Public Target Interface

Core Builtins


Header:	`<bedrockcoreintrin.h>`

Core Intrinsics

Name	C interface	Lowering	Availability, constraint, and effect
`breakpoint`	`void(void)`	`BKPT`	Any privilege; always raises the architectural breakpoint/debug exception.
`cpuid`	`u64(u64)`	`CPUID`	User allowed; reads discovery state and not C memory.
`rdpmc`	`u64(performance_counter)`	`RDPMC`	Unprivileged; ID is a constant in 0 through 65535; zero and unavailable IDs fault with `INVALID_SELECTOR`.
`read_status`	`u16(void)`	`RDSTATUS`	Unprivileged; reads `STATUS` and not C memory.
`relax`	`void(void)`	`RELAX`	Unprivileged PAUSE-like convergent hint; retires normally and permits a finite implementation-selected delay, including zero, before subsequent execution.
`trace`	`void(u32)`	`TRACE`	Unprivileged; marker is a constant in 0 through 65535; emits a marker when enabled and otherwise acts as a NOP.
`yield`	`void(void)`	`YIELD`	Any privilege; scheduling hint, not a compiler or memory barrier.

Memory Builtins


Header:	`<bedrockmemoryintrin.h>`

Memory Intrinsics

Name	C interface	Lowering	Availability, constraint, and effect
`address_fence`	`void(void)`	`AFENCE`	Full compiler memory clobber and hardware address fence.
`nontemporal_store_u16`	`void(u16 *,u16)`	`MOVNT.W`	One writable 16-bit object; one hinted store, neither volatile nor a fence.
`nontemporal_store_u32`	`void(u32 *,u32)`	`MOVNT.L`	One writable 32-bit object; one hinted store, neither volatile nor a fence.
`nontemporal_store_u64`	`void(u64 *,u64)`	`MOVNT.Q`	One writable 64-bit object; one hinted store, neither volatile nor a fence.
`nontemporal_store_u8`	`void(u8 *,u8)`	`MOVNT.B`	One writable byte; one hinted store, neither volatile nor a fence.
`read_fence`	`void(void)`	`RFENCE`	Compiler and hardware read-ordering barrier.
`write_fence`	`void(void)`	`WFENCE`	Compiler and hardware write-ordering barrier.

Integer Builtins


Header:	`<bedrockintegerintrin.h>`

Integer Intrinsics

Name	C interface	Lowering	Availability, constraint, and effect
`clmul_u16`	`u16(u16,u16)`	`CLMUL.W`	Pure; low 16 bits of the carryless product.
`clmul_u32`	`u32(u32,u32)`	`CLMUL.L`	Pure; low 32 bits of the carryless product.
`clmul_u64`	`u64(u64,u64)`	`CLMUL.Q`	Pure; low 64 bits of the carryless product.
`clmul_u8`	`u8(u8,u8)`	`CLMUL.B`	Pure; low 8 bits of the carryless product.

Floating-Point Builtins


Header:	`<bedrockfpuintrin.h>`

Floating-Point Intrinsics

Name	C interface	Lowering	Availability, constraint, and effect
`facosa_f32`	`float(float)`	`FACOSA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`facosa_f64`	`double(double)`	`FACOSA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fasina_f32`	`float(float)`	`FASINA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fasina_f64`	`double(double)`	`FASINA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fatana_f32`	`float(float)`	`FATANA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fatana_f64`	`double(double)`	`FATANA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fatanha_f32`	`float(float)`	`FATANHA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fatanha_f64`	`double(double)`	`FATANHA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fclass_f32`	`u16(float)`	`FCLASS.S`	FPU required; pure ten-class one-hot classification; does not update `FFLAGS`.
`fclass_f64`	`u16(double)`	`FCLASS.D`	FPU required; pure ten-class one-hot classification; does not update `FFLAGS`.
`fcosa_f32`	`float(float)`	`FCOSA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fcosa_f64`	`double(double)`	`FCOSA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fcosha_f32`	`float(float)`	`FCOSHA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fcosha_f64`	`double(double)`	`FCOSHA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fetoxa_f32`	`float(float)`	`FETOXA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fetoxa_f64`	`double(double)`	`FETOXA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fetoxm1a_f32`	`float(float)`	`FETOXM1A.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fetoxm1a_f64`	`double(double)`	`FETOXM1A.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flog10a_f32`	`float(float)`	`FLOG10A.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flog10a_f64`	`double(double)`	`FLOG10A.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flog2a_f32`	`float(float)`	`FLOG2A.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flog2a_f64`	`double(double)`	`FLOG2A.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flogna_f32`	`float(float)`	`FLOGNA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flogna_f64`	`double(double)`	`FLOGNA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flognp1a_f32`	`float(float)`	`FLOGNP1A.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`flognp1a_f64`	`double(double)`	`FLOGNP1A.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fsina_f32`	`float(float)`	`FSINA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fsina_f64`	`double(double)`	`FSINA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fsincosa_f32`	`void(float,float ,float )`	`FSINCOSA.S`	FP and FPTRANSA required; writes sine through the first result pointer and cosine through the second; ISA accuracy, special-value, and floating-point event contracts apply to both results.
`fsincosa_f64`	`void(double,double ,double )`	`FSINCOSA.D`	FP and FPTRANSA required; writes sine through the first result pointer and cosine through the second; ISA accuracy, special-value, and floating-point event contracts apply to both results.
`fsinha_f32`	`float(float)`	`FSINHA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`fsinha_f64`	`double(double)`	`FSINHA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftana_f32`	`float(float)`	`FTANA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftana_f64`	`double(double)`	`FTANA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftanha_f32`	`float(float)`	`FTANHA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftanha_f64`	`double(double)`	`FTANHA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftentoxa_f32`	`float(float)`	`FTENTOXA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftentoxa_f64`	`double(double)`	`FTENTOXA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftwotoxa_f32`	`float(float)`	`FTWOTOXA.S`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`ftwotoxa_f64`	`double(double)`	`FTWOTOXA.D`	FP and FPTRANSA required; ISA accuracy, special-value, and floating-point event contracts apply.
`read_fflags`	`u16(void)`	`RDFFLAGS`	Unprivileged; reads accrued floating-point exception flags.
`read_fstatus`	`u16(void)`	`RDFSTATUS`	Unprivileged; access to the floating-point environment.
`write_fflags`	`void(u16)`	`WRFFLAGS`	Unprivileged; reserved bits follow the ISA; writes accrued flags.
`write_fstatus`	`void(u16)`	`WRFSTATUS`	Unprivileged; reserved bits follow the ISA; changes the floating-point environment.

Approximate Transcendental Lowering Policy

A strict language-library operation shall not be lowered to an FPTRANSA instruction. A compiler may select FPTRANSA only when the source uses a separately defined approximate interface or an active compiler mode permits approximate mathematics. The selected instruction remains subject to its ISA accuracy, special-value, and floating-point exception-condition contract.