Podcast
Questions and Answers
What is the primary difference between Unified and Harvard Memory Architecture?
What is the primary difference between Unified and Harvard Memory Architecture?
- Memory size for Instruction and Data
- Memory location for Instruction and Data (correct)
- Memory access time
- Memory allocation for Instruction and Data
What is the formula for MemStall in Unified Memory Architecture?
What is the formula for MemStall in Unified Memory Architecture?
- (MIss Rate)(Miss Penalty) (correct)
- (Miss Rate)(Miss Penalty) + (Fraction of Instruction)(Fraction of Data)(Miss Penalty)
- (Fraction of Instruction)(Miss Rate on Instruction)(Miss Penalty) + (Fraction of Data)(Miss Rate on Data)(Miss Penalty)
- (MemStall for Instruction + MemStall for Data)
What does %inst represent in the formula for MemStall?
What does %inst represent in the formula for MemStall?
- Fraction of Data
- Miss Penalty
- Fraction of Instruction (correct)
- Miss Rate on Instruction
What is the formula for %inst?
What is the formula for %inst?
What does MemStall for Data represent in Harvard Memory Architecture?
What does MemStall for Data represent in Harvard Memory Architecture?
Which of the following is the correct formula for MemStall in a Harvard (Separate) Memory Architecture?
Which of the following is the correct formula for MemStall in a Harvard (Separate) Memory Architecture?
What is the formula for CPIStall in terms of MemStall?
What is the formula for CPIStall in terms of MemStall?
In the example given, what is the value of MemStall for the Unified Memory Architecture?
In the example given, what is the value of MemStall for the Unified Memory Architecture?
What is the formula for Speedup in terms of CPI?
What is the formula for Speedup in terms of CPI?
In the example given, what is the value of CPI for the Separate Memory Architecture?
In the example given, what is the value of CPI for the Separate Memory Architecture?
What is the Speedup of the Separate Memory Architecture compared to the Unified Memory Architecture?
What is the Speedup of the Separate Memory Architecture compared to the Unified Memory Architecture?
Flashcards are hidden until you start studying
Study Notes
Data and Instruction Level 1 Cache
- Unified Cache (Princeton Memory Architecture):
- Instruction and Data are located on the same memory
- MemStall = (Miss Rate)(Miss Penalty)
- Separate Cache (Harvard Memory Architecture):
- Instruction and Data have separate memory locations
- MemStall = MemStall for Instruction + MemStall for Data
MemStall for Instruction and Data
- MemStall for Instruction = (Fraction of Instruction)(Miss Rate on Instruction)(Miss Penalty)
- Fraction of Instruction or %inst = 1/(MemAccess/Instruction) or 1/(1 + %load + %store)
- MemStall for Data = (Fraction of Data)(Miss Rate on Data)(Miss Penalty)
- Fraction of Data or %data = (%load + %store)/(MemAccess/Instruction) or (%load + %store)/(1 + %load + %store)
CPIStall for Harvard (Separate) Memory Architecture
- CPIStall = (1 + %load + %store)(MemStall)
- CPIStall = (1 + %load + %store)((1/(1 + %load + %store))(Im1) + ((%load + %store)/(1 + %load + %store))(Dm1))(M)
- CPIStall = (Im1 + (%load + %store)(Dm1))(M)
CPI and Speedup Examples
- Unified Cache:
- MemStall = (Miss Rate)(Miss Penalty)
- CPIStall = (1 + %load + %store)(MemStall)
- CPI = CPIPerfect + CPIStall
- Speedup = CPI/CPIPerfect
- Separate Cache:
- MemStall = ((%inst)(Im1) + (%data)(Dm1))(Miss Penalty)
- CPIStall = (1 + %load + %store)(MemStall)
- CPI = CPIPerfect + CPIStall
- Speedup = CPI/CPIPerfect
Studying That Suits You
Use AI to generate personalized quizzes and flashcards to suit your learning preferences.
