Module rustc_apfloat::ieee::sig

Implementation details of IeeeFloat significands, such as big integer arithmetic. As a rule of thumb, no functions in this module should dynamically allocate.

Functions

add

a += b + c where c is zero or one. Returns the carry flag.

add_or_sub

a += b or a -= b. Does not preserve b.

clear_bit

Clear the given bit.

cmp

Comparison (unsigned) of two significands.

decrement

Decrement in-place, return the borrow flag.

div

quotient = dividend / divisor. Returns the lost fraction. Does not preserve dividend or divisor.

each_chunk

For every consecutive chunk of bits bits from limbs, going from most significant to the least significant bits, call f to transform those bits and store the result back.

extract

Copies the bit vector of width src_bits from src, starting at bit SRC_LSB, to dst, such that the bit SRC_LSB becomes the least significant bit of dst. All high bits above src_bits in dst are zero-filled.

from_limbs

We want the most significant PRECISION bits of src. There may not be that many; extract what we can.

get_bit

Extracts the given bit.

increment

Increment in-place, return the carry flag.

is_all_zeros

mul

dst = a * b (for normal a and b). Returns the lost fraction.

olsb

One, not zero, based LSB. That is, returns 0 for a zeroed significand.

omsb

One, not zero, based MSB. That is, returns 0 for a zeroed significand.

set_bit

Sets the given bit.

shift_left

Shifts dst left bits bits, subtract bits from its exponent.

shift_right

Shifts dst right bits bits noting lost fraction.

sub

a -= b + c where c is zero or one. Returns the borrow flag.

widening_mul

[low, high] = a * b.