Quiz 4 Solution - PowerPoint PPT Presentation

About This Presentation
Title:

Quiz 4 Solution

Description:

L1-ICACHE : 2% miss rate, 32-byte blocks. L1-DCACHE :5% miss rate, 16-byte blocks ... 1-Bigger L1, cut missrate to half, L1 hit= 1 cycle =0.4 ns ... – PowerPoint PPT presentation

Number of Views:15
Avg rating:3.0/5.0
Slides: 7
Provided by: Amir82
Category:
Tags: hit1 | quiz | solution

less

Transcript and Presenter's Notes

Title: Quiz 4 Solution


1
Quiz 4 Solution
2
  • Frequency 2.5GHz , CLK 0.4ns
  • CPI 0.4, 30 loads and stores,
  • L1 hit 0,
  • L1-ICACHE 2 miss rate, 32-byte blocks
  • L1-DCACHE 5 miss rate, 16-byte blocks
  • L2-Cache 10 miss rate, access time15 ns,
    64-byte blocks
  • Memory access time 75 ns
  • Miss penalty for data between L1 and
    L215ns16/162.5ns17.5ns
  • Miss penalty for inst between L1 and L2
    15ns32/162.5ns 20ns
  • Miss penalty between L2 and Memory
    75ns64/167.5ns 105ns
  • Part a)
  • DATA_AMAT L1 hit time L1 data miss rate ( L2
    hit rate(Data in L2 penalty)L2 miss rate (Data
    not in L2 but in memory penalty))
  • 00.05(0.9(17.5)0.1(17.5105))1.4
  • INST_AMAT L1 hit time L1 inst miss rate ( L2
    hit rate(inst in L2 penalty)L2 miss rate (inst
    not in L2 but in memory penalty))
  • 00.02(0.9(20)0.1(20105))0.61

3
  • Frequency 2.5GHz , CLK 0.4ns
  • CPI 0.4, 30 loads and stores,
  • L1 hit 0,
  • L1-ICACHE 2 miss rate, 32-byte blocks
  • L1-DCACHE 5 miss rate, 16-byte blocks
  • L2-Cache 10 miss rate, access time15 ns,
    64-byte blocks
  • Memory access time 75 ns
  • DATA_AMAT
  • 00.05(0.9(17.5)0.1(17.5105))1.4
  • INST_AMAT
  • 00.02(0.9(20)0.1(20105))0.61
  • Part b)
  • Over-all CPI Ideal CPIimpact of data
    accessesimpact of inst accesses
  • 0.4(0.3(AMAT_DATA)1(AMAT_INST))xfrequency
  • We need to take frequency into account as this is
    CPI and not execution time.
  • 0.4((0.3(1.4)1(0.61))2.5GHZ)2.975

4
  • Frequency 2.5GHz , CLK 0.4ns
  • CPI 0.4, 30 loads and stores,
  • L1 hit 0,
  • L1-ICACHE 2 miss rate, 32-byte blocks
  • L1-DCACHE 5 miss rate, 16-byte blocks
  • L2-Cache 10 miss rate, access time15 ns,
    64-byte blocks
  • Memory access time 75 ns
  • Over-all CPI0.4((0.3(1.4)1(0.61))2.5GHZ)2.975
  • Part c) Replace 2.5GHZ processor w/ 3.6GHZ.
    Improvement?
  • New CPI0.4((0.3(1.4)1(0.61))3.6GHZ)4.1
  • execution timeCPIClock period
  • Improvementold execution time/new execution time
  • (2.975/2.5)/(4.1/3.6)
    1.19/1.131.05
  • 5 improvement!

5
  • Part d)
  • 1-Bigger L1, cut missrate to half, L1 hit 1
    cycle 0.4 ns
  • DATA_AMAT L1 hit time L1 data miss rate ( L2
    hit rate(Data in L2 penalty)L2 miss rate (Data
    not in L2 but in memory penalty))
  • 0.40.025(0.9(17.5)0.1(17.5105))1.1
  • INST_AMAT L1 hit time L1 inst miss rate ( L2
    hit rate(inst in L2 penalty)L2 miss rate (inst
    not in L2 but in memory penalty))
  • 0.40.01(0.9(20)0.1(20105))0.7
  • Over-all CPI Ideal CPIimpact of data
    accessesimpact of inst accesses
  • 0.4(0.3(AMAT_DATA)1(AMAT_INST))xfrequency
  • 0.4(0.3(1.1)1(0.7))frequency2.975. No
    improvement.

6
  • Part d)
  • 2-Smaller L2, L2 access time 10 , L2 miss
    rate15
  • Miss penalty for data between L1 and
    L210ns16/162.5ns12.5ns
  • Miss penalty for inst between L1 and L2
    10ns32/162.5ns 15ns
  • DATA_AMAT L1 hit time L1 data miss rate ( L2
    hit rate(Data in L2 penalty)L2 miss rate (Data
    not in L2 but in memory penalty))
  • 00.05(0.85(12.5)0.15(12.5105))1.41
  • INST_AMAT L1 hit time L1 inst miss rate ( L2
    hit rate(inst in L2 penalty)L2 miss rate (inst
    not in L2 but in memory penalty))
  • 00.02(0.85(15)0.15(15105))0.61
  • Over-all CPI Ideal CPIimpact of data
    accessesimpact of inst accesses
  • 0.4(0.3(AMAT_DATA)1(AMAT_INST))xfrequency
  • 0.4(0.3(1.41)1(0.61))frequency2.982. lose
    performance.
Write a Comment
User Comments (0)
About PowerShow.com