Search all intrinsics
SIMD ISAReturn TypeNameArgumentsInstruction Group
Neonbfloat16_tvcvth_bf16_f32(float32_t a)Data type conversion / Conversions
Description
Floating-point convert from single-precision to BFloat16 format (scalar) converts the single-precision floating-point value in the 32-bit SIMD&FP source register to BFloat16 format and writes the result in the 16-bit SIMD&FP destination register.
Results
Hd result
This intrinsic compiles to the following instructions:

BFCVT Hd,Sn

Argument Preparation
a register: Sn
Architectures
A32, A64

Operation

CheckFPAdvSIMDEnabled64();

bits(32) operand = V[n];
FPCRType fpcr    = FPCR[];
boolean merge    = IsMerging(fpcr);
bits(128) result = if merge then V[d] else Zeros();

Elem[result, 0, 16] = FPConvertBF(operand, fpcr);

V[d] = result;