Lab 08

Deadline: EOD (11PM PT) Wednesday, April 07

Objectives:

  • The student would explore the workings of virtual memory, specifically the TLB and the Page Table.
  • The student would be able to analyze TLB hit rate and Page Table hit rate and figure out what accesses optimize these values.

Setup

Pull the lab 8 files from the lab starter repository with

git pull starter master

Write your answers in the provided answers.txt file. For numerical answers, write the number instead of spelling it out (e.g. “7” instead of “seven”). The autograder assumes that the original formatting will not be changed, so please don’t add additional lines, switch existing lines, or otherwise modify the current formatting.

For this lab we will mostly be using the virtual memory simulation features of Camera, a cache and virtual memory simulator. You may also find the cache simulations interesting, however we won’t be working with those here. Unfortunately, Camera is known to have issues when trying to run it on the Hive or Linux machines, so it’s recommend to download Camera from here, and simply double click on the jar file to run it on your own (non-Linux) laptop. If you’re on a Mac, you may need to go to “Security & Privacy” in your settings and click “Open Anyway” to allow Camera to run. Some displays don’t seem to play nice with the standard Camera app, if the values in memory are all squished together, try running this version of Camera. If you are unable to find a way to get Camera working on a machine, please partner up with someone who does.

Once Camera opens up, select the virtual memory option to open a visualization of the virtual memory system. In the top left you can see the contents of physical memory. Just below that is a listing of all the pages of virtual memory for this process. To the right of these items are the contents of the TLB and the Page Table. At this point these should all be empty as we haven’t done anything yet. Read about the statistics of your memory system in the “PROGRESS UPDATE” box at the bottom of the window. This area will keep you updated on your status through the simulation as it progresses. You can move the simulation forward, backward or start it over from the beginning using the buttons to the right of the “PROGRESS UPDATE” box.

Exercise 0 - Sanity Check

Before you continue, MAKE SURE THAT YOU OPENED THE VM SIMULATOR AND NOT THE CACHE SIMULATOR.

Exercise 1 - Working with CAMERA

Click the button labeled “Auto Generate Add. Ref. Str.” at the right-hand side of the window. This will generate a set of ten address references. You can think of these as a series of RISC-V “load word” instructions reading from the memory address specified. Click the button labeled “Next” to begin the simulation.

For the rest of this exercise you are at the mercy of the “PROGRESS UPDATE” box. After each click of the “Next” button examine the contents of the box and the current state of the memory system. Try to really get an understanding of what is going on in the TLB, the Page Table, and Physical Memory at each step.

Checkpoint

  1. Given the way the address was broken down, how big (in words) are the pages in this model? Leave out the units in your answer (e.g. 4 instead of 4 words).
  2. How many TLB Hits and Misses did we have for the randomly generated set of ten addresses? What about for Page Hits and Page Faults? Your answer should be a comma separated list (e.g. 1, 2, 3, 4).
  3. Did you have any Page Hits? (Why?) Can you think of a set of ten addresses that would result in a Page Hit? Your answer should be two [yes/no]’s separated by a comma.
  4. What is the process by which we access memory given a virtual address on the very first access assuming a page fault? Order the steps given below. Give your answer as numbers delimited by by a comma with a space, i.e. [num], [num], … for example, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12.
    1. We update the page table to map the corresponding VPN to the PPN.
    2. Calculate number of virtual page number (VPN) bits through .
    3. TLB does not contain the VPN so we access the page table for the corresponding VPN.
    4. We access the corresponding byte using the offset.
    5. We bring the corresponding virtual page into physical memory from disk.
    6. The page table’s entry for VPN has a valid bit of 0.
    7. Get VPN by taking VA[address bits - 1 : offset bits].
    8. Calculate number of offset bits by taking log 2 of page size.
    9. We update the TLB with the corresponding PT entry.
    10. Get offset by taking VA[offset bits - 1 ; 0].
    11. Access TLB for corresponding VPN.
    12. Access given virtual address.
  5. How many PPN and VPN bits are there respectively? Format your answer as NUM\_PPN\_BITS, NUM\_VPN\_BITS.
  6. How many physical and virtual pages are there respectively? Format your answer as NUM\_PHYSICAL\PAGES, NUM\_VIRTUAL\_PAGES.

Exercise 2 - Misses

Now that you’ve seen what a random workload looks like in the VM system let’s try creating a custom workload with a specific property. Your goal for this exercise is to create a workload of ten memory accesses that will cause ten TLB misses and ten Page Faults. You should be able to come up with such a workload on paper, but then you should run it in CAMERA to verify your work. You can specify a custom workload in CAMERA by clicking the button labeled “Self Generate Add. Ref. Str.” and entering in the addresses you want to reference one at a time.

Checkpoint

  1. Write down your ten memory accesses. The answer should be formatted as a comma separated list of hex values (e.g. 00, 01, 02, 03, 04, 05, 06, 07, 08, 09).

Exercise 3 - Fixing our Faults

Given your sequence of memory accesses from Exercise 2, can you find a change to a single parameter (e.g. TLB Size, Physical Memory Size, Virtual Memory Size, Page Size) that would result in the same number (ten) of TLB misses but result in fewer than ten page faults?

Checkpoint

  1. Write down two parameters for which if each got individually changed (while all other parameters stay the same) would result in ten TLB misses but fewer than ten page faults. Format your two answers in alphabetical order as such: A_ANS, B_ANS. For example, if your answers are daisy, cotton ball, write it as cotton ball, daisy.

Exercise 4 - Bringing it All Together

We used VMSIM, another Virtual Memory simulator, to create this question. You have two options for this exercise.

  • Watch this webm of a VMSIM simulation.
  • Use the appletviewer command in your Terminal like so (doesn’t work on Hive):
$ appletviewer https://denninginstitute.com/workbenches/vmsim/vm.html 

Observe what is happnening and answer the following questions:

What is different about the setup of the system in this question as compared to the setup in CAMERA? In particular, what are P1, P2, P3, and P4? If you watch closely you’ll see that this simulation reports a much higher percentage of TLB misses than random runs on CAMERA did. Why might this be? (If you have trouble following the simulation, use the appletviewer and turn down the speed using the slider on the bottom right.)

Checkpoint

  1. Select all factors that would cause the .webm simulation to show a lower TLB hit rate than our original model. Give your answer in numerical order delimited by by a comma with a space, i.e. [num], [num], … an example would be 1, 2, 3.
    1. A single process is running in .webm.
    2. Multiple processes are running in .webm.
    3. Running multiple processes requires the system to context switch.
    4. Running a single process requires context switches.
    5. Context switches preserve TLB state.
    6. Context switches invalidate the TLB.

Submission

Please submit to the Lab Autograder assignment.

Note that the autograder is whitespace and case insensitive, but otherwise very simple and thus incapable of recognizing typos or misformatted answers.

The autograder looks for the answers.txt file.