSC 231
16 students total
14 participate in the survey,
1 student dropped the class.

79. Evaluation
2017 Winter Semester CPSC 231
16 students total
14 participate in the survey,
1 student dropped the class.
Lacks any statistical significance. We use it to get
an idea of how students perceived the labs

80. Evaluation
2017 Winter Semester CPSC 231
16 students total
14 participate in the survey,
1 student dropped the class.
1=‘Strongly disagree’, 2=‘Disagree’, 3=‘Neutral’, 4=‘Agree’, and 5=‘Strongly agree’.
16 students total
14 participate in the survey,
1 student dropped the class.

81. Evaluation
2017 Winter Semester CPSC 231
16 students total
14 participate in the survey,
1 student dropped the class.
1=‘Strongly disagree’, 2=‘Disagree’, 3=‘Neutral’, 4=‘Agree’, and 5=‘Strongly agree’.

82. Evaluation
2017 Winter Semester CPSC 231
16 students total
14 participate in the survey,
1 student dropped the class.
1=‘Strongly disagree’, 2=‘Disagree’, 3=‘Neutral’, 4=‘Agree’, and 5=‘Strongly agree’.

83. Disadvantages

84. Disadvantages
• Students will not be using MCUs in their workplace. They’ll be using
desktop architectures.

85. Disadvantages
• Students will not be using MCUs in their workplace. They’ll be using
desktop architectures.
• MCUs have no MMU

86. Disadvantages
• Students will not be using MCUs in their workplace. They’ll be using
desktop architectures.
• MCUs have no MMU
• MCUs are still complex!

87. Disadvantages
• Students will not be using MCUs in their workplace. They’ll be using
desktop architectures
• MCUs have no MMU
• MCUs are still complex!
• MCUs have a Harvard architecture. We need to teach Von
Neumman! (Someone please think of the children!!)

88. That’s it!
Thank You