Computer Science & Engineering Questions & Answers
QUESTION BANK - MODULE 6
Explain various latency hiding techniques
OR
What do you mean by Release Consistency (RC) memory model? Give the
conditions to ensure RC. [4]
OR
1 [8]
Explain the distributed caching. [4]
OR
Explain any two latency hiding techniques used in distributed shared memory
multi computers [6]
ANSWER
Four latency hiding mechanisms are used to increase scalability and programmability.
1. Pre fetching techniques
2. Distributed Coherent caches
3. Relaxed memory consistency
4. Multiple context
• Prefetching technique:
• Which bring instruction or data close to the processor before they are actually
needed.
• Distributed coherent caches:
• Supported by hardware to reduce caches misses.
• Relaxed memory consistency models:
• By allowing buffering and pipelining of memory references.
• Multiple context support:
• To allow a processor to switch from one context to another
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 1
, Computer Science & Engineering Questions & Answers
1. PRE FETCHING TECHNIQUES
SOFTWARE CONTROLLED PRE –FETCHING
• In this approach, explicit "pre-fetch" instructions are issued for data that is "known" to be
remote.
• Pre-fetching is done selectively
• Disadvantages
• Extra instruction overhead
• Need of sophisticated software intervention
HARDWARE CONTROLLED PRE–FETCHING
• The hardware pre-fetches at run time.
• It provide better dynamic information
• No instruction overhead to issue pre-fetches
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 2
, Computer Science & Engineering Questions & Answers
• Difficult to access memory access pattern
2. DISTRIBUTED COHERENT CACHES
• Cache coherence is maintained using snoopy bus protocols for bus based system
• Coherence is maintained using directory based protocol for n/w based system
Benefits of caching
• Improves the performance
• Reduces the no: of cycles wasted due to read misses
• Cycles wasted due to write misses are also reduced
3. RELAXED MEMORY CONSISTENCY
• Memory consistency model is a set of rules that govern how memory system will process
memory operation ( Read/Write) from multiple processors
• Consistency model are used in distributed system like distributed shared memory system
or distributed data stores
Following are the relaxed memory consistency model
• Processor consistency ( PC)
• Release consistency (RC)
PROCESSOR CONSISTENCY ( PC)
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 3
QUESTION BANK - MODULE 6
Explain various latency hiding techniques
OR
What do you mean by Release Consistency (RC) memory model? Give the
conditions to ensure RC. [4]
OR
1 [8]
Explain the distributed caching. [4]
OR
Explain any two latency hiding techniques used in distributed shared memory
multi computers [6]
ANSWER
Four latency hiding mechanisms are used to increase scalability and programmability.
1. Pre fetching techniques
2. Distributed Coherent caches
3. Relaxed memory consistency
4. Multiple context
• Prefetching technique:
• Which bring instruction or data close to the processor before they are actually
needed.
• Distributed coherent caches:
• Supported by hardware to reduce caches misses.
• Relaxed memory consistency models:
• By allowing buffering and pipelining of memory references.
• Multiple context support:
• To allow a processor to switch from one context to another
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 1
, Computer Science & Engineering Questions & Answers
1. PRE FETCHING TECHNIQUES
SOFTWARE CONTROLLED PRE –FETCHING
• In this approach, explicit "pre-fetch" instructions are issued for data that is "known" to be
remote.
• Pre-fetching is done selectively
• Disadvantages
• Extra instruction overhead
• Need of sophisticated software intervention
HARDWARE CONTROLLED PRE–FETCHING
• The hardware pre-fetches at run time.
• It provide better dynamic information
• No instruction overhead to issue pre-fetches
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 2
, Computer Science & Engineering Questions & Answers
• Difficult to access memory access pattern
2. DISTRIBUTED COHERENT CACHES
• Cache coherence is maintained using snoopy bus protocols for bus based system
• Coherence is maintained using directory based protocol for n/w based system
Benefits of caching
• Improves the performance
• Reduces the no: of cycles wasted due to read misses
• Cycles wasted due to write misses are also reduced
3. RELAXED MEMORY CONSISTENCY
• Memory consistency model is a set of rules that govern how memory system will process
memory operation ( Read/Write) from multiple processors
• Consistency model are used in distributed system like distributed shared memory system
or distributed data stores
Following are the relaxed memory consistency model
• Processor consistency ( PC)
• Release consistency (RC)
PROCESSOR CONSISTENCY ( PC)
KTU-S7-CS405-Computer System Architecture [ MODULE-6] 3
