v2-n-gpus

See readme for details.

Download largemm executable using the link below.

v1.2-n-streams-lut

Release Notes - Version 1.2

New Features and Enhancements

N-Streams Support

• N-Streams Functionality: Version 1.2 introduces the ability to specify the number of streams as a command-line argument. Users can now provide the desired number of streams when running the program. For example, users can use the following command to specify 4 streams:

./your_program -s 4 -m {matrix_width} -a {algorithm}

Lookup Tables Integration

• Lookup Tables Integration: This release includes the integration of lookup tables. Users can specify the path to a CSV file containing a LUT using the -f flag. Based on the matrix size, the program will determine the best number of streams to use at run time. This feature allows dynamically deciding the number of streams based on the matrix size input.

./your_program -s -1 -m {matrix_width} -a {algorithm} -f path_to_lookup_table.csv

Full Changelog: https://github.com/Zlisch/LargeMM/commits/v1.2-n-streams-lut