triton.language.extra.cann.libdevice

Libdevice Ops

abs	Computes the absolute value of the input parameter.
acos	Computes the inverse cosine (arccos) of the input parameter.
acosh	Computes the inverse hyperbolic cosine of the input parameter.
add_rd	Floating-point addition with round-down (toward negative infinity) rounding mode.
add_rn	Floating-point addition with round-to-nearest-even rounding mode.
add_ru	Floating-point addition with round-up (toward positive infinity) rounding mode.
add_rz	Floating-point addition with round-toward-zero rounding mode.
asin	Computes the inverse sine (arcsin) of the input parameter.
asinh	Computes the inverse hyperbolic sine of the input parameter.
atan	Computes the inverse tangent (arctan) of the input parameter.
atan2	Two-argument inverse tangent function, computes the arctangent of x / y.
atanh	Inverse hyperbolic tangent function, computes the inverse hyperbolic tangent of the input parameter.
brev	Bit reversal function, reverses the bit order of a 32-bit integer.
byte_perm	Selects bytes from two 32-bit integers x and y according to selector s and combines them into a new integer.
cbrt	Computes the cube root of the input parameter.
ceil	Ceiling operation, returns the smallest integer greater than or equal to x.
clz	Counts the number of leading zeros in a 32-bit integer.
copysign	Generates a floating-point number with magnitude equal to the magnitude of x and sign equal to the sign of y.
cos	Computes the cosine of the input parameter (in radians).
cosh	Computes the hyperbolic cosine of the input parameter.
cospi	Computes the value of cos(π × x).
cyl_bessel_i0	Computes the modified Bessel function of the first kind, order 0, of the input parameter.
cyl_bessel_i1	Computes the modified Bessel function of the first kind, order 1, of the input parameter.
div_rd	Floating-point division with round-down (toward negative infinity) rounding mode.
div_rn	Floating-point division with round-to-nearest-even rounding mode.
div_ru	Floating-point division with round-up (toward positive infinity) rounding mode.
div_rz	Floating-point division with round-toward-zero rounding mode.
erf	Computes the error function of the input parameter.
erfc	Computes the complementary error function of the input parameter, i.e., 1 - erf(x).
erfcinv	Inverse complementary error function, finds the value y such that x = erfc(y).
erfcx	Computes the scaled complementary error function of the input parameter, i.e., exp(x²) × erfc(x).
erfinv	Inverse error function, finds the value y such that x = erf(y).
exp	Exponential function, computes e raised to the power of x.
exp10	Base-10 exponential function, computes 10 raised to the power of x.
exp2	Base-2 exponential function, computes 2 raised to the power of x.
expm1	Computes e raised to the power of x, minus 1.
fast_cosf	Fast approximate cosine function.
fast_dividef	Fast approximate division.
fast_exp10f	Fast approximate base-10 exponential function.
fast_expf	Fast approximate exponential function.
fast_log10f	Fast approximate base-10 logarithm function.
fast_log2f	Fast approximate base-2 logarithm function.
fast_logf	Fast approximate natural logarithm function.
fast_powf	Fast approximate power function.
fast_sinf	Fast approximate sine function.
fast_tanf	Fast approximate tangent function.
fast_tanhf	Computes the hyperbolic tangent of x using a fast approximation.
fdim	Computes the positive difference between x and y.
ffs	Finds the first bit set to 1 and returns the index of the lowest bit set to 1.
finitef	Determines whether the input is a finite floating-point number.
float2half_rn	Converts x from a 32-bit floating-point value to a 16-bit floating-point value using round-to-nearest-even.
float2int_rd	Converts a floating-point number to a 32-bit integer with round-down mode.
float2int_rn	Converts a floating-point number to a 32-bit integer with round-to-nearest-even mode.
float2int_ru	Converts a floating-point number to a 32-bit integer with round-up mode.
float2int_rz	Converts a floating-point number to a 32-bit integer with round-toward-zero mode.
float2ll_rd	Converts a floating-point number to a 64-bit integer with round-down mode.
float2ll_rn	Converts a floating-point number to a 64-bit integer with round-to-nearest-even mode.
float2ll_ru	Converts a floating-point number to a 64-bit integer with round-up mode.
float2ll_rz	Converts a floating-point number to a 64-bit integer with round-toward-zero mode.
float2uint_rd	Converts a floating-point number to a 32-bit unsigned integer with round-down mode.
float2uint_rn	Converts a floating-point number to a 32-bit unsigned integer with round-to-nearest-even mode.
float2uint_ru	Converts a floating-point number to a 32-bit unsigned integer with round-up mode.
float2uint_rz	Converts a floating-point number to a 32-bit unsigned integer with round-toward-zero mode.
float2ull_rd	Converts a floating-point number to a 64-bit unsigned integer with round-down mode.
float2ull_rn	Converts a floating-point number to a 64-bit unsigned integer with round-to-nearest-even mode.
float2ull_ru	Converts a floating-point number to a 64-bit unsigned integer with round-up mode.
float2ull_rz	Converts a floating-point number to a 64-bit unsigned integer with round-toward-zero mode.
float_as_int	Reinterprets the bit pattern of a floating-point number as a 32-bit integer.
float_as_uint	Reinterprets the bit pattern of a floating-point number as a 32-bit unsigned integer.
floor	Floor operation, returns the largest integer less than or equal to x.
fma	Fused multiply-add, computes x × y + z.
fma_rd	Fused multiply-add operation with round-down rounding mode.
fma_rn	Fused multiply-add operation with round-to-nearest-even rounding mode.
fma_ru	Fused multiply-add operation with round-up rounding mode.
fma_rz	Fused multiply-add operation with round-toward-zero rounding mode.
fmod	Floating-point modulo, computes the remainder of x / y, with the same sign as x.
gamma	Computes the gamma function of the input parameter.
hadd	Computes the average of x and y.
half2float	Converts x from a 16-bit floating-point value to a 32-bit floating-point value.
hypot	Computes the Euclidean distance between x and y.
ilogb	Extracts the unbiased exponent (base-2 integer logarithm) of a floating-point number.
int2float_rd	Converts a 32-bit integer to a floating-point number with round-down mode.
int2float_rn	Converts a 32-bit integer to a floating-point number with round-to-nearest-even mode.
int2float_ru	Converts a 32-bit integer to a floating-point number with round-up mode.
int2float_rz	Converts a 32-bit integer to a floating-point number with round-toward-zero mode.
int_as_float	Reinterprets the bit pattern of a 32-bit integer as a floating-point number.
isinf	Determines whether the input is infinity.
isnan	Determines whether the input is NaN (not a number).
j0	Computes the Bessel function of the first kind of order 0 of the input parameter.
j1	Computes the Bessel function of the first kind of order 1 of the input parameter.
jn	Computes the Bessel function of the first kind of order n of the input parameter.
ldexp	Computes the value of x × 2^exp.
lgamma	Computes the natural logarithm of the absolute value of the gamma function for input x.
ll2float_rd	Converts a 64-bit integer to a floating-point number with round-down mode.
ll2float_rn	Converts a 64-bit integer to a floating-point number with round-to-nearest-even mode.
ll2float_ru	Converts a 64-bit integer to a floating-point number with round-up mode.
ll2float_rz	Converts a 64-bit integer to a floating-point number with round-toward-zero mode.
llrint	Rounds a floating-point number to the nearest 64-bit integer value.
llround	Rounds a floating-point number to the nearest 64-bit integer value.
log	Computes the natural (base-e) logarithm of input x.
log10	Computes the base-10 logarithm of input x.
log1p	Computes the value of log(1 + x).
log2	Computes the base-2 logarithm of input x.
logb	Extracts the exponent value of a floating-point number.
max	Computes the element-wise maximum of x and y.
min	Computes the element-wise minimum of x and y.
mul24	Computes the lower 24-bit multiplication result of x and y.
mul_rd	Floating-point multiplication with round-down rounding mode.
mul_rn	Floating-point multiplication with round-to-nearest-even rounding mode.
mul_ru	Floating-point multiplication with round-up rounding mode.
mul_rz	Floating-point multiplication with round-toward-zero rounding mode.
mulhi	Computes the high 32 bits of the multiplication result of x and y.
nan	Generates a NaN value from x.
nearbyint	Converts x to the nearest integer.
nextafter	Computes the next representable floating-point number from x toward y.
norm3d	Computes the Euclidean norm of a 3D vector, i.e., sqrt(x² + y² + z²).
norm4d	Computes the Euclidean norm of a 4D vector, i.e., sqrt(x² + y² + z² + w²).
normcdf	Computes the cumulative distribution function of the standard normal distribution.
normcdfinv	Computes the inverse of the cumulative distribution function of the standard normal distribution.
popc	Counts the number of bits set to 1 in x.
pow	Power function, computes x raised to the power of y.
rcbrt	Computes the reciprocal cube root of x.
rcp_rd	Floating-point reciprocal with round-down rounding mode.
rcp_rn	Floating-point reciprocal with round-to-nearest-even rounding mode.
rcp_ru	Floating-point reciprocal with round-up rounding mode.
rcp_rz	Floating-point reciprocal with round-toward-zero rounding mode.
reciprocal	Computes the reciprocal of x (i.e., 1 / x).
relu	Rectified linear unit function, returns x when x > 0, otherwise returns 0.
remainder	Computes the remainder of x divided by y, where r = x - ny, and n is the nearest integer to x / y.
rhadd	Computes the rounded average of x and y.
rhypot	Computes the reciprocal of the Euclidean distance between x and y.
rint	Computes the nearest integer to x using round-to-nearest-even rounding mode.
rnorm3d	Computes the reciprocal of the Euclidean norm of a 3D vector.
rnorm4d	Computes the reciprocal of the Euclidean norm of a 4D vector.
round	Computes the nearest integer to x using round-to-nearest-even rounding mode.
rsqrt	Computes the reciprocal square root of x.
rsqrt_rn	Computes the reciprocal square root of x using round-to-nearest-even rounding mode.
sad	Computes abs(x - y) + z for signed or unsigned 32-bit integers.
saturatef	Clamps x to the range [+0.0, 1.0].
scalbn	Computes the value of x × 2^n.
signbit	Extracts the sign bit of x.
sin	Computes the sine of the input parameter x (in radians).
sinh	Computes the hyperbolic sine of input parameter x.
sinpi	Computes the value of sin(π × x).
sqrt	Computes the square root of x.
sqrt_rd	Computes the square root of x with round-down rounding mode.
sqrt_rn	Computes the square root of x with round-to-nearest-even rounding mode.
sqrt_ru	Computes the square root of x with round-up rounding mode.
sqrt_rz	Computes the square root of x with round-toward-zero rounding mode.
sub_rd	Floating-point subtraction with round-down rounding mode.
sub_rn	Floating-point subtraction with round-to-nearest-even rounding mode.
sub_ru	Floating-point subtraction with round-up rounding mode.
sub_rz	Floating-point subtraction with round-toward-zero rounding mode.
tan	Computes the tangent of input parameter x (in radians).
tanh	Computes the hyperbolic tangent of input parameter x.
tgamma	Computes the gamma function of the input parameter.
trunc	Truncation operation, rounds toward zero to the nearest integer.
uint2float_rd	Converts a 32-bit unsigned integer to a floating-point number with round-down mode.
uint2float_rn	Converts a 32-bit unsigned integer to a floating-point number with round-to-nearest-even mode.
uint2float_ru	Converts a 32-bit unsigned integer to a floating-point number with round-up mode.
uint2float_rz	Converts a 32-bit unsigned integer to a floating-point number with round-toward-zero mode.
uint_as_float	Reinterprets the bit pattern of a 32-bit unsigned integer as a floating-point number.
ull2float_rd	Converts a 64-bit unsigned integer to a floating-point number with round-down mode.
ull2float_rn	Converts a 64-bit unsigned integer to a floating-point number with round-to-nearest-even mode.
ull2float_ru	Converts a 64-bit unsigned integer to a floating-point number with round-up mode.
ull2float_rz	Converts a 64-bit unsigned integer to a floating-point number with round-toward-zero mode.
y0	Computes the Bessel function of the second kind of order 0 of the input parameter.
y1	Computes the Bessel function of the second kind of order 1 of the input parameter.
yn	Computes the Bessel function of the second kind of order n of the input parameter.