dymokomi/luce-kernel
Source of release 0.1.2
0.1.2 / src / engine / lanes_unary.lucb
# Generated by tools/lanes.py from its tables; edit those, not this file.
# Unary lane operations and float classifications: d = op a.
## Run a unary or classifying instruction; the lowest faulting lane, or the lane count.
func unary(w: Worker*, ins: Instruction) -> usize:
match ins.op:
.neg:
match ins.kind:
.f32: neg_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: neg_f64(w.f64s(ins.dst), w.f64s(ins.a))
.i32: return neg_i32(w.i32s(ins.dst), w.i32s(ins.a), w.on())
.i64: return neg_i64(w.i64s(ins.dst), w.i64s(ins.a), w.on())
_: return w.n
.wneg:
match ins.kind:
.i32: wneg_i32(w.i32s(ins.dst), w.i32s(ins.a))
.i64: wneg_i64(w.i64s(ins.dst), w.i64s(ins.a))
.u32: wneg_u32(w.u32s(ins.dst), w.u32s(ins.a))
.u64: wneg_u64(w.u64s(ins.dst), w.u64s(ins.a))
_: return w.n
.abs:
match ins.kind:
.f32: abs_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: abs_f64(w.f64s(ins.dst), w.f64s(ins.a))
.i32: return abs_i32(w.i32s(ins.dst), w.i32s(ins.a), w.on())
.i64: return abs_i64(w.i64s(ins.dst), w.i64s(ins.a), w.on())
_: return w.n
.bnot:
match ins.kind:
.i32: bnot_i32(w.i32s(ins.dst), w.i32s(ins.a))
.i64: bnot_i64(w.i64s(ins.dst), w.i64s(ins.a))
.u32: bnot_u32(w.u32s(ins.dst), w.u32s(ins.a))
.u64: bnot_u64(w.u64s(ins.dst), w.u64s(ins.a))
.mask: bnot_mask(w.masks(ins.dst), w.masks(ins.a))
_: return w.n
.sqrt:
match ins.kind:
.f32: sqrt_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: sqrt_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.cbrt:
match ins.kind:
.f32: cbrt_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: cbrt_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.exp:
match ins.kind:
.f32: exp_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: exp_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.exp2:
match ins.kind:
.f32: exp2_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: exp2_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.log:
match ins.kind:
.f32: log_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: log_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.log2:
match ins.kind:
.f32: log2_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: log2_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.log10:
match ins.kind:
.f32: log10_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: log10_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.sin:
match ins.kind:
.f32: sin_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: sin_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.cos:
match ins.kind:
.f32: cos_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: cos_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.tan:
match ins.kind:
.f32: tan_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: tan_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.asin:
match ins.kind:
.f32: asin_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: asin_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.acos:
match ins.kind:
.f32: acos_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: acos_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.atan:
match ins.kind:
.f32: atan_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: atan_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.sinh:
match ins.kind:
.f32: sinh_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: sinh_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.cosh:
match ins.kind:
.f32: cosh_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: cosh_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.tanh:
match ins.kind:
.f32: tanh_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: tanh_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.floor:
match ins.kind:
.f32: floor_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: floor_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.ceil:
match ins.kind:
.f32: ceil_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: ceil_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.round:
match ins.kind:
.f32: round_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: round_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.trunc:
match ins.kind:
.f32: trunc_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: trunc_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.sign:
match ins.kind:
.f32: sign_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: sign_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.frac:
match ins.kind:
.f32: frac_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: frac_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.degrees:
match ins.kind:
.f32: degrees_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: degrees_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.radians:
match ins.kind:
.f32: radians_f32(w.f32s(ins.dst), w.f32s(ins.a))
.f64: radians_f64(w.f64s(ins.dst), w.f64s(ins.a))
_: return w.n
.isnan:
match ins.kind:
.f32: isnan_f32(w.masks(ins.dst), w.f32s(ins.a))
.f64: isnan_f64(w.masks(ins.dst), w.f64s(ins.a))
_: return w.n
.isinf:
match ins.kind:
.f32: isinf_f32(w.masks(ins.dst), w.f32s(ins.a))
.f64: isinf_f64(w.masks(ins.dst), w.f64s(ins.a))
_: return w.n
.isfinite:
match ins.kind:
.f32: isfinite_f32(w.masks(ins.dst), w.f32s(ins.a))
.f64: isfinite_f64(w.masks(ins.dst), w.f64s(ins.a))
_: return w.n
_: return w.n
return w.n
func neg_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -a[i]
func neg_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -a[i]
func neg_i32(d: i32[], a: const i32[], on: Active) -> usize:
for i in 0..<d.length:
if a[i] == (-2147483648):
if on.at(i): return i
d[i] = a[i]
else:
d[i] = -a[i]
return d.length
func neg_i64(d: i64[], a: const i64[], on: Active) -> usize:
for i in 0..<d.length:
if a[i] == (-9223372036854775808):
if on.at(i): return i
d[i] = a[i]
else:
d[i] = -a[i]
return d.length
func wneg_i32(d: i32[], a: const i32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -%a[i]
func wneg_i64(d: i64[], a: const i64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -%a[i]
func wneg_u32(d: u32[], a: const u32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -%a[i]
func wneg_u64(d: u64[], a: const u64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = -%a[i]
func abs_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.abs(a[i])
func abs_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.abs(a[i])
func abs_i32(d: i32[], a: const i32[], on: Active) -> usize:
for i in 0..<d.length:
if a[i] == (-2147483648):
if on.at(i): return i
d[i] = a[i]
else:
d[i] = a[i] if a[i] >= 0 else -a[i]
return d.length
func abs_i64(d: i64[], a: const i64[], on: Active) -> usize:
for i in 0..<d.length:
if a[i] == (-9223372036854775808):
if on.at(i): return i
d[i] = a[i]
else:
d[i] = a[i] if a[i] >= 0 else -a[i]
return d.length
func bnot_i32(d: i32[], a: const i32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = ~a[i]
func bnot_i64(d: i64[], a: const i64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = ~a[i]
func bnot_u32(d: u32[], a: const u32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = ~a[i]
func bnot_u64(d: u64[], a: const u64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = ~a[i]
func bnot_mask(d: u8[], a: const u8[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] ^ 1
func sqrt_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.sqrt(a[i])
func sqrt_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.sqrt(a[i])
func cbrt_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.cbrt(a[i])
func cbrt_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.cbrt(a[i])
func exp_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.exp(a[i])
func exp_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.exp(a[i])
func exp2_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.exp2(a[i])
func exp2_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.exp2(a[i])
func log_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.log(a[i])
func log_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.log(a[i])
func log2_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.log2(a[i])
func log2_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.log2(a[i])
func log10_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.log10(a[i])
func log10_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.log10(a[i])
func sin_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.sin(a[i])
func sin_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.sin(a[i])
func cos_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.cos(a[i])
func cos_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.cos(a[i])
func tan_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.tan(a[i])
func tan_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.tan(a[i])
func asin_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.asin(a[i])
func asin_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.asin(a[i])
func acos_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.acos(a[i])
func acos_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.acos(a[i])
func atan_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.atan(a[i])
func atan_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.atan(a[i])
func sinh_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.sinh(a[i])
func sinh_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.sinh(a[i])
func cosh_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.cosh(a[i])
func cosh_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.cosh(a[i])
func tanh_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.tanh(a[i])
func tanh_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.tanh(a[i])
func floor_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.floor(a[i])
func floor_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.floor(a[i])
func ceil_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.ceil(a[i])
func ceil_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.ceil(a[i])
func round_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.round(a[i])
func round_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.round(a[i])
func trunc_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.trunc(a[i])
func trunc_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.trunc(a[i])
func sign_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math32.sign(a[i])
func sign_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = math.sign(a[i])
func frac_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] - math32.floor(a[i])
func frac_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] - math.floor(a[i])
func degrees_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] * (180.0 / math32.pi)
func degrees_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] * (180.0 / math.pi)
func radians_f32(d: f32[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] * (math32.pi / 180.0)
func radians_f64(d: f64[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = a[i] * (math.pi / 180.0)
func isnan_f32(d: u8[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math32.is_nan(a[i])
func isnan_f64(d: u8[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math.is_nan(a[i])
func isinf_f32(d: u8[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math32.is_infinite(a[i])
func isinf_f64(d: u8[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math.is_infinite(a[i])
func isfinite_f32(d: u8[], a: const f32[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math32.is_finite(a[i])
func isfinite_f64(d: u8[], a: const f64[]):
assert(a.length == d.length)
for i in 0..<d.length: d[i] = (u8)math.is_finite(a[i])