Dhrystone

  1. Unzip: tar -xzvf dhrystone-2.1.tar.gz
  2. Compile: make. Any errors?
  3. Fix the compilation errors
  4. Why do you think they put those function definition there if they are available in the system?
  5. Run the benchmark: ./gcc_dry2. Set the number of iterations to 100000000.
  6. Compare the results with the colleague next to you. Are they different?
  7. Run with register variables: ./gcc_dry2reg. Set the number of iterations to 100000000.
  8. Let's discuss about register variables.

CoreMark

  1. Unzip: tar -xzvf coremark_v1.0.tgz
  2. Compile: make
  3. Run benchmark: make run
  4. Compare the results with the colleague next to you. Are they different?

lmbench

  1. Unzip: tar -xzvf lmbench-3.0-a9.tgz
  2. Compile: make
  3. Run lat_mem_rd: ./bin/x86_64-linux-gnu/lat_mem_rd 250 16 32 64 128 256 512 1024
  4. Guess the cache hierarchy and size.

Evaluate context switch overhead

  1. run benchmark ./bin/x86_64-linux-gnu/lat_ctx -s 80 2 4 6 8 10 12 14 16 18 20 24 32 48 96
  2. Why do you think the results are so close at the beginning?

Run lmbench default template

  1. cd scripts
  2. ./config-run with MB: 100
  3. ./results
  4. cd ./results
  5. make
sesiuni/virtualization-networking/session4-2014.txt · Last modified: 2014/07/08 16:13 by freescale