Behaviour: Underflow flag not set at subnormal/normal boundary for FMA. According to IEEE 754, the underflow for after rounding should be raise when a non-zero result computed as though the exponent range were unbounded would lie strictly between ±b^(emin). The floatrix app has found a situation where the operand_a is 16'h03ee, operand_b is 16'h3c12 and operand_c is 16'b8000 result into the computed result that lied at the boundary of subnormal/normal boundary, and should be inexact with underflow flag.
Related ticket: A similar ticket was also found in cvfpu: openhwgroup/cvfpu#94
Behaviour: Underflow flag not set at subnormal/normal boundary for FMA. According to IEEE 754, the underflow for after rounding should be raise when a non-zero result computed as though the exponent range were unbounded would lie strictly between ±b^(emin). The floatrix app has found a situation where the operand_a is 16'h03ee, operand_b is 16'h3c12 and operand_c is 16'b8000 result into the computed result that lied at the boundary of subnormal/normal boundary, and should be inexact with underflow flag.
Related ticket: A similar ticket was also found in cvfpu: openhwgroup/cvfpu#94