oprecomp/flexfloat

Issues

• compressed nan/inf representations used for new fp8? e4m3 e5m2

  #16 opened 2 years ago
  0

• Project doesn't install anything

  #15 opened 2 years ago
  0

• flexfloat_get_bits() gives wrong results

  #14 opened 2 years ago
  1

• The test value_representation_half fails

  #9 opened 2 years ago
  1

• No sqrt, fabs?

  #8 opened 5 years ago
  2

• NaN handling in tests makes too many assumptions

  #7 opened 2 years ago
  0