store (simple alias of xchg)

add

and

cmpxchg (up here, because it's needed by the fallback implementations)

cmpxchg

dec

dec

inc

inc

load

min

or

or

constructors:

store

store

sub

xchg

xchg

xor

xor

sum of all components / #components

+Z backward vector

compresses the data in "input" and writes it to a newly allocated vector, returns the compressed data as a vector of the correct size on success, or r

computes (n choose k), the binomial coefficient NOTE: only safe to call up to n = 67, after this results may no longer fit into 64-bit

computes (n choose k), the binomial coefficient NOTE: this allows for larger n than binomial(n, k), but recursiveness gets ugly for n > 80

device bitness NOTE: we only support 64-bit now

blits the "src" image onto this image, returns true on success NOTE: dim must be identical, format must be compatible TODO: implement this everywhere

asynchronously blits the "src" image onto this image, returns true if encoding was successful, waits for all fences specified in "wait_fences" and sig

returns true if this context can compile programs from source code at run-time

returns true if external memory can be used (i.e. Vulkan buffer/image sharing)

returns a canonicalized copy of this quaternion (r will be positive)

returns a canonicalized copy of this dual quaternion (r component in rq will be positive)

cubic catmull-rom interpolation (with *this = a, interpolation between a and b, and values in order [a_prev, a, b, b_next])

clamps all components of this vector to [min, max]

zeros/clears the complete memory object, returns true on success

returns an int vector containing the number of leading 0-bits of each component in this vector

returns the combined form of the conjugate of this dual quaternion (third conjugate)