Xilinx RAM COE Generator

Overview

This third party python script with no dependency other than native python package is used to generate Xilinx *.coe files for RAM data initializing

Xilinx RAM COE files aka coefficient files is used to initiate the data associated with their addresses inside Block RAM or other types of RAM.

The reference for Xilinx *.coe files can be found here at COE File Syntax on Xilinx support website.

Motivation

Although it is possible to use one RAM controller for each data with large FPGA setup, saving multiple data on single RAM usually costs us less resource requirement and provides routing simplicity thus decreasing development time. However, this usually comes with the expenses of doing Memory Mapping on available variables.

The *coe files is read and write in linear fashion, ie. it will start from address 0x0 all the ways until 0x100000 (your end). Therefore, saving a series of data starting from arbitary addresses (eg. 0x1A32) usually is a hard task as we might need to count the data manually.

Besides, data collision in same address is also possible as chucks of data might overlap with each other

This python script is use to translate addresses and data specified by the input file, then arrange them to generate *.coe files

Using it

• Clone or download the repository
• Update the config.ini files
• Write your own input file using the syntax specified at the following section
• run python xilinx-coe-generator.py

Available setting in config.ini

Default setting values is defined in the [default] section, user can overwrite the setting in [user] section

; value in USER section overwrite the default config value
[USER]
INPUT_FILE = example/my_new_file.txt

; The value in DEFAULT section shall be left untouched
[DEFAULT]
; The input file name written with defined syntax
INPUT_FILE = example/normal.txt
; The file name for output coe file
OUTPUT_FILE = output.coe

; the maximum range for each bit, 16 for hex
ACCEPT_RADIX = 16
; maximum size of the targetted RAM
MAX_SIZE = 150000
; default data for the remaining data of RAM
DEFAULT_DATA = 0
; number of datas in each line of coe output, wont affect Xilinx RAM but affect our readability
NUM_DATA_EACH = 10

This will cause the script to take "example/my_new_file.txt" as input, then follow default value for other variables.

Documentation

This python script is developed on python 3.7.1 on Ubuntu 18.04 LTS

Accepted syntaxes for input files

Here are 4 of the syntaxes acceptable for this script
The spaces in between = or * is optional.

0x[address] = [data] * [number of repeat]
0x[address] = [data]
[data] * [number of repeat]
[data]

In the following sections, please note that we will be using w/ "0x" for address, w/o "0x" for data. Both are in hexadecimal.

Syntax:

0x[address] = [data]

Eg. Saving data 0xAB3432 into address 0x123AB shall be

0x123AB = AB3432 

Saving chucks of data

To save this chuck of data one after another

Address	Data
0x10	0x20
0x11	0x21
...	...
0x1F	0x2F

Shall be written like this in input file

0x10 = 20
21
22
...
2F

Saving repetitive data

One can use " * " as indicator for repetitive data. Notes that [ number of repeat ] is in decimal of base 10

Syntax:

0x[address] = [data] * [number of repeat]

Eg. saving 12 data with value of 0xABC in address range of 0x110 to 0x11C

0x110 = ABC * 12

Address	Data
0x110	0xABC
0x111	0xABC
...	...
0x11B	0xABC

Using two types of syntax together

It is also possible to generate repeative values after assigning address on previous line.

0x100 = 1102321C
20932123
32212CDA * 3
BC312212

Will result in

Address	Data
0x100	0x1102321C
0x101	0x20932123
0x102	0x32212CDA
0x103	0x32212CDA
0x104	0x32212CDA
0x105	0xBC312212

Using examples

Will be added in the future

To do list

• Accepting comment in input file
• Adding more example
• Adding command line interface (CLI) support
• Adding Collision alert. ie. warning for saving multiple data on same address
• Reading files directly as binary

Disclaimer

Although the contributor of this project had made every attempt to ensure the best possible accuracy of this script. However, this python script is provided "as is" without warranty of any kind. The contributor does not take any responsibility or liability for the damage of lost caused by the usage of this script.

By using this script, you are hereby agree to the condition stated in the disclaimer section of this project.

Wrapping up!

This project is maintained as side project of an undergraduate. So updates might come a bit occasionally

For any issues, please submit an issues or contact the author at [email protected]

Contributing

Have all the fun forking the repository, check out, then submit your new feature as pull request,
I will be more than happy to accept it :D