Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/isaaccorley/python-cache-simulator
Graduate Computer Architecture Course Project - Cache Simulator
https://github.com/isaaccorley/python-cache-simulator
Last synced: about 1 month ago
JSON representation
Graduate Computer Architecture Course Project - Cache Simulator
- Host: GitHub
- URL: https://github.com/isaaccorley/python-cache-simulator
- Owner: isaaccorley
- Created: 2019-03-22T05:57:10.000Z (almost 6 years ago)
- Default Branch: master
- Last Pushed: 2019-03-22T06:15:41.000Z (almost 6 years ago)
- Last Synced: 2024-10-14T10:15:28.636Z (3 months ago)
- Language: Python
- Homepage:
- Size: 1.99 MB
- Stars: 1
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Python-Cache-Simulator
Cache Performance Simulator using a Least Recently Used (LRU) replacement policy.
## Usage:
usage: cache_sim.py [-h] -trace TRACE [-grid] [-config CONFIG]
optional arguments:
-h, --help show this help message and exit
-trace TRACE Path to memory address trace .trc file
-grid (Optional) Perform grid search across various configurations
-config CONFIG Path to simulation configuration .cfg file
### Example Usage:
Using config and trace files.
```
python3 cache_sim.py -trace example.trc -config example.cfg
```
Performing grid search across hardcoded config parameters (output is written to grid_search.csv).
```
python3 cache_sim.py -trace example.trc -grid
```
### Example Output:
100%|--------------------------------| 695521/695521 [00:31<00:00, 22304.47it/s]
Cache Size 32768 - Block Size 64 - Associativity 4
Num Blocks 512 - Num Sets 128
Tag Length 19 - Index Length 7 - Offset Length 6
Num Accesses 695521 - Num Hits 594110 - Num Misses 101411
Hit Rate: 85.42
### Dependencies (can be installed via pip):
- python==3.6.8
- numpy>=1.15.4
- tqdm>=4.29.1
## Implementation Details:
### Configuration Calculations:
Given cache size, block size, associativity (n, direct, or fully), address length = 32
- num_blocks = cache size / block size
- if associativity = direct
- associativity = 1
- num_sets = num_blocks
- elif associativity = fully
- associativity = num_blocks
- num_sets = 1
- else
- associativity = n <- the provided associativity
- num_sets = num_blocks // associativity
- offset length = log_2(block_size)
- index length = log_2(num_sets)
- tag length = address length - (index length + offset length)
### Tag, Index, Offset Calculations:
- offset = address & 2^(offset length) - 1
- tag = (address >> (index length + offset length)) & 2^(tag length) - 1
- if num_sets = 1
- index = 0 # only one set
- else
- index = (address >> (offset length)) & 2^(index length) - 1
### Default Data Structures:
Blocks: Python dictionary with the below default parameters
```
block = {
'valid': 0,
'tag': None,
'last_used': 0
}
```
Set: Python dictionary of n (associativity) block data structures indexed by number (0, 1, ..., n)
```
set = {n: block for n in range(associativity)}
```
Cache: Python dictionary of k (num_sets) set data structures indexed by number (0, 1, ..., k)
```
cache = {k: set for k in range(num_sets)}
```
### Cache Algorithm:
For Address in Trace
- Calculate Tag, Index, and Offset from Address
- If a Block in the Indexed Set has an equivalent Tag
- Update the Least Recently Used (LRU) parameter
- Hit = True
- Elif a Block in the Indexed Set has Valid bit = 0
- Store Tag in the Block
- Set Valid bit = 1
- Update the Least Recently Used (LRU) parameter
- Hit = False (Miss)
- Else
- Overwrite the LRU Block's Tag
- Update the Least Recently Used (LRU) parameter
- Hit = False (Miss)
After simulation, the hit rate = sum(hits) / # accesses * 100%
## Grid Search Results:
Grid search was performed with the following parameter space:
cache_sizes = [8*1024, 64*1024, 256*1024, 1024*1024]
associativity = [direct, 2, 4, 8, fully]
block_size = 32
| Cache Size | Block Size | Associativity | Hit Rate |
|:----------:|:----------:|:-------------:|:--------:|
| 8192 | 32 | direct | 75.5 |
| 65536 | 32 | direct | 75.76 |
| 262144 | 32 | direct | 77.85 |
| 1048576 | 32 | direct | 85.51 |
| 8192 | 32 | 2 | 75.52 |
| 65536 | 32 | 2 | 75.83 |
| 262144 | 32 | 2 | 77.85 |
| 1048576 | 32 | 2 | 86.95 |
| 8192 | 32 | 4 | 75.53 |
| 65536 | 32 | 4 | 75.92 |
| 262144 | 32 | 4 | 77.98 |
| 1048576 | 32 | 4 | 88.02 |
| 8192 | 32 | 8 | 75.54 |
| 65536 | 32 | 8 | 76.01 |
| 262144 | 32 | 8 | 78.81 |
| 1048576 | 32 | 8 | 88.9 |
| 8192 | 32 | fully | 75.56 |
| 65536 | 32 | fully | 76.07 |
| 262144 | 32 | fully | 79.08 |
| 1048576 | 32 | fully | 91.13 |