A simple C project to visualize temporal locality of memory accesses using the stack distance method. It helps evaluate how well a program exhibits locality, a key factor in memory performance.
| File Name | Description |
|---|---|
stack_distance_tool.h |
Header file defining stack size and histogram bins |
stack_distance_tool.c |
Implements stack distance tracking and histogram printing |
main_test_runner.c |
Reads a fixed access sequence and prints the histogram |
generate_sequence.c |
Generates a synthetic address access pattern with localities |
Compile with gcc:
# Compile main analysis tool
gcc -o analyze main_test_runner.c stack_distance_tool.c
# Compile synthetic trace generator
gcc -o generate generate_sequence.c./generateThis will output a C array of 100 memory addresses simulating local and remote accesses.
Paste the generated array into main_test_runner.c → addresses[], then run:
./analyzeYou will see output like:
=== 栈距离分布柱状图（时间局部性） ===
距离 0: 10 | **********
距离 1: 6 | ******
距离 2: 4 | ****
...
距离 19: 0 |
This histogram shows how many times each stack distance occurred — smaller distances imply higher temporal locality.
Stack distance is the number of unique addresses accessed between two references to the same address.
- Low stack distance = recent reuse ⇒ high temporal locality
- High stack distance = poor reuse ⇒ low locality
- A program with many accesses in distance 0–10 likely performs well with small caches.
- You can visually identify locality by observing how “left-heavy” the histogram is.
- Evaluate how loop-structured programs exhibit locality
- Design better caching policies
- Teach memory performance concepts in architecture courses
MIT License.
Built for learning and analysis of temporal locality.
Feel free to fork and extend for file input, multiple program traces, or real-world benchmarks!