CEN 316, 1252 - Computer Architecture
Homework #7
Name UTID
KHALID FARAH SULIMAN ALANAZI 421007347
Section 1: Multiple Choice Questions (10 Questions)
1. Which of the following best describes a SIMD architecture?
a) Multiple processors, each executing different instructions
b) One instruction stream controlling multiple data streams in lockstep
c) One instruction and one data stream
d) Each processor has its own private memory
2. What is the primary drawback of bus-based multiprocessor systems?
a) Lack of flexibility
b) High reliability
c) Performance limited by bus cycle time
d) Inability to expand with more processors
3. Which cache coherence protocol allows only one writer and multiple readers?
a) Write-update
b) Write-through
c) Write-invalidate
d) MESI-None
4. Which of the following MESI states indicates that the cache line is the same as memory and not shared with any
other cache?
a) Shared
b) Modified
c) Invalid
d) Exclusive
5. A group of interconnected whole computers working together as one unified system is called:
a) NUMA
b) SMP
c) Cluster
d) Grid
6. What is the main purpose of the snoopy protocol?
a) Broadcast messages between operating systems
b) Manage threads
c) Maintain cache coherence
d) Schedule I/O operations
7. What is a defining feature of Nonuniform Memory Access (NUMA)?
a) All processors access the same memory at the same speed
b) Access time depends on the memory region
c) It is exclusive to SIMD
d) It uses paging exclusively
8. Which system is generally easier to manage and configure for legacy software?
a) SMP
b) Cluster
c) NUMA
d) Grid
9. In MESI protocol, the state “Invalid” means:
a) The data is private to this cache
b) The data matches main memory
c) The data is not valid or usable
d) The data is modified
10. In a directory protocol, the coherence is:
a) Maintained by a central table tracking cache contents
b) Based on broadcasting updates
c) Ignored for read operations
d) Done only at compile time
b c c d c c b a c a
🔹 Section 2: True or False (10 Questions)
1. MIMD systems can execute multiple instruction streams on multiple data sets simultaneously.
2. Bus-based multiprocessor systems are completely immune to cache coherence issues.
3. Write-update and write-invalidate are the two basic approaches in snoopy protocols.
4. In the MESI protocol, “Modified” means the data is the same in cache and memory.
5. Clusters require specialized hardware to function as one system.
6. In NUMA systems, some memory access takes longer depending on the processor’s proximity.
7. Cache coherence can be enforced through both hardware and software approaches.
8. SMP systems do not require synchronization between processors.
9. Clustering provides better fault tolerance and high availability than SMP.
10. Cache consistency is only a concern in uniprocessor systems.
True False True False False True True False True False
🔹 Section 3: List & Compare (5 Questions)
1. List the four types of Flynn’s classification of processor organizations.
Single instruction, single data (SISD) stream
Multiple instruction, single data (MISD) stream
Single instruction, multiple data (SIMD) stream
Multiple instruction, multiple data (MIMD) stream.
2. List three design considerations for multiprocessor operating systems.
Simultaneous concurrent processes
Scheduling
Synchronization
3. Compare SMP and Clusters in terms of scalability, manageability, and availability.
Scalability: clusters is absolute scalability.
Manageability: clusters more complex than SMP.
Availability: clusters has high availability.
4. List the four MESI cache line states and what each state means.
Modified: The line in the cache has been modified and is available only in this cache
Exclusive: The line in the cache is the same as that in main memory and is not present in any other cache
Shared: The line in the cache is the same as that in main memory and may be present in another cache
Invalid: The line in the cache does not contain valid data
5. List two advantages and two disadvantages of NUMA compared to SMP.
Advantages :
Better scalability
faster in access to local memory
Disadvantages:
more complex
non-uniform access time
🔹 Section 4: Figure-Based Questions (5 Questions)
(Use Figures from the slides such as diagrams of SMP, MESI Table, NUMA, etc.)
6. Refer to the MESI state table figure. Describe how a cache line transitions from Shared to Modified.
Requires a write and must hit.
7. Using the figure showing cluster systems, explain what is meant by "failover" and "failback."
Failover: Automatically switching to a backup system upon failure.
Failback: Returning to the original system once it’s recovered.
8. From the NUMA diagram, explain why access latency differs for different processors.
Access latency varies because local memory access is faster than remote memory
9. Analyze the figure of a symmetric multiprocessor (SMP) organization. Explain why cache coherence becomes an
issue in this design.
Multiple processors have separate caches When they access the same memory data may become inconsistent
unless coherence protocols are used.
10. Using the cache coherence figure, describe the behavior of a snoopy protocol using a write-invalidate approach.
On a write, all other caches invalidate their copies of the cache line to ensure only one valid
