Java SoftFloat

The new release contains the following changes:

• Added half-precision (16-bit) floating-point operations
• Updated single-precision (32-bit) and double-precision (64-bit) floating-point operations
• Removed double-extended-precision and quadruple-precision (128-bit) floating-point operations
• Updated test cases for 16, 32, and 64-bit floating-point operations

Download: https://forge.ispras.ru/projects/jsoftfloat/files

The build fully supports four floating-point formats:

• 32-bit single-precision;
• 64-bit double-precision;
• 80-bit double-extended-precision;
• 128-bit quadruple-precision.

JSoftFloat can be downloaded from the page http://forge.ispras.ru/projects/jsoftfloat/files.