NOTE: The intrinsics described in this document (usually called ve intrinsics) were deprecated in the LLVM-VE compiler som time during 2020. They have been replaced by variants that include in each vector instruction the vector length for the instruction described by the intrinsic: the vel intrinsics. Carrying the vector length along with the intrinsic allows for reordering the instructions to some extent and generate automatically instructions that load the proper vector length. The new list can be found here: https://sx-aurora-dev.github.io/velintrin.html .
This document contains the list of vector intrinsics that were added to the LLVM backend for the SX-Aurora TSUBASA vector engine. The proprietary ncc compiler from NEC is not exposing vector instrinsics, there the focus is on auto-vectorization and optimization, thus intrinsics are actually not needed.
The intrinsics in this document resemble closely the vector ISA of the SX-Aurora VE. LLVM does not (yet) support properly predication (vector masks) and the variable vector length, there are efforts underway for supporting these (SVE/EVL, VP) and making them some day first-class entities.
0. Vector Transfer Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VLD | __vr _ve_vld_vss(unsigned long int sy, void const* sz) | vld | |
| VLD | __vr _ve_vld_vss(unsigned long int I, void const* sz) | vld | |
| VLDU | __vr _ve_vldu_vss(unsigned long int sy, void const* sz) | vldu | |
| VLDU | __vr _ve_vldu_vss(unsigned long int I, void const* sz) | vldu | |
| VLDL | __vr _ve_vldlsx_vss(unsigned long int sy, void const* sz) | vldl.sx | |
| VLDL | __vr _ve_vldlsx_vss(unsigned long int I, void const* sz) | vldl.sx | |
| VLDL | __vr _ve_vldlzx_vss(unsigned long int sy, void const* sz) | vldl.zx | |
| VLDL | __vr _ve_vldlzx_vss(unsigned long int I, void const* sz) | vldl.zx | |
| VLD2D | __vr _ve_vld2d_vss(unsigned long int sy, void const* sz) | vld2d | |
| VLD2D | __vr _ve_vld2d_vss(unsigned long int I, void const* sz) | vld2d | |
| VLDU2D | __vr _ve_vldu2d_vss(unsigned long int sy, void const* sz) | vldu2d | |
| VLDU2D | __vr _ve_vldu2d_vss(unsigned long int I, void const* sz) | vldu2d | |
| VLDL2D | __vr _ve_vldl2dsx_vss(unsigned long int sy, void const* sz) | vldl2d.sx | |
| VLDL2D | __vr _ve_vldl2dsx_vss(unsigned long int I, void const* sz) | vldl2d.sx | |
| VLDL2D | __vr _ve_vldl2dzx_vss(unsigned long int sy, void const* sz) | vldl2d.zx | |
| VLDL2D | __vr _ve_vldl2dzx_vss(unsigned long int I, void const* sz) | vldl2d.zx | |
| VST | void _ve_vst_vss(__vr vx, unsigned long int sy, void* sz) | vst | |
| VST | void _ve_vst_vss(__vr vx, unsigned long int I, void* sz) | vst | |
| VST | void _ve_vstot_vss(__vr vx, unsigned long int sy, void* sz) | vst.ot | |
| VST | void _ve_vstot_vss(__vr vx, unsigned long int I, void* sz) | vst.ot | |
| VSTU | void _ve_vstu_vss(__vr vx, unsigned long int sy, void* sz) | vstu | |
| VSTU | void _ve_vstu_vss(__vr vx, unsigned long int I, void* sz) | vstu | |
| VSTU | void _ve_vstuot_vss(__vr vx, unsigned long int sy, void* sz) | vstu.ot | |
| VSTU | void _ve_vstuot_vss(__vr vx, unsigned long int I, void* sz) | vstu.ot | |
| VSTL | void _ve_vstl_vss(__vr vx, unsigned long int sy, void* sz) | vstl | |
| VSTL | void _ve_vstl_vss(__vr vx, unsigned long int I, void* sz) | vstl | |
| VSTL | void _ve_vstlot_vss(__vr vx, unsigned long int sy, void* sz) | vstl.ot | |
| VSTL | void _ve_vstlot_vss(__vr vx, unsigned long int I, void* sz) | vstl.ot | |
| VST2D | void _ve_vst2d_vss(__vr vx, unsigned long int sy, void* sz) | vst2d | |
| VST2D | void _ve_vst2d_vss(__vr vx, unsigned long int I, void* sz) | vst2d | |
| VST2D | void _ve_vst2dot_vss(__vr vx, unsigned long int sy, void* sz) | vst2d.ot | |
| VST2D | void _ve_vst2dot_vss(__vr vx, unsigned long int I, void* sz) | vst2d.ot | |
| VSTU2D | void _ve_vstu2d_vss(__vr vx, unsigned long int sy, void* sz) | vstu2d | |
| VSTU2D | void _ve_vstu2d_vss(__vr vx, unsigned long int I, void* sz) | vstu2d | |
| VSTU2D | void _ve_vstu2dot_vss(__vr vx, unsigned long int sy, void* sz) | vstu2d.ot | |
| VSTU2D | void _ve_vstu2dot_vss(__vr vx, unsigned long int I, void* sz) | vstu2d.ot | |
| VSTL2D | void _ve_vstl2d_vss(__vr vx, unsigned long int sy, void* sz) | vstl2d | |
| VSTL2D | void _ve_vstl2d_vss(__vr vx, unsigned long int I, void* sz) | vstl2d | |
| VSTL2D | void _ve_vstl2dot_vss(__vr vx, unsigned long int sy, void* sz) | vstl2d.ot | |
| VSTL2D | void _ve_vstl2dot_vss(__vr vx, unsigned long int I, void* sz) | vstl2d.ot | |
| PFCHV | void _ve_pfchv(long int sy, void const* sz) | pfchv | |
| PFCHV | void _ve_pfchv(long int I, void const* sz) | pfchv | |
| LSV | __vr _ve_lsv_vvss(__vr vx, unsigned int sy, unsigned long int sz) | lsv | |
| LVS | unsigned long int _ve_lvs_svs_u64(__vr vx, unsigned int sy) | lvs | |
| LVS | double _ve_lvs_svs_f64(__vr vx, unsigned int sy) | lvs | |
| LVS | float _ve_lvs_svs_f32(__vr vx, unsigned int sy) | lvs | |
| LVM | __vm256 _ve_lvm_mmss(__vm256 vmd, unsigned long int sy, unsigned long int sz) | lvm | |
| LVM | __vm256 _ve_lvm_mmss(__vm256 vmd, unsigned long int N, unsigned long int sz) | lvm | |
| LVM | __vm512 _ve_lvm_MMss(__vm512 vmd, unsigned long int N, unsigned long int sz) | lvm | |
| SVM | unsigned long int _ve_svm_sms(__vm256 vmz, unsigned long int sy) | svm | |
| SVM | unsigned long int _ve_svm_sms(__vm256 vmz, unsigned long int N) | svm | |
| SVM | unsigned long int _ve_svm_sMs(__vm512 vmz, unsigned long int N) | svm | |
| VBRD | __vr _ve_vbrd_vs_f64(double sy) | vbrd | vx[:] = sy |
| VBRD | __vr _ve_vbrd_vsmv_f64(double sy, __vm256 vm, __vr vd) | vbrd | vx[:] = vm[:] ? sy : vd[:] |
| VBRD | __vr _ve_vbrd_vs_i64(long int sy) | vbrd | vx[:] = sy |
| VBRD | __vr _ve_vbrd_vsmv_i64(long int sy, __vm256 vm, __vr vd) | vbrd | vx[:] = vm[:] ? sy : vd[:] |
| VBRD | __vr _ve_vbrd_vs_i64(long int I) | vbrd | vx[:] = I |
| VBRD | __vr _ve_vbrd_vsmv_i64(long int I, __vm256 vm, __vr vd) | vbrd | vx[:] = vm[:] ? I : vd[:] |
| VBRD | __vr _ve_vbrdu_vs_f32(float sy) | vbrdu | vx[:] = sy |
| VBRD | __vr _ve_vbrdu_vsmv_f32(float sy, __vm256 vm, __vr vd) | vbrdu | vx[:] = vm[:] ? sy : vd[:] |
| VBRD | __vr _ve_vbrdl_vs_i32(int sy) | vbrdl | vx[:] = sy |
| VBRD | __vr _ve_vbrdl_vsmv_i32(int sy, __vm256 vm, __vr vd) | vbrdl | vx[:] = vm[:] ? sy : vd[:] |
| VBRD | __vr _ve_vbrdl_vs_i32(int I) | vbrdl | vx[:] = I |
| VBRD | __vr _ve_vbrdl_vsmv_i32(int I, __vm256 vm, __vr vd) | vbrdl | vx[:] = vm[:] ? I : vd[:] |
| VBRD | __vr _ve_pvbrd_vs_i64(unsigned long int sy) | pvbrd | vx[:] = sy |
| VBRD | __vr _ve_pvbrd_vsMv_i64(unsigned long int sy, __vm512 vm, __vr vd) | pvbrd | vx[:] = vm[:] ? sy : vd[:] |
| VMV | __vr _ve_vmv_vsv(unsigned int sy, __vr vz) | vmv | |
| VMV | __vr _ve_vmv_vsv(unsigned int N, __vr vz) | vmv |
1. Vector Fixed-Point Arithmetic Operation Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VADD | __vr _ve_vaddul_vvv(__vr vy, __vr vz) | vaddu.l | vx[:] = vy[:] + vz[:] |
| VADD | __vr _ve_vaddul_vsv(unsigned long int sy, __vr vz) | vaddu.l | vx[:] = sy + vz[:] |
| VADD | __vr _ve_vaddul_vsv(unsigned long int I, __vr vz) | vaddu.l | vx[:] = I + vz[:] |
| VADD | __vr _ve_vaddul_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vaddu.l | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADD | __vr _ve_vaddul_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vaddu.l | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADD | __vr _ve_vaddul_vsvmv(unsigned long int I, __vr vz, __vm256 vm, __vr vd) | vaddu.l | vx[:] = vm[:] ? I + vz[:] : vd[:] |
| VADD | __vr _ve_vadduw_vvv(__vr vy, __vr vz) | vaddu.w | vx[:] = vy[:] + vz[:] |
| VADD | __vr _ve_vadduw_vsv(unsigned int sy, __vr vz) | vaddu.w | vx[:] = sy + vz[:] |
| VADD | __vr _ve_vadduw_vsv(unsigned int I, __vr vz) | vaddu.w | vx[:] = I + vz[:] |
| VADD | __vr _ve_vadduw_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vaddu.w | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADD | __vr _ve_vadduw_vsvmv(unsigned int sy, __vr vz, __vm256 vm, __vr vd) | vaddu.w | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADD | __vr _ve_vadduw_vsvmv(unsigned int I, __vr vz, __vm256 vm, __vr vd) | vaddu.w | vx[:] = vm[:] ? I + vz[:] : vd[:] |
| VADD | __vr _ve_pvaddu_vvv(__vr vy, __vr vz) | pvaddu | vx[:] = vy[:] + vz[:] |
| VADD | __vr _ve_pvaddu_vsv(unsigned long int sy, __vr vz) | pvaddu | vx[:] = sy + vz[:] |
| VADD | __vr _ve_pvaddu_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvaddu | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADD | __vr _ve_pvaddu_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvaddu | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswsx_vvv(__vr vy, __vr vz) | vadds.w.sx | vx[:] = vy[:] + vz[:] |
| VADS | __vr _ve_vaddswsx_vsv(int sy, __vr vz) | vadds.w.sx | vx[:] = sy + vz[:] |
| VADS | __vr _ve_vaddswsx_vsv(int I, __vr vz) | vadds.w.sx | vx[:] = I + vz[:] |
| VADS | __vr _ve_vaddswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vadds.w.sx | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vadds.w.sx | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vadds.w.sx | vx[:] = vm[:] ? I + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswzx_vvv(__vr vy, __vr vz) | vadds.w.zx | vx[:] = vy[:] + vz[:] |
| VADS | __vr _ve_vaddswzx_vsv(int sy, __vr vz) | vadds.w.zx | vx[:] = sy + vz[:] |
| VADS | __vr _ve_vaddswzx_vsv(int I, __vr vz) | vadds.w.zx | vx[:] = I + vz[:] |
| VADS | __vr _ve_vaddswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vadds.w.zx | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vadds.w.zx | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADS | __vr _ve_vaddswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vadds.w.zx | vx[:] = vm[:] ? I + vz[:] : vd[:] |
| VADS | __vr _ve_pvadds_vvv(__vr vy, __vr vz) | pvadds | vx[:] = vy[:] + vz[:] |
| VADS | __vr _ve_pvadds_vsv(unsigned long int sy, __vr vz) | pvadds | vx[:] = sy + vz[:] |
| VADS | __vr _ve_pvadds_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvadds | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADS | __vr _ve_pvadds_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvadds | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADX | __vr _ve_vaddsl_vvv(__vr vy, __vr vz) | vadds.l | vx[:] = vy[:] + vz[:] |
| VADX | __vr _ve_vaddsl_vsv(long int sy, __vr vz) | vadds.l | vx[:] = sy + vz[:] |
| VADX | __vr _ve_vaddsl_vsv(long int I, __vr vz) | vadds.l | vx[:] = I + vz[:] |
| VADX | __vr _ve_vaddsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vadds.l | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VADX | __vr _ve_vaddsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vadds.l | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VADX | __vr _ve_vaddsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vadds.l | vx[:] = vm[:] ? I + vz[:] : vd[:] |
| VSUB | __vr _ve_vsubul_vvv(__vr vy, __vr vz) | vsubu.l | vx[:] = vy[:] - vz[:] |
| VSUB | __vr _ve_vsubul_vsv(unsigned long int sy, __vr vz) | vsubu.l | vx[:] = sy - vz[:] |
| VSUB | __vr _ve_vsubul_vsv(unsigned long int I, __vr vz) | vsubu.l | vx[:] = I - vz[:] |
| VSUB | __vr _ve_vsubul_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vsubu.l | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSUB | __vr _ve_vsubul_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vsubu.l | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSUB | __vr _ve_vsubul_vsvmv(unsigned long int I, __vr vz, __vm256 vm, __vr vd) | vsubu.l | vx[:] = vm[:] ? I - vz[:] : vd[:] |
| VSUB | __vr _ve_vsubuw_vvv(__vr vy, __vr vz) | vsubu.w | vx[:] = vy[:] - vz[:] |
| VSUB | __vr _ve_vsubuw_vsv(unsigned int sy, __vr vz) | vsubu.w | vx[:] = sy - vz[:] |
| VSUB | __vr _ve_vsubuw_vsv(unsigned int I, __vr vz) | vsubu.w | vx[:] = I - vz[:] |
| VSUB | __vr _ve_vsubuw_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vsubu.w | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSUB | __vr _ve_vsubuw_vsvmv(unsigned int sy, __vr vz, __vm256 vm, __vr vd) | vsubu.w | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSUB | __vr _ve_vsubuw_vsvmv(unsigned int I, __vr vz, __vm256 vm, __vr vd) | vsubu.w | vx[:] = vm[:] ? I - vz[:] : vd[:] |
| VSUB | __vr _ve_pvsubu_vvv(__vr vy, __vr vz) | pvsubu | vx[:] = vy[:] - vz[:] |
| VSUB | __vr _ve_pvsubu_vsv(unsigned long int sy, __vr vz) | pvsubu | vx[:] = sy - vz[:] |
| VSUB | __vr _ve_pvsubu_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvsubu | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSUB | __vr _ve_pvsubu_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvsubu | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswsx_vvv(__vr vy, __vr vz) | vsubs.w.sx | vx[:] = vy[:] - vz[:] |
| VSBS | __vr _ve_vsubswsx_vsv(int sy, __vr vz) | vsubs.w.sx | vx[:] = sy - vz[:] |
| VSBS | __vr _ve_vsubswsx_vsv(int I, __vr vz) | vsubs.w.sx | vx[:] = I - vz[:] |
| VSBS | __vr _ve_vsubswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vsubs.w.sx | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vsubs.w.sx | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vsubs.w.sx | vx[:] = vm[:] ? I - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswzx_vvv(__vr vy, __vr vz) | vsubs.w.zx | vx[:] = vy[:] - vz[:] |
| VSBS | __vr _ve_vsubswzx_vsv(int sy, __vr vz) | vsubs.w.zx | vx[:] = sy - vz[:] |
| VSBS | __vr _ve_vsubswzx_vsv(int I, __vr vz) | vsubs.w.zx | vx[:] = I - vz[:] |
| VSBS | __vr _ve_vsubswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vsubs.w.zx | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vsubs.w.zx | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSBS | __vr _ve_vsubswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vsubs.w.zx | vx[:] = vm[:] ? I - vz[:] : vd[:] |
| VSBS | __vr _ve_pvsubs_vvv(__vr vy, __vr vz) | pvsubs | vx[:] = vy[:] - vz[:] |
| VSBS | __vr _ve_pvsubs_vsv(unsigned long int sy, __vr vz) | pvsubs | vx[:] = sy - vz[:] |
| VSBS | __vr _ve_pvsubs_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvsubs | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSBS | __vr _ve_pvsubs_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvsubs | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSBX | __vr _ve_vsubsl_vvv(__vr vy, __vr vz) | vsubs.l | vx[:] = vy[:] - vz[:] |
| VSBX | __vr _ve_vsubsl_vsv(long int sy, __vr vz) | vsubs.l | vx[:] = sy - vz[:] |
| VSBX | __vr _ve_vsubsl_vsv(long int I, __vr vz) | vsubs.l | vx[:] = I - vz[:] |
| VSBX | __vr _ve_vsubsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vsubs.l | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VSBX | __vr _ve_vsubsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vsubs.l | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VSBX | __vr _ve_vsubsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vsubs.l | vx[:] = vm[:] ? I - vz[:] : vd[:] |
| VMPY | __vr _ve_vmulul_vvv(__vr vy, __vr vz) | vmulu.l | vx[:] = vy[:] * vz[:] |
| VMPY | __vr _ve_vmulul_vsv(unsigned long int sy, __vr vz) | vmulu.l | vx[:] = sy * vz[:] |
| VMPY | __vr _ve_vmulul_vsv(unsigned long int I, __vr vz) | vmulu.l | vx[:] = I * vz[:] |
| VMPY | __vr _ve_vmulul_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmulu.l | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VMPY | __vr _ve_vmulul_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vmulu.l | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VMPY | __vr _ve_vmulul_vsvmv(unsigned long int I, __vr vz, __vm256 vm, __vr vd) | vmulu.l | vx[:] = vm[:] ? I * vz[:] : vd[:] |
| VMPY | __vr _ve_vmuluw_vvv(__vr vy, __vr vz) | vmulu.w | vx[:] = vy[:] * vz[:] |
| VMPY | __vr _ve_vmuluw_vsv(unsigned int sy, __vr vz) | vmulu.w | vx[:] = sy * vz[:] |
| VMPY | __vr _ve_vmuluw_vsv(unsigned int I, __vr vz) | vmulu.w | vx[:] = I * vz[:] |
| VMPY | __vr _ve_vmuluw_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmulu.w | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VMPY | __vr _ve_vmuluw_vsvmv(unsigned int sy, __vr vz, __vm256 vm, __vr vd) | vmulu.w | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VMPY | __vr _ve_vmuluw_vsvmv(unsigned int I, __vr vz, __vm256 vm, __vr vd) | vmulu.w | vx[:] = vm[:] ? I * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswsx_vvv(__vr vy, __vr vz) | vmuls.w.sx | vx[:] = vy[:] * vz[:] |
| VMPS | __vr _ve_vmulswsx_vsv(int sy, __vr vz) | vmuls.w.sx | vx[:] = sy * vz[:] |
| VMPS | __vr _ve_vmulswsx_vsv(int I, __vr vz) | vmuls.w.sx | vx[:] = I * vz[:] |
| VMPS | __vr _ve_vmulswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmuls.w.sx | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmuls.w.sx | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmuls.w.sx | vx[:] = vm[:] ? I * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswzx_vvv(__vr vy, __vr vz) | vmuls.w.zx | vx[:] = vy[:] * vz[:] |
| VMPS | __vr _ve_vmulswzx_vsv(int sy, __vr vz) | vmuls.w.zx | vx[:] = sy * vz[:] |
| VMPS | __vr _ve_vmulswzx_vsv(int I, __vr vz) | vmuls.w.zx | vx[:] = I * vz[:] |
| VMPS | __vr _ve_vmulswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmuls.w.zx | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmuls.w.zx | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VMPS | __vr _ve_vmulswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmuls.w.zx | vx[:] = vm[:] ? I * vz[:] : vd[:] |
| VMPX | __vr _ve_vmulsl_vvv(__vr vy, __vr vz) | vmuls.l | vx[:] = vy[:] * vz[:] |
| VMPX | __vr _ve_vmulsl_vsv(long int sy, __vr vz) | vmuls.l | vx[:] = sy * vz[:] |
| VMPX | __vr _ve_vmulsl_vsv(long int I, __vr vz) | vmuls.l | vx[:] = I * vz[:] |
| VMPX | __vr _ve_vmulsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmuls.l | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VMPX | __vr _ve_vmulsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vmuls.l | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VMPX | __vr _ve_vmulsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vmuls.l | vx[:] = vm[:] ? I * vz[:] : vd[:] |
| VMPD | __vr _ve_vmulslw_vvv(__vr vy, __vr vz) | vmuls.l.w | vx[:] = vy[:] * vz[:] |
| VMPD | __vr _ve_vmulslw_vsv(int sy, __vr vz) | vmuls.l.w | vx[:] = sy * vz[:] |
| VMPD | __vr _ve_vmulslw_vsv(int I, __vr vz) | vmuls.l.w | vx[:] = I * vz[:] |
| VDIV | __vr _ve_vdivul_vvv(__vr vy, __vr vz) | vdivu.l | vx[:] = vy[:] / vz[:] |
| VDIV | __vr _ve_vdivul_vsv(unsigned long int sy, __vr vz) | vdivu.l | vx[:] = sy / vz[:] |
| VDIV | __vr _ve_vdivul_vsv(unsigned long int I, __vr vz) | vdivu.l | vx[:] = I / vz[:] |
| VDIV | __vr _ve_vdivul_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vdivu.l | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivul_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vdivu.l | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivul_vsvmv(unsigned long int I, __vr vz, __vm256 vm, __vr vd) | vdivu.l | vx[:] = vm[:] ? I / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivuw_vvv(__vr vy, __vr vz) | vdivu.w | vx[:] = vy[:] / vz[:] |
| VDIV | __vr _ve_vdivuw_vsv(unsigned int sy, __vr vz) | vdivu.w | vx[:] = sy / vz[:] |
| VDIV | __vr _ve_vdivuw_vsv(unsigned int I, __vr vz) | vdivu.w | vx[:] = I / vz[:] |
| VDIV | __vr _ve_vdivuw_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vdivu.w | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivuw_vsvmv(unsigned int sy, __vr vz, __vm256 vm, __vr vd) | vdivu.w | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivuw_vsvmv(unsigned int I, __vr vz, __vm256 vm, __vr vd) | vdivu.w | vx[:] = vm[:] ? I / vz[:] : vd[:] |
| VDIV | __vr _ve_vdivul_vvs(__vr vy, unsigned long int sy) | vdivu.l | vx[:] = vy[:] / sy |
| VDIV | __vr _ve_vdivul_vvs(__vr vy, unsigned long int I) | vdivu.l | vx[:] = vy[:] / I |
| VDIV | __vr _ve_vdivul_vvsmv(__vr vy, unsigned long int sy, __vm256 vm, __vr vd) | vdivu.l | vx[:] = vm[:] ? vy[:] / sy : vd[:] |
| VDIV | __vr _ve_vdivul_vvsmv(__vr vy, unsigned long int I, __vm256 vm, __vr vd) | vdivu.l | vx[:] = vm[:] ? vy[:] / I : vd[:] |
| VDIV | __vr _ve_vdivuw_vvs(__vr vy, unsigned int sy) | vdivu.w | vx[:] = vy[:] / sy |
| VDIV | __vr _ve_vdivuw_vvs(__vr vy, unsigned int I) | vdivu.w | vx[:] = vy[:] / I |
| VDIV | __vr _ve_vdivuw_vvsmv(__vr vy, unsigned int sy, __vm256 vm, __vr vd) | vdivu.w | vx[:] = vm[:] ? vy[:] / sy : vd[:] |
| VDIV | __vr _ve_vdivuw_vvsmv(__vr vy, unsigned int I, __vm256 vm, __vr vd) | vdivu.w | vx[:] = vm[:] ? vy[:] / I : vd[:] |
| VDVS | __vr _ve_vdivswsx_vvv(__vr vy, __vr vz) | vdivs.w.sx | vx[:] = vy[:] / vz[:] |
| VDVS | __vr _ve_vdivswsx_vsv(int sy, __vr vz) | vdivs.w.sx | vx[:] = sy / vz[:] |
| VDVS | __vr _ve_vdivswsx_vsv(int I, __vr vz) | vdivs.w.sx | vx[:] = I / vz[:] |
| VDVS | __vr _ve_vdivswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vdivs.w.sx | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vdivs.w.sx | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vdivs.w.sx | vx[:] = vm[:] ? I / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswzx_vvv(__vr vy, __vr vz) | vdivs.w.zx | vx[:] = vy[:] / vz[:] |
| VDVS | __vr _ve_vdivswzx_vsv(int sy, __vr vz) | vdivs.w.zx | vx[:] = sy / vz[:] |
| VDVS | __vr _ve_vdivswzx_vsv(int I, __vr vz) | vdivs.w.zx | vx[:] = I / vz[:] |
| VDVS | __vr _ve_vdivswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vdivs.w.zx | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vdivs.w.zx | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vdivs.w.zx | vx[:] = vm[:] ? I / vz[:] : vd[:] |
| VDVS | __vr _ve_vdivswsx_vvs(__vr vy, int sy) | vdivs.w.sx | vx[:] = vy[:] / sy |
| VDVS | __vr _ve_vdivswsx_vvs(__vr vy, int I) | vdivs.w.sx | vx[:] = vy[:] / I |
| VDVS | __vr _ve_vdivswsx_vvsmv(__vr vy, int sy, __vm256 vm, __vr vd) | vdivs.w.sx | vx[:] = vm[:] ? vy[:] / sy : vd[:] |
| VDVS | __vr _ve_vdivswsx_vvsmv(__vr vy, int I, __vm256 vm, __vr vd) | vdivs.w.sx | vx[:] = vm[:] ? vy[:] / I : vd[:] |
| VDVS | __vr _ve_vdivswzx_vvs(__vr vy, int sy) | vdivs.w.zx | vx[:] = vy[:] / sy |
| VDVS | __vr _ve_vdivswzx_vvs(__vr vy, int I) | vdivs.w.zx | vx[:] = vy[:] / I |
| VDVS | __vr _ve_vdivswzx_vvsmv(__vr vy, int sy, __vm256 vm, __vr vd) | vdivs.w.zx | vx[:] = vm[:] ? vy[:] / sy : vd[:] |
| VDVS | __vr _ve_vdivswzx_vvsmv(__vr vy, int I, __vm256 vm, __vr vd) | vdivs.w.zx | vx[:] = vm[:] ? vy[:] / I : vd[:] |
| VDVX | __vr _ve_vdivsl_vvv(__vr vy, __vr vz) | vdivs.l | vx[:] = vy[:] / vz[:] |
| VDVX | __vr _ve_vdivsl_vsv(long int sy, __vr vz) | vdivs.l | vx[:] = sy / vz[:] |
| VDVX | __vr _ve_vdivsl_vsv(long int I, __vr vz) | vdivs.l | vx[:] = I / vz[:] |
| VDVX | __vr _ve_vdivsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vdivs.l | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VDVX | __vr _ve_vdivsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vdivs.l | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VDVX | __vr _ve_vdivsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vdivs.l | vx[:] = vm[:] ? I / vz[:] : vd[:] |
| VDVX | __vr _ve_vdivsl_vvs(__vr vy, long int sy) | vdivs.l | vx[:] = vy[:] / sy |
| VDVX | __vr _ve_vdivsl_vvs(__vr vy, long int I) | vdivs.l | vx[:] = vy[:] / I |
| VDVX | __vr _ve_vdivsl_vvsmv(__vr vy, long int sy, __vm256 vm, __vr vd) | vdivs.l | vx[:] = vm[:] ? vy[:] / sy : vd[:] |
| VDVX | __vr _ve_vdivsl_vvsmv(__vr vy, long int I, __vm256 vm, __vr vd) | vdivs.l | vx[:] = vm[:] ? vy[:] / I : vd[:] |
| VCMP | __vr _ve_vcmpul_vvv(__vr vy, __vr vz) | vcmpu.l | vx[:] = compare(vy[:], vz[:]) |
| VCMP | __vr _ve_vcmpul_vsv(unsigned long int sy, __vr vz) | vcmpu.l | vx[:] = compare(sy, vz[:]) |
| VCMP | __vr _ve_vcmpul_vsv(unsigned long int I, __vr vz) | vcmpu.l | vx[:] = compare(I, vz[:]) |
| VCMP | __vr _ve_vcmpul_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vcmpu.l | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCMP | __vr _ve_vcmpul_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vcmpu.l | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCMP | __vr _ve_vcmpul_vsvmv(unsigned long int I, __vr vz, __vm256 vm, __vr vd) | vcmpu.l | vx[:] = vm[:] ? compare(I, vz[:]) : vd[:] |
| VCMP | __vr _ve_vcmpuw_vvv(__vr vy, __vr vz) | vcmpu.w | vx[:] = compare(vy[:], vz[:]) |
| VCMP | __vr _ve_vcmpuw_vsv(unsigned int sy, __vr vz) | vcmpu.w | vx[:] = compare(sy, vz[:]) |
| VCMP | __vr _ve_vcmpuw_vsv(unsigned int I, __vr vz) | vcmpu.w | vx[:] = compare(I, vz[:]) |
| VCMP | __vr _ve_vcmpuw_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vcmpu.w | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCMP | __vr _ve_vcmpuw_vsvmv(unsigned int sy, __vr vz, __vm256 vm, __vr vd) | vcmpu.w | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCMP | __vr _ve_vcmpuw_vsvmv(unsigned int I, __vr vz, __vm256 vm, __vr vd) | vcmpu.w | vx[:] = vm[:] ? compare(I, vz[:]) : vd[:] |
| VCMP | __vr _ve_pvcmpu_vvv(__vr vy, __vr vz) | pvcmpu | vx[:] = compare(vy[:], vz[:]) |
| VCMP | __vr _ve_pvcmpu_vsv(unsigned long int sy, __vr vz) | pvcmpu | vx[:] = compare(sy, vz[:]) |
| VCMP | __vr _ve_pvcmpu_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvcmpu | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCMP | __vr _ve_pvcmpu_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvcmpu | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswsx_vvv(__vr vy, __vr vz) | vcmps.w.sx | vx[:] = compare(vy[:], vz[:]) |
| VCPS | __vr _ve_vcmpswsx_vsv(int sy, __vr vz) | vcmps.w.sx | vx[:] = compare(sy, vz[:]) |
| VCPS | __vr _ve_vcmpswsx_vsv(int I, __vr vz) | vcmps.w.sx | vx[:] = compare(I, vz[:]) |
| VCPS | __vr _ve_vcmpswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vcmps.w.sx | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vcmps.w.sx | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vcmps.w.sx | vx[:] = vm[:] ? compare(I, vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswzx_vvv(__vr vy, __vr vz) | vcmps.w.zx | vx[:] = compare(vy[:], vz[:]) |
| VCPS | __vr _ve_vcmpswzx_vsv(int sy, __vr vz) | vcmps.w.zx | vx[:] = compare(sy, vz[:]) |
| VCPS | __vr _ve_vcmpswzx_vsv(int I, __vr vz) | vcmps.w.zx | vx[:] = compare(I, vz[:]) |
| VCPS | __vr _ve_vcmpswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vcmps.w.zx | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vcmps.w.zx | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCPS | __vr _ve_vcmpswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vcmps.w.zx | vx[:] = vm[:] ? compare(I, vz[:]) : vd[:] |
| VCPS | __vr _ve_pvcmps_vvv(__vr vy, __vr vz) | pvcmps | vx[:] = compare(vy[:], vz[:]) |
| VCPS | __vr _ve_pvcmps_vsv(unsigned long int sy, __vr vz) | pvcmps | vx[:] = compare(sy, vz[:]) |
| VCPS | __vr _ve_pvcmps_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvcmps | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCPS | __vr _ve_pvcmps_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvcmps | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCPX | __vr _ve_vcmpsl_vvv(__vr vy, __vr vz) | vcmps.l | vx[:] = compare(vy[:], vz[:]) |
| VCPX | __vr _ve_vcmpsl_vsv(long int sy, __vr vz) | vcmps.l | vx[:] = compare(sy, vz[:]) |
| VCPX | __vr _ve_vcmpsl_vsv(long int I, __vr vz) | vcmps.l | vx[:] = compare(I, vz[:]) |
| VCPX | __vr _ve_vcmpsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vcmps.l | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VCPX | __vr _ve_vcmpsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vcmps.l | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VCPX | __vr _ve_vcmpsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vcmps.l | vx[:] = vm[:] ? compare(I, vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswsx_vvv(__vr vy, __vr vz) | vmaxs.w.sx | vx[:] = max(vy[:], vz[:]) |
| VCMS | __vr _ve_vmaxswsx_vsv(int sy, __vr vz) | vmaxs.w.sx | vx[:] = max(sy, vz[:]) |
| VCMS | __vr _ve_vmaxswsx_vsv(int I, __vr vz) | vmaxs.w.sx | vx[:] = max(I, vz[:]) |
| VCMS | __vr _ve_vmaxswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.sx | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.sx | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.sx | vx[:] = vm[:] ? max(I, vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswzx_vvv(__vr vy, __vr vz) | vmaxs.w.zx | vx[:] = max(vy[:], vz[:]) |
| VCMS | __vr _ve_vmaxswzx_vsv(int sy, __vr vz) | vmaxs.w.zx | vx[:] = max(sy, vz[:]) |
| VCMS | __vr _ve_vmaxswzx_vsv(int I, __vr vz) | vmaxs.w.zx | vx[:] = max(I, vz[:]) |
| VCMS | __vr _ve_vmaxswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.zx | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.zx | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VCMS | __vr _ve_vmaxswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmaxs.w.zx | vx[:] = vm[:] ? max(I, vz[:]) : vd[:] |
| VCMS | __vr _ve_pvmaxs_vvv(__vr vy, __vr vz) | pvmaxs | vx[:] = max(vy[:], vz[:]) |
| VCMS | __vr _ve_pvmaxs_vsv(unsigned long int sy, __vr vz) | pvmaxs | vx[:] = max(sy, vz[:]) |
| VCMS | __vr _ve_pvmaxs_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvmaxs | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_pvmaxs_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvmaxs | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswsx_vvv(__vr vy, __vr vz) | vmins.w.sx | vx[:] = min(vy[:], vz[:]) |
| VCMS | __vr _ve_vminswsx_vsv(int sy, __vr vz) | vmins.w.sx | vx[:] = min(sy, vz[:]) |
| VCMS | __vr _ve_vminswsx_vsv(int I, __vr vz) | vmins.w.sx | vx[:] = min(I, vz[:]) |
| VCMS | __vr _ve_vminswsx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmins.w.sx | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswsx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmins.w.sx | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswsx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmins.w.sx | vx[:] = vm[:] ? min(I, vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswzx_vvv(__vr vy, __vr vz) | vmins.w.zx | vx[:] = min(vy[:], vz[:]) |
| VCMS | __vr _ve_vminswzx_vsv(int sy, __vr vz) | vmins.w.zx | vx[:] = min(sy, vz[:]) |
| VCMS | __vr _ve_vminswzx_vsv(int I, __vr vz) | vmins.w.zx | vx[:] = min(I, vz[:]) |
| VCMS | __vr _ve_vminswzx_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmins.w.zx | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswzx_vsvmv(int sy, __vr vz, __vm256 vm, __vr vd) | vmins.w.zx | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VCMS | __vr _ve_vminswzx_vsvmv(int I, __vr vz, __vm256 vm, __vr vd) | vmins.w.zx | vx[:] = vm[:] ? min(I, vz[:]) : vd[:] |
| VCMS | __vr _ve_pvmins_vvv(__vr vy, __vr vz) | pvmins | vx[:] = min(vy[:], vz[:]) |
| VCMS | __vr _ve_pvmins_vsv(unsigned long int sy, __vr vz) | pvmins | vx[:] = min(sy, vz[:]) |
| VCMS | __vr _ve_pvmins_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvmins | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VCMS | __vr _ve_pvmins_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvmins | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VCMX | __vr _ve_vmaxsl_vvv(__vr vy, __vr vz) | vmaxs.l | vx[:] = max(vy[:], vz[:]) |
| VCMX | __vr _ve_vmaxsl_vsv(long int sy, __vr vz) | vmaxs.l | vx[:] = max(sy, vz[:]) |
| VCMX | __vr _ve_vmaxsl_vsv(long int I, __vr vz) | vmaxs.l | vx[:] = max(I, vz[:]) |
| VCMX | __vr _ve_vmaxsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmaxs.l | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VCMX | __vr _ve_vmaxsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vmaxs.l | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VCMX | __vr _ve_vmaxsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vmaxs.l | vx[:] = vm[:] ? max(I, vz[:]) : vd[:] |
| VCMX | __vr _ve_vminsl_vvv(__vr vy, __vr vz) | vmins.l | vx[:] = min(vy[:], vz[:]) |
| VCMX | __vr _ve_vminsl_vsv(long int sy, __vr vz) | vmins.l | vx[:] = min(sy, vz[:]) |
| VCMX | __vr _ve_vminsl_vsv(long int I, __vr vz) | vmins.l | vx[:] = min(I, vz[:]) |
| VCMX | __vr _ve_vminsl_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vmins.l | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VCMX | __vr _ve_vminsl_vsvmv(long int sy, __vr vz, __vm256 vm, __vr vd) | vmins.l | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VCMX | __vr _ve_vminsl_vsvmv(long int I, __vr vz, __vm256 vm, __vr vd) | vmins.l | vx[:] = vm[:] ? min(I, vz[:]) : vd[:] |
2. Vector Logical Arithmetic Operation Instructions
| Instruction | Function | asm | Description | |
|---|---|---|---|---|
| VAND | __vr _ve_vand_vvv(__vr vy, __vr vz) | vand | vx[:] = vy[:] & vz[:] | |
| VAND | __vr _ve_vand_vsv(unsigned long int sy, __vr vz) | vand | vx[:] = sy & vz[:] | |
| VAND | __vr _ve_vand_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vand | vx[:] = vm[:] ? vy[:] & vz[:] : vd[:] | |
| VAND | __vr _ve_vand_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vand | vx[:] = vm[:] ? sy & vz[:] : vd[:] | |
| VAND | __vr _ve_pvand_vvv(__vr vy, __vr vz) | pvand | vx[:] = vy[:] & vz[:] | |
| VAND | __vr _ve_pvand_vsv(unsigned long int sy, __vr vz) | pvand | vx[:] = sy & vz[:] | |
| VAND | __vr _ve_pvand_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvand | vx[:] = vm[:] ? vy[:] & vz[:] : vd[:] | |
| VAND | __vr _ve_pvand_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvand | vx[:] = vm[:] ? sy & vz[:] : vd[:] | |
| VOR | __vr _ve_vor_vvv(__vr vy, __vr vz) | vor | vx[:] = vy[:] | vz[:] |
| VOR | __vr _ve_vor_vsv(unsigned long int sy, __vr vz) | vor | vx[:] = sy | vz[:] |
| VOR | __vr _ve_vor_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vor | vx[:] = vm[:] ? vy[:] | vz[:] : vd[:] |
| VOR | __vr _ve_vor_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vor | vx[:] = vm[:] ? sy | vz[:] : vd[:] |
| VOR | __vr _ve_pvor_vvv(__vr vy, __vr vz) | pvor | vx[:] = vy[:] | vz[:] |
| VOR | __vr _ve_pvor_vsv(unsigned long int sy, __vr vz) | pvor | vx[:] = sy | vz[:] |
| VOR | __vr _ve_pvor_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvor | vx[:] = vm[:] ? vy[:] | vz[:] : vd[:] |
| VOR | __vr _ve_pvor_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvor | vx[:] = vm[:] ? sy | vz[:] : vd[:] |
| VXOR | __vr _ve_vxor_vvv(__vr vy, __vr vz) | vxor | vx[:] = vy[:] ^ vz[:] | |
| VXOR | __vr _ve_vxor_vsv(unsigned long int sy, __vr vz) | vxor | vx[:] = sy ^ vz[:] | |
| VXOR | __vr _ve_vxor_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vxor | vx[:] = vm[:] ? vy[:] ^ vz[:] : vd[:] | |
| VXOR | __vr _ve_vxor_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | vxor | vx[:] = vm[:] ? sy ^ vz[:] : vd[:] | |
| VXOR | __vr _ve_pvxor_vvv(__vr vy, __vr vz) | pvxor | vx[:] = vy[:] ^ vz[:] | |
| VXOR | __vr _ve_pvxor_vsv(unsigned long int sy, __vr vz) | pvxor | vx[:] = sy ^ vz[:] | |
| VXOR | __vr _ve_pvxor_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvxor | vx[:] = vm[:] ? vy[:] ^ vz[:] : vd[:] | |
| VXOR | __vr _ve_pvxor_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvxor | vx[:] = vm[:] ? sy ^ vz[:] : vd[:] | |
| VEQV | __vr _ve_veqv_vvv(__vr vy, __vr vz) | veqv | vx[:] = ~(vy[:] ^ vz[:]) | |
| VEQV | __vr _ve_veqv_vsv(unsigned long int sy, __vr vz) | veqv | vx[:] = ~(sy ^ vz[:]) | |
| VEQV | __vr _ve_veqv_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | veqv | vx[:] = vm[:] ? ~(vy[:] ^ vz[:]) : vd[:] | |
| VEQV | __vr _ve_veqv_vsvmv(unsigned long int sy, __vr vz, __vm256 vm, __vr vd) | veqv | vx[:] = vm[:] ? ~(sy ^ vz[:]) : vd[:] | |
| VEQV | __vr _ve_pveqv_vvv(__vr vy, __vr vz) | pveqv | vx[:] = ~(vy[:] ^ vz[:]) | |
| VEQV | __vr _ve_pveqv_vsv(unsigned long int sy, __vr vz) | pveqv | vx[:] = ~(sy ^ vz[:]) | |
| VEQV | __vr _ve_pveqv_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pveqv | vx[:] = vm[:] ? ~(vy[:] ^ vz[:]) : vd[:] | |
| VEQV | __vr _ve_pveqv_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pveqv | vx[:] = vm[:] ? ~(sy ^ vz[:]) : vd[:] | |
| VLDZ | not yet implemented | |||
| VPCNT | not yet implemented | |||
| VBRV | not yet implemented | |||
| VSEQ | __vr _ve_vseq_v() | vseq | vx[:] = i | |
| VSEQ | __vr _ve_pvseqlo_v() | pvseq.lo | vx[:] = i | |
| VSEQ | __vr _ve_pvsequp_v() | pvseq.up | vx[:] = i | |
| VSEQ | __vr _ve_pvseq_v() | pvseq | vx[:] = i |
3. Vector Shift Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VSLL | __vr _ve_vsll_vvv(__vr vz, __vr vy) | vsll | vx[:] = vz[:] « (vy[:] & 0x3f) |
| VSLL | __vr _ve_vsll_vvs(__vr vz, unsigned long int sy) | vsll | vx[:] = vz[:] « (sy & 0x3f) |
| VSLL | __vr _ve_vsll_vvs(__vr vz, unsigned long int N) | vsll | vx[:] = vz[:] « (N & 0x3f) |
| VSLL | __vr _ve_vsll_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsll | vx[:] = vm[:] ? vz[:] « (vy[:] & 0x3f) : vd[:] |
| VSLL | __vr _ve_vsll_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsll | vx[:] = vm[:] ? vz[:] « (sy & 0x3f) : vd[:] |
| VSLL | __vr _ve_vsll_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsll | vx[:] = vm[:] ? vz[:] « (N & 0x3f) : vd[:] |
| VSLL | __vr _ve_pvsll_vvv(__vr vz, __vr vy) | pvsll | vx[:] = vz[:] « (vy[:] & 0x1f) |
| VSLL | __vr _ve_pvsll_vvs(__vr vz, unsigned long int sy) | pvsll | vx[:] = vz[:] « (sy & 0x1f) |
| VSLL | __vr _ve_pvsll_vvvMv(__vr vz, __vr vy, __vm512 vm, __vr vd) | pvsll | vx[:] = vm[:] ? vz[:] « (vy[:] & 0x1f) : vd[:] |
| VSLL | __vr _ve_pvsll_vvsMv(__vr vz, unsigned long int sy, __vm512 vm, __vr vd) | pvsll | vx[:] = vm[:] ? vz[:] « (sy & 0x1f) : vd[:] |
| VSLD | not yet implemented | ||
| VSRL | __vr _ve_vsrl_vvv(__vr vz, __vr vy) | vsrl | vx[:] = vz[:] » (vy[:] & 0x3f) |
| VSRL | __vr _ve_vsrl_vvs(__vr vz, unsigned long int sy) | vsrl | vx[:] = vz[:] » (sy & 0x3f) |
| VSRL | __vr _ve_vsrl_vvs(__vr vz, unsigned long int N) | vsrl | vx[:] = vz[:] » (N & 0x3f) |
| VSRL | __vr _ve_vsrl_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsrl | vx[:] = vm[:] ? vz[:] » (vy[:] & 0x3f) : vd[:] |
| VSRL | __vr _ve_vsrl_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsrl | vx[:] = vm[:] ? vz[:] » (sy & 0x3f) : vd[:] |
| VSRL | __vr _ve_vsrl_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsrl | vx[:] = vm[:] ? vz[:] » (N & 0x3f) : vd[:] |
| VSRL | __vr _ve_pvsrl_vvv(__vr vz, __vr vy) | pvsrl | vx[:] = vz[:] » (vy[:] & 0x1f) |
| VSRL | __vr _ve_pvsrl_vvs(__vr vz, unsigned long int sy) | pvsrl | vx[:] = vz[:] » (sy & 0x1f) |
| VSRL | __vr _ve_pvsrl_vvvMv(__vr vz, __vr vy, __vm512 vm, __vr vd) | pvsrl | vx[:] = vm[:] ? vz[:] » (vy[:] & 0x1f) : vd[:] |
| VSRL | __vr _ve_pvsrl_vvsMv(__vr vz, unsigned long int sy, __vm512 vm, __vr vd) | pvsrl | vx[:] = vm[:] ? vz[:] » (sy & 0x1f) : vd[:] |
| VSRD | not yet implemented | ||
| VSLA | __vr _ve_vslaw_vvv(__vr vz, __vr vy) | vsla.w | vx[:] = vz[:] « (vy[:] & 0x1f) |
| VSLA | __vr _ve_vslaw_vvs(__vr vz, unsigned long int sy) | vsla.w | vx[:] = vz[:] « (sy & 0x1f) |
| VSLA | __vr _ve_vslaw_vvs(__vr vz, unsigned long int N) | vsla.w | vx[:] = vz[:] « (N & 0x1f) |
| VSLA | __vr _ve_vslaw_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsla.w | vx[:] = vm[:] ? vz[:] « (vy[:] & 0x1f) : vd[:] |
| VSLA | __vr _ve_vslaw_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsla.w | vx[:] = vm[:] ? vz[:] « (sy & 0x1f) : vd[:] |
| VSLA | __vr _ve_vslaw_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsla.w | vx[:] = vm[:] ? vz[:] « (N & 0x1f) : vd[:] |
| VSLA | __vr _ve_pvsla_vvv(__vr vz, __vr vy) | pvsla | vx[:] = vz[:] « (vy[:] & 0x1f) |
| VSLA | __vr _ve_pvsla_vvs(__vr vz, unsigned long int sy) | pvsla | vx[:] = vz[:] « (sy & 0x1f) |
| VSLA | __vr _ve_pvsla_vvvMv(__vr vz, __vr vy, __vm512 vm, __vr vd) | pvsla | vx[:] = vm[:] ? vz[:] « (vy[:] & 0x1f) : vd[:] |
| VSLA | __vr _ve_pvsla_vvsMv(__vr vz, unsigned long int sy, __vm512 vm, __vr vd) | pvsla | vx[:] = vm[:] ? vz[:] « (sy & 0x1f) : vd[:] |
| VSLAX | __vr _ve_vslal_vvv(__vr vz, __vr vy) | vsla.l | vx[:] = vz[:] « (vy[:] & 0x3f) |
| VSLAX | __vr _ve_vslal_vvs(__vr vz, unsigned long int sy) | vsla.l | vx[:] = vz[:] « (sy & 0x3f) |
| VSLAX | __vr _ve_vslal_vvs(__vr vz, unsigned long int N) | vsla.l | vx[:] = vz[:] « (N & 0x3f) |
| VSLAX | __vr _ve_vslal_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsla.l | vx[:] = vm[:] ? vz[:] « (vy[:] & 0x3f) : vd[:] |
| VSLAX | __vr _ve_vslal_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsla.l | vx[:] = vm[:] ? vz[:] « (sy & 0x3f) : vd[:] |
| VSLAX | __vr _ve_vslal_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsla.l | vx[:] = vm[:] ? vz[:] « (N & 0x3f) : vd[:] |
| VSRA | __vr _ve_vsraw_vvv(__vr vz, __vr vy) | vsra.w | vx[:] = vz[:] » (vy[:] & 0x1f) |
| VSRA | __vr _ve_vsraw_vvs(__vr vz, unsigned long int sy) | vsra.w | vx[:] = vz[:] » (sy & 0x1f) |
| VSRA | __vr _ve_vsraw_vvs(__vr vz, unsigned long int N) | vsra.w | vx[:] = vz[:] » (N & 0x1f) |
| VSRA | __vr _ve_vsraw_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsra.w | vx[:] = vm[:] ? vz[:] » (vy[:] & 0x1f) : vd[:] |
| VSRA | __vr _ve_vsraw_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsra.w | vx[:] = vm[:] ? vz[:] » (sy & 0x1f) : vd[:] |
| VSRA | __vr _ve_vsraw_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsra.w | vx[:] = vm[:] ? vz[:] » (N & 0x1f) : vd[:] |
| VSRA | __vr _ve_pvsra_vvv(__vr vz, __vr vy) | pvsra | vx[:] = vz[:] » (vy[:] & 0x1f) |
| VSRA | __vr _ve_pvsra_vvs(__vr vz, unsigned long int sy) | pvsra | vx[:] = vz[:] » (sy & 0x1f) |
| VSRA | __vr _ve_pvsra_vvvMv(__vr vz, __vr vy, __vm512 vm, __vr vd) | pvsra | vx[:] = vm[:] ? vz[:] » (vy[:] & 0x1f) : vd[:] |
| VSRA | __vr _ve_pvsra_vvsMv(__vr vz, unsigned long int sy, __vm512 vm, __vr vd) | pvsra | vx[:] = vm[:] ? vz[:] » (sy & 0x1f) : vd[:] |
| VSRAX | __vr _ve_vsral_vvv(__vr vz, __vr vy) | vsra.l | vx[:] = vz[:] » (vy[:] & 0x3f) |
| VSRAX | __vr _ve_vsral_vvs(__vr vz, unsigned long int sy) | vsra.l | vx[:] = vz[:] » (sy & 0x3f) |
| VSRAX | __vr _ve_vsral_vvs(__vr vz, unsigned long int N) | vsra.l | vx[:] = vz[:] » (N & 0x3f) |
| VSRAX | __vr _ve_vsral_vvvmv(__vr vz, __vr vy, __vm256 vm, __vr vd) | vsra.l | vx[:] = vm[:] ? vz[:] » (vy[:] & 0x3f) : vd[:] |
| VSRAX | __vr _ve_vsral_vvsmv(__vr vz, unsigned long int sy, __vm256 vm, __vr vd) | vsra.l | vx[:] = vm[:] ? vz[:] » (sy & 0x3f) : vd[:] |
| VSRAX | __vr _ve_vsral_vvsmv(__vr vz, unsigned long int N, __vm256 vm, __vr vd) | vsra.l | vx[:] = vm[:] ? vz[:] » (N & 0x3f) : vd[:] |
| VSFA | __vr _ve_vsfa_vvss(__vr vz, unsigned long int sy, unsigned long int sz) | vsfa | vx[:] = (vz[:] « (sy & 0x7)) + sz |
| VSFA | __vr _ve_vsfa_vvss(__vr vz, unsigned long int I, unsigned long int sz) | vsfa | vx[:] = (vz[:] « (I & 0x7)) + sz |
| VSFA | __vr _ve_vsfa_vvssmv(__vr vz, unsigned long int sy, unsigned long int sz, __vm256 vm, __vr vd) | vsfa | vx[:] = vm[:] ? (vz[:] « (sy & 0x7)) + sz : vd[:] |
| VSFA | __vr _ve_vsfa_vvssmv(__vr vz, unsigned long int I, unsigned long int sz, __vm256 vm, __vr vd) | vsfa | vx[:] = vm[:] ? (vz[:] « (I & 0x7)) + sz : vd[:] |
4. Vector Floating-Point Operation Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VFAD | __vr _ve_vfaddd_vvv(__vr vy, __vr vz) | vfadd.d | vx[:] = vy[:] + vz[:] |
| VFAD | __vr _ve_vfaddd_vsv(double sy, __vr vz) | vfadd.d | vx[:] = sy + vz[:] |
| VFAD | __vr _ve_vfaddd_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfadd.d | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VFAD | __vr _ve_vfaddd_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfadd.d | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VFAD | __vr _ve_vfadds_vvv(__vr vy, __vr vz) | vfadd.s | vx[:] = vy[:] + vz[:] |
| VFAD | __vr _ve_vfadds_vsv(float sy, __vr vz) | vfadd.s | vx[:] = sy + vz[:] |
| VFAD | __vr _ve_vfadds_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfadd.s | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VFAD | __vr _ve_vfadds_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfadd.s | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VFAD | __vr _ve_pvfadd_vvv(__vr vy, __vr vz) | pvfadd | vx[:] = vy[:] + vz[:] |
| VFAD | __vr _ve_pvfadd_vsv(unsigned long int sy, __vr vz) | pvfadd | vx[:] = sy + vz[:] |
| VFAD | __vr _ve_pvfadd_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfadd | vx[:] = vm[:] ? vy[:] + vz[:] : vd[:] |
| VFAD | __vr _ve_pvfadd_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfadd | vx[:] = vm[:] ? sy + vz[:] : vd[:] |
| VFSB | __vr _ve_vfsubd_vvv(__vr vy, __vr vz) | vfsub.d | vx[:] = vy[:] - vz[:] |
| VFSB | __vr _ve_vfsubd_vsv(double sy, __vr vz) | vfsub.d | vx[:] = sy - vz[:] |
| VFSB | __vr _ve_vfsubd_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfsub.d | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VFSB | __vr _ve_vfsubd_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfsub.d | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VFSB | __vr _ve_vfsubs_vvv(__vr vy, __vr vz) | vfsub.s | vx[:] = vy[:] - vz[:] |
| VFSB | __vr _ve_vfsubs_vsv(float sy, __vr vz) | vfsub.s | vx[:] = sy - vz[:] |
| VFSB | __vr _ve_vfsubs_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfsub.s | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VFSB | __vr _ve_vfsubs_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfsub.s | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VFSB | __vr _ve_pvfsub_vvv(__vr vy, __vr vz) | pvfsub | vx[:] = vy[:] - vz[:] |
| VFSB | __vr _ve_pvfsub_vsv(unsigned long int sy, __vr vz) | pvfsub | vx[:] = sy - vz[:] |
| VFSB | __vr _ve_pvfsub_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfsub | vx[:] = vm[:] ? vy[:] - vz[:] : vd[:] |
| VFSB | __vr _ve_pvfsub_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfsub | vx[:] = vm[:] ? sy - vz[:] : vd[:] |
| VFMP | __vr _ve_vfmuld_vvv(__vr vy, __vr vz) | vfmul.d | vx[:] = vy[:] * vz[:] |
| VFMP | __vr _ve_vfmuld_vsv(double sy, __vr vz) | vfmul.d | vx[:] = sy * vz[:] |
| VFMP | __vr _ve_vfmuld_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmul.d | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VFMP | __vr _ve_vfmuld_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfmul.d | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VFMP | __vr _ve_vfmuls_vvv(__vr vy, __vr vz) | vfmul.s | vx[:] = vy[:] * vz[:] |
| VFMP | __vr _ve_vfmuls_vsv(float sy, __vr vz) | vfmul.s | vx[:] = sy * vz[:] |
| VFMP | __vr _ve_vfmuls_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmul.s | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VFMP | __vr _ve_vfmuls_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfmul.s | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VFMP | __vr _ve_pvfmul_vvv(__vr vy, __vr vz) | pvfmul | vx[:] = vy[:] * vz[:] |
| VFMP | __vr _ve_pvfmul_vsv(unsigned long int sy, __vr vz) | pvfmul | vx[:] = sy * vz[:] |
| VFMP | __vr _ve_pvfmul_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfmul | vx[:] = vm[:] ? vy[:] * vz[:] : vd[:] |
| VFMP | __vr _ve_pvfmul_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfmul | vx[:] = vm[:] ? sy * vz[:] : vd[:] |
| VFDV | __vr _ve_vfdivd_vvv(__vr vy, __vr vz) | vfdiv.d | vx[:] = vy[:] / vz[:] |
| VFDV | __vr _ve_vfdivd_vsv(double sy, __vr vz) | vfdiv.d | vx[:] = sy / vz[:] |
| VFDV | __vr _ve_vfdivd_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfdiv.d | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VFDV | __vr _ve_vfdivd_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfdiv.d | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| VFDV | __vr _ve_vfdivs_vvv(__vr vy, __vr vz) | vfdiv.s | vx[:] = vy[:] / vz[:] |
| VFDV | __vr _ve_vfdivs_vsv(float sy, __vr vz) | vfdiv.s | vx[:] = sy / vz[:] |
| VFDV | __vr _ve_vfdivs_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfdiv.s | vx[:] = vm[:] ? vy[:] / vz[:] : vd[:] |
| VFDV | __vr _ve_vfdivs_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfdiv.s | vx[:] = vm[:] ? sy / vz[:] : vd[:] |
| __vr _ve_vfdivsA_vvv(__vr vy, __vr vz) | vx[:] = vy[:] / vz[:] | ||
| __vr _ve_vfdivsA_vsv(float sy, __vr vz) | vx[:] = sy / vz[:] | ||
| __vr _ve_pvfdivA_vvv(__vr vy, __vr vz) | vx[:] = vy[:] / vz[:] | ||
| VFSQRT | __vr _ve_vfsqrtd_vv(__vr vy) | vfsqrt.d | vx[:] = std::sqrt(vy[:]) |
| VFSQRT | __vr _ve_vfsqrts_vv(__vr vy) | vfsqrt.s | vx[:] = std::sqrt(vy[:]) |
| VFCP | __vr _ve_vfcmpd_vvv(__vr vy, __vr vz) | vfcmp.d | vx[:] = compare(vy[:], vz[:]) |
| VFCP | __vr _ve_vfcmpd_vsv(double sy, __vr vz) | vfcmp.d | vx[:] = compare(sy, vz[:]) |
| VFCP | __vr _ve_vfcmpd_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfcmp.d | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VFCP | __vr _ve_vfcmpd_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfcmp.d | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VFCP | __vr _ve_vfcmps_vvv(__vr vy, __vr vz) | vfcmp.s | vx[:] = compare(vy[:], vz[:]) |
| VFCP | __vr _ve_vfcmps_vsv(float sy, __vr vz) | vfcmp.s | vx[:] = compare(sy, vz[:]) |
| VFCP | __vr _ve_vfcmps_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfcmp.s | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VFCP | __vr _ve_vfcmps_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfcmp.s | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VFCP | __vr _ve_pvfcmp_vvv(__vr vy, __vr vz) | pvfcmp | vx[:] = compare(vy[:], vz[:]) |
| VFCP | __vr _ve_pvfcmp_vsv(unsigned long int sy, __vr vz) | pvfcmp | vx[:] = compare(sy, vz[:]) |
| VFCP | __vr _ve_pvfcmp_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfcmp | vx[:] = vm[:] ? compare(vy[:], vz[:]) : vd[:] |
| VFCP | __vr _ve_pvfcmp_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfcmp | vx[:] = vm[:] ? compare(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmaxd_vvv(__vr vy, __vr vz) | vfmax.d | vx[:] = max(vy[:], vz[:]) |
| VFCM | __vr _ve_vfmaxd_vsv(double sy, __vr vz) | vfmax.d | vx[:] = max(sy, vz[:]) |
| VFCM | __vr _ve_vfmaxd_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmax.d | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmaxd_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfmax.d | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmaxs_vvv(__vr vy, __vr vz) | vfmax.s | vx[:] = max(vy[:], vz[:]) |
| VFCM | __vr _ve_vfmaxs_vsv(float sy, __vr vz) | vfmax.s | vx[:] = max(sy, vz[:]) |
| VFCM | __vr _ve_vfmaxs_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmax.s | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmaxs_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfmax.s | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_pvfmax_vvv(__vr vy, __vr vz) | pvfmax | vx[:] = max(vy[:], vz[:]) |
| VFCM | __vr _ve_pvfmax_vsv(unsigned long int sy, __vr vz) | pvfmax | vx[:] = max(sy, vz[:]) |
| VFCM | __vr _ve_pvfmax_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfmax | vx[:] = vm[:] ? max(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_pvfmax_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfmax | vx[:] = vm[:] ? max(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmind_vvv(__vr vy, __vr vz) | vfmin.d | vx[:] = min(vy[:], vz[:]) |
| VFCM | __vr _ve_vfmind_vsv(double sy, __vr vz) | vfmin.d | vx[:] = min(sy, vz[:]) |
| VFCM | __vr _ve_vfmind_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmin.d | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmind_vsvmv(double sy, __vr vz, __vm256 vm, __vr vd) | vfmin.d | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmins_vvv(__vr vy, __vr vz) | vfmin.s | vx[:] = min(vy[:], vz[:]) |
| VFCM | __vr _ve_vfmins_vsv(float sy, __vr vz) | vfmin.s | vx[:] = min(sy, vz[:]) |
| VFCM | __vr _ve_vfmins_vvvmv(__vr vy, __vr vz, __vm256 vm, __vr vd) | vfmin.s | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_vfmins_vsvmv(float sy, __vr vz, __vm256 vm, __vr vd) | vfmin.s | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VFCM | __vr _ve_pvfmin_vvv(__vr vy, __vr vz) | pvfmin | vx[:] = min(vy[:], vz[:]) |
| VFCM | __vr _ve_pvfmin_vsv(unsigned long int sy, __vr vz) | pvfmin | vx[:] = min(sy, vz[:]) |
| VFCM | __vr _ve_pvfmin_vvvMv(__vr vy, __vr vz, __vm512 vm, __vr vd) | pvfmin | vx[:] = vm[:] ? min(vy[:], vz[:]) : vd[:] |
| VFCM | __vr _ve_pvfmin_vsvMv(unsigned long int sy, __vr vz, __vm512 vm, __vr vd) | pvfmin | vx[:] = vm[:] ? min(sy, vz[:]) : vd[:] |
| VFMAD | __vr _ve_vfmadd_vvvv(__vr vy, __vr vz, __vr vw) | vfmad.d | vx[:] = vz[:] * vw[:] + vy[:] |
| VFMAD | __vr _ve_vfmadd_vsvv(double sy, __vr vz, __vr vw) | vfmad.d | vx[:] = vz[:] * vw[:] + sy |
| VFMAD | __vr _ve_vfmadd_vvsv(__vr vy, double sy, __vr vw) | vfmad.d | vx[:] = sy * vw[:] + vy[:] |
| VFMAD | __vr _ve_vfmadd_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmad.d | vx[:] = vm[:] ? vz[:] * vw[:] + vy[:] : vd[:] |
| VFMAD | __vr _ve_vfmadd_vsvvmv(double sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmad.d | vx[:] = vm[:] ? vz[:] * vw[:] + sy : vd[:] |
| VFMAD | __vr _ve_vfmadd_vvsvmv(__vr vy, double sy, __vr vw, __vm256 vm, __vr vd) | vfmad.d | vx[:] = vm[:] ? sy * vw[:] + vy[:] : vd[:] |
| VFMAD | __vr _ve_vfmads_vvvv(__vr vy, __vr vz, __vr vw) | vfmad.s | vx[:] = vz[:] * vw[:] + vy[:] |
| VFMAD | __vr _ve_vfmads_vsvv(float sy, __vr vz, __vr vw) | vfmad.s | vx[:] = vz[:] * vw[:] + sy |
| VFMAD | __vr _ve_vfmads_vvsv(__vr vy, float sy, __vr vw) | vfmad.s | vx[:] = sy * vw[:] + vy[:] |
| VFMAD | __vr _ve_vfmads_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmad.s | vx[:] = vm[:] ? vz[:] * vw[:] + vy[:] : vd[:] |
| VFMAD | __vr _ve_vfmads_vsvvmv(float sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmad.s | vx[:] = vm[:] ? vz[:] * vw[:] + sy : vd[:] |
| VFMAD | __vr _ve_vfmads_vvsvmv(__vr vy, float sy, __vr vw, __vm256 vm, __vr vd) | vfmad.s | vx[:] = vm[:] ? sy * vw[:] + vy[:] : vd[:] |
| VFMAD | __vr _ve_pvfmad_vvvv(__vr vy, __vr vz, __vr vw) | pvfmad | vx[:] = vz[:] * vw[:] + vy[:] |
| VFMAD | __vr _ve_pvfmad_vsvv(unsigned long int sy, __vr vz, __vr vw) | pvfmad | vx[:] = vz[:] * vw[:] + sy |
| VFMAD | __vr _ve_pvfmad_vvsv(__vr vy, unsigned long int sy, __vr vw) | pvfmad | vx[:] = sy * vw[:] + vy[:] |
| VFMAD | __vr _ve_pvfmad_vvvvMv(__vr vy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfmad | vx[:] = vm[:] ? vz[:] * vw[:] + vy[:] : vd[:] |
| VFMAD | __vr _ve_pvfmad_vsvvMv(unsigned long int sy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfmad | vx[:] = vm[:] ? vz[:] * vw[:] + sy : vd[:] |
| VFMAD | __vr _ve_pvfmad_vvsvMv(__vr vy, unsigned long int sy, __vr vw, __vm512 vm, __vr vd) | pvfmad | vx[:] = vm[:] ? sy * vw[:] + vy[:] : vd[:] |
| VFMSB | __vr _ve_vfmsbd_vvvv(__vr vy, __vr vz, __vr vw) | vfmsb.d | vx[:] = vz[:] * vw[:] - vy[:] |
| VFMSB | __vr _ve_vfmsbd_vsvv(double sy, __vr vz, __vr vw) | vfmsb.d | vx[:] = vz[:] * vw[:] - sy |
| VFMSB | __vr _ve_vfmsbd_vvsv(__vr vy, double sy, __vr vw) | vfmsb.d | vx[:] = sy * vw[:] - vy[:] |
| VFMSB | __vr _ve_vfmsbd_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmsb.d | vx[:] = vm[:] ? vz[:] * vw[:] - vy[:] : vd[:] |
| VFMSB | __vr _ve_vfmsbd_vsvvmv(double sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmsb.d | vx[:] = vm[:] ? vz[:] * vw[:] - sy : vd[:] |
| VFMSB | __vr _ve_vfmsbd_vvsvmv(__vr vy, double sy, __vr vw, __vm256 vm, __vr vd) | vfmsb.d | vx[:] = vm[:] ? sy * vw[:] - vy[:] : vd[:] |
| VFMSB | __vr _ve_vfmsbs_vvvv(__vr vy, __vr vz, __vr vw) | vfmsb.s | vx[:] = vz[:] * vw[:] - vy[:] |
| VFMSB | __vr _ve_vfmsbs_vsvv(float sy, __vr vz, __vr vw) | vfmsb.s | vx[:] = vz[:] * vw[:] - sy |
| VFMSB | __vr _ve_vfmsbs_vvsv(__vr vy, float sy, __vr vw) | vfmsb.s | vx[:] = sy * vw[:] - vy[:] |
| VFMSB | __vr _ve_vfmsbs_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmsb.s | vx[:] = vm[:] ? vz[:] * vw[:] - vy[:] : vd[:] |
| VFMSB | __vr _ve_vfmsbs_vsvvmv(float sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfmsb.s | vx[:] = vm[:] ? vz[:] * vw[:] - sy : vd[:] |
| VFMSB | __vr _ve_vfmsbs_vvsvmv(__vr vy, float sy, __vr vw, __vm256 vm, __vr vd) | vfmsb.s | vx[:] = vm[:] ? sy * vw[:] - vy[:] : vd[:] |
| VFMSB | __vr _ve_pvfmsb_vvvv(__vr vy, __vr vz, __vr vw) | pvfmsb | vx[:] = vz[:] * vw[:] - vy[:] |
| VFMSB | __vr _ve_pvfmsb_vsvv(unsigned long int sy, __vr vz, __vr vw) | pvfmsb | vx[:] = vz[:] * vw[:] - sy |
| VFMSB | __vr _ve_pvfmsb_vvsv(__vr vy, unsigned long int sy, __vr vw) | pvfmsb | vx[:] = sy * vw[:] - vy[:] |
| VFMSB | __vr _ve_pvfmsb_vvvvMv(__vr vy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfmsb | vx[:] = vm[:] ? vz[:] * vw[:] - vy[:] : vd[:] |
| VFMSB | __vr _ve_pvfmsb_vsvvMv(unsigned long int sy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfmsb | vx[:] = vm[:] ? vz[:] * vw[:] - sy : vd[:] |
| VFMSB | __vr _ve_pvfmsb_vvsvMv(__vr vy, unsigned long int sy, __vr vw, __vm512 vm, __vr vd) | pvfmsb | vx[:] = vm[:] ? sy * vw[:] - vy[:] : vd[:] |
| VFNMAD | __vr _ve_vfnmadd_vvvv(__vr vy, __vr vz, __vr vw) | vfnmad.d | vx[:] = - (vz[:] * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_vfnmadd_vsvv(double sy, __vr vz, __vr vw) | vfnmad.d | vx[:] = - (vz[:] * vw[:] + sy) |
| VFNMAD | __vr _ve_vfnmadd_vvsv(__vr vy, double sy, __vr vw) | vfnmad.d | vx[:] = - (sy * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_vfnmadd_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmad.d | vx[:] = vm[:] ? - (vz[:] * vw[:] + vy[:]) : vd[:] |
| VFNMAD | __vr _ve_vfnmadd_vsvvmv(double sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmad.d | vx[:] = vm[:] ? - (vz[:] * vw[:] + sy) : vd[:] |
| VFNMAD | __vr _ve_vfnmadd_vvsvmv(__vr vy, double sy, __vr vw, __vm256 vm, __vr vd) | vfnmad.d | vx[:] = vm[:] ? - (sy * vw[:] + vy[:]) : vd[:] |
| VFNMAD | __vr _ve_vfnmads_vvvv(__vr vy, __vr vz, __vr vw) | vfnmad.s | vx[:] = - (vz[:] * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_vfnmads_vsvv(float sy, __vr vz, __vr vw) | vfnmad.s | vx[:] = - (vz[:] * vw[:] + sy) |
| VFNMAD | __vr _ve_vfnmads_vvsv(__vr vy, float sy, __vr vw) | vfnmad.s | vx[:] = - (sy * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_vfnmads_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmad.s | vx[:] = vm[:] ? - (vz[:] * vw[:] + vy[:]) : vd[:] |
| VFNMAD | __vr _ve_vfnmads_vsvvmv(float sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmad.s | vx[:] = vm[:] ? - (vz[:] * vw[:] + sy) : vd[:] |
| VFNMAD | __vr _ve_vfnmads_vvsvmv(__vr vy, float sy, __vr vw, __vm256 vm, __vr vd) | vfnmad.s | vx[:] = vm[:] ? - (sy * vw[:] + vy[:]) : vd[:] |
| VFNMAD | __vr _ve_pvfnmad_vvvv(__vr vy, __vr vz, __vr vw) | pvfnmad | vx[:] = - (vz[:] * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_pvfnmad_vsvv(unsigned long int sy, __vr vz, __vr vw) | pvfnmad | vx[:] = - (vz[:] * vw[:] + sy) |
| VFNMAD | __vr _ve_pvfnmad_vvsv(__vr vy, unsigned long int sy, __vr vw) | pvfnmad | vx[:] = - (sy * vw[:] + vy[:]) |
| VFNMAD | __vr _ve_pvfnmad_vvvvMv(__vr vy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfnmad | vx[:] = vm[:] ? - (vz[:] * vw[:] + vy[:]) : vd[:] |
| VFNMAD | __vr _ve_pvfnmad_vsvvMv(unsigned long int sy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfnmad | vx[:] = vm[:] ? - (vz[:] * vw[:] + sy) : vd[:] |
| VFNMAD | __vr _ve_pvfnmad_vvsvMv(__vr vy, unsigned long int sy, __vr vw, __vm512 vm, __vr vd) | pvfnmad | vx[:] = vm[:] ? - (sy * vw[:] + vy[:]) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbd_vvvv(__vr vy, __vr vz, __vr vw) | vfnmsb.d | vx[:] = - (vz[:] * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_vfnmsbd_vsvv(double sy, __vr vz, __vr vw) | vfnmsb.d | vx[:] = - (vz[:] * vw[:] - sy) |
| VFNMSB | __vr _ve_vfnmsbd_vvsv(__vr vy, double sy, __vr vw) | vfnmsb.d | vx[:] = - (sy * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_vfnmsbd_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmsb.d | vx[:] = vm[:] ? - (vz[:] * vw[:] - vy[:]) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbd_vsvvmv(double sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmsb.d | vx[:] = vm[:] ? - (vz[:] * vw[:] - sy) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbd_vvsvmv(__vr vy, double sy, __vr vw, __vm256 vm, __vr vd) | vfnmsb.d | vx[:] = vm[:] ? - (sy * vw[:] - vy[:]) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbs_vvvv(__vr vy, __vr vz, __vr vw) | vfnmsb.s | vx[:] = - (vz[:] * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_vfnmsbs_vsvv(float sy, __vr vz, __vr vw) | vfnmsb.s | vx[:] = - (vz[:] * vw[:] - sy) |
| VFNMSB | __vr _ve_vfnmsbs_vvsv(__vr vy, float sy, __vr vw) | vfnmsb.s | vx[:] = - (sy * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_vfnmsbs_vvvvmv(__vr vy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmsb.s | vx[:] = vm[:] ? - (vz[:] * vw[:] - vy[:]) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbs_vsvvmv(float sy, __vr vz, __vr vw, __vm256 vm, __vr vd) | vfnmsb.s | vx[:] = vm[:] ? - (vz[:] * vw[:] - sy) : vd[:] |
| VFNMSB | __vr _ve_vfnmsbs_vvsvmv(__vr vy, float sy, __vr vw, __vm256 vm, __vr vd) | vfnmsb.s | vx[:] = vm[:] ? - (sy * vw[:] - vy[:]) : vd[:] |
| VFNMSB | __vr _ve_pvfnmsb_vvvv(__vr vy, __vr vz, __vr vw) | pvfnmsb | vx[:] = - (vz[:] * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_pvfnmsb_vsvv(unsigned long int sy, __vr vz, __vr vw) | pvfnmsb | vx[:] = - (vz[:] * vw[:] - sy) |
| VFNMSB | __vr _ve_pvfnmsb_vvsv(__vr vy, unsigned long int sy, __vr vw) | pvfnmsb | vx[:] = - (sy * vw[:] - vy[:]) |
| VFNMSB | __vr _ve_pvfnmsb_vvvvMv(__vr vy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfnmsb | vx[:] = vm[:] ? - (vz[:] * vw[:] - vy[:]) : vd[:] |
| VFNMSB | __vr _ve_pvfnmsb_vsvvMv(unsigned long int sy, __vr vz, __vr vw, __vm512 vm, __vr vd) | pvfnmsb | vx[:] = vm[:] ? - (vz[:] * vw[:] - sy) : vd[:] |
| VFNMSB | __vr _ve_pvfnmsb_vvsvMv(__vr vy, unsigned long int sy, __vr vw, __vm512 vm, __vr vd) | pvfnmsb | vx[:] = vm[:] ? - (sy * vw[:] - vy[:]) : vd[:] |
| VRCP | __vr _ve_vrcpd_vv(__vr vy) | vrcp.d | vx[:] = 1.0f / vy[:] |
| VRCP | __vr _ve_vrcps_vv(__vr vy) | vrcp.s | vx[:] = 1.0f / vy[:] |
| VRCP | __vr _ve_pvrcp_vv(__vr vy) | pvrcp | vx[:] = 1.0f / vy[:] |
| VRSQRT | __vr _ve_vrsqrtd_vv(__vr vy) | vrsqrt.d | vx[:] = 1.0f / std::sqrt(vy[:]) |
| VRSQRT | __vr _ve_vrsqrts_vv(__vr vy) | vrsqrt.s | vx[:] = 1.0f / std::sqrt(vy[:]) |
| VRSQRT | __vr _ve_pvrsqrt_vv(__vr vy) | pvrsqrt | vx[:] = 1.0f / std::sqrt(vy[:]) |
| VRSQRTnex | not yet implemented | ||
| VFIX | __vr _ve_vcvtwdsx_vv(__vr vy) | vcvt.w.d.sx | vx[:] = (int)(vy[:]+0.5) |
| VFIX | __vr _ve_vcvtwdsx_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.d.sx | vx[:] = vm[:] ? (int)(vy[:]+0.5) : vd[:] |
| VFIX | __vr _ve_vcvtwdsxrz_vv(__vr vy) | vcvt.w.d.sx.rz | vx[:] = (int)(vy[:]) |
| VFIX | __vr _ve_vcvtwdsxrz_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.d.sx.rz | vx[:] = vm[:] ? (int)(vy[:]) : vd[:] |
| VFIX | __vr _ve_vcvtwdzx_vv(__vr vy) | vcvt.w.d.zx | vx[:] = (unsigned int)(vy[:]+0.5) |
| VFIX | __vr _ve_vcvtwdzx_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.d.zx | vx[:] = vm[:] ? (unsigned int)(vy[:]+0.5) : vd[:] |
| VFIX | __vr _ve_vcvtwdzxrz_vv(__vr vy) | vcvt.w.d.zx.rz | vx[:] = (unsigned int)(vy[:]) |
| VFIX | __vr _ve_vcvtwdzxrz_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.d.zx.rz | vx[:] = vm[:] ? (unsigned int)(vy[:]) : vd[:] |
| VFIX | __vr _ve_vcvtwssx_vv(__vr vy) | vcvt.w.s.sx | vx[:] = (int)(vy[:]+0.5) |
| VFIX | __vr _ve_vcvtwssx_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.s.sx | vx[:] = vm[:] ? (int)(vy[:]+0.5) : vd[:] |
| VFIX | __vr _ve_vcvtwssxrz_vv(__vr vy) | vcvt.w.s.sx.rz | vx[:] = (int)(vy[:]) |
| VFIX | __vr _ve_vcvtwssxrz_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.s.sx.rz | vx[:] = vm[:] ? (int)(vy[:]) : vd[:] |
| VFIX | __vr _ve_vcvtwszx_vv(__vr vy) | vcvt.w.s.zx | vx[:] = (unsigned int)(vy[:]+0.5) |
| VFIX | __vr _ve_vcvtwszx_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.s.zx | vx[:] = vm[:] ? (unsigned int)(vy[:]+0.5) : vd[:] |
| VFIX | __vr _ve_vcvtwszxrz_vv(__vr vy) | vcvt.w.s.zx.rz | vx[:] = (unsigned int)(vy[:]) |
| VFIX | __vr _ve_vcvtwszxrz_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.w.s.zx.rz | vx[:] = vm[:] ? (unsigned int)(vy[:]) : vd[:] |
| VFIX | __vr _ve_pvcvtws_vv(__vr vy) | pvcvt.w.s | vx[:] = (int)(vy[:]+0.5) |
| VFIX | __vr _ve_pvcvtws_vvMv(__vr vy, __vm512 vm, __vr vd) | pvcvt.w.s | vx[:] = vm[:] ? (int)(vy[:]+0.5) : vd[:] |
| VFIX | __vr _ve_pvcvtwsrz_vv(__vr vy) | pvcvt.w.s.rz | vx[:] = (int)(vy[:]) |
| VFIX | __vr _ve_pvcvtwsrz_vvMv(__vr vy, __vm512 vm, __vr vd) | pvcvt.w.s.rz | vx[:] = vm[:] ? (int)(vy[:]) : vd[:] |
| VFIXX | __vr _ve_vcvtld_vv(__vr vy) | vcvt.l.d | vx[:] = (long long)(vy[:]+0.5) |
| VFIXX | __vr _ve_vcvtld_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.l.d | vx[:] = vm[:] ? (long long)(vy[:]+0.5) : vd[:] |
| VFIXX | __vr _ve_vcvtldrz_vv(__vr vy) | vcvt.l.d.rz | vx[:] = (long long)(vy[:]) |
| VFIXX | __vr _ve_vcvtldrz_vvmv(__vr vy, __vm256 vm, __vr vd) | vcvt.l.d.rz | vx[:] = vm[:] ? (long long)(vy[:]) : vd[:] |
| VFLT | __vr _ve_vcvtdw_vv(__vr vy) | vcvt.d.w | vx[:] = (double)vy[:] |
| VFLT | __vr _ve_vcvtsw_vv(__vr vy) | vcvt.s.w | vx[:] = (float)vy[:] |
| VFLT | __vr _ve_pvcvtsw_vv(__vr vy) | pvcvt.s.w | vx[:] = (float)vy[:] |
| VFLTX | __vr _ve_vcvtdl_vv(__vr vy) | vcvt.d.l | vx[:] = (double)vy[:] |
| VCVD | __vr _ve_vcvtds_vv(__vr vy) | vcvt.d.s | vx[:] = (double)vy[:] |
| VCVS | __vr _ve_vcvtsd_vv(__vr vy) | vcvt.s.d | vx[:] = (float)vy[:] |
5. Vector Mask Arithmetic Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VMRG | __vr _ve_vmrg_vvvm(__vr vy, __vr vz, __vm256 vm) | vmrg | |
| VMRG | __vr _ve_vmrgw_vvvM(__vr vy, __vr vz, __vm512 vm) | vmrg.w | |
| VSHF | __vr _ve_vshf_vvvs(__vr vy, __vr vz, unsigned long int sy) | vshf | |
| VSHF | __vr _ve_vshf_vvvs(__vr vy, __vr vz, unsigned long int N) | vshf | |
| VCP | __vr _ve_vcp_vvmv(__vr vz, __vm256 vm, __vr vd) | vcp | |
| VEX | __vr _ve_vex_vvmv(__vr vz, __vm256 vm, __vr vd) | vex | |
| VFMK | __vm256 _ve_vfmkl_mcv(int cc, __vr vz) | vfmk.l | |
| VFMK | __vm256 _ve_vfmkl_mcvm(int cc, __vr vz, __vm256 vm) | vfmk.l | |
| VFMK | __vm256 _ve_vfmkat_m() | vfmk.at | |
| VFMK | __vm256 _ve_vfmkaf_m() | vfmk.af | |
| VFMK | __vm512 _ve_pvfmkat_M() | ||
| VFMK | __vm512 _ve_pvfmkaf_M() | ||
| VFMS | __vm256 _ve_vfmkw_mcv(int cc, __vr vz) | vfmk.w | |
| VFMS | __vm256 _ve_vfmkw_mcvm(int cc, __vr vz, __vm256 vm) | vfmk.w | |
| VFMS | __vm512 _ve_pvfmkw_Mcv(int cc, __vr vz) | ||
| VFMS | __vm512 _ve_pvfmkw_McvM(int cc, __vr vz, __vm512 vm) | ||
| VFMF | __vm256 _ve_vfmkd_mcv(int cc, __vr vz) | vfmk.d | |
| VFMF | __vm256 _ve_vfmkd_mcvm(int cc, __vr vz, __vm256 vm) | vfmk.d | |
| VFMF | __vm256 _ve_vfmks_mcv(int cc, __vr vz) | vfmk.s | |
| VFMF | __vm256 _ve_vfmks_mcvm(int cc, __vr vz, __vm256 vm) | vfmk.s | |
| VFMF | __vm512 _ve_pvfmks_Mcv(int cc, __vr vz) | ||
| VFMF | __vm512 _ve_pvfmks_McvM(int cc, __vr vz, __vm512 vm) |
6. Vector Recursive Relation Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VSUMS | __vr _ve_vsumwsx_vv(__vr vy) | vsum.w.sx | |
| VSUMS | __vr _ve_vsumwsx_vvm(__vr vy, __vm256 vm) | vsum.w.sx | |
| VSUMS | __vr _ve_vsumwzx_vv(__vr vy) | vsum.w.zx | |
| VSUMS | __vr _ve_vsumwzx_vvm(__vr vy, __vm256 vm) | vsum.w.zx | |
| VSUMX | __vr _ve_vsuml_vv(__vr vy) | vsum.l | |
| VSUMX | __vr _ve_vsuml_vvm(__vr vy, __vm256 vm) | vsum.l | |
| VFSUM | __vr _ve_vfsumd_vv(__vr vy) | vfsum.d | |
| VFSUM | __vr _ve_vfsumd_vvm(__vr vy, __vm256 vm) | vfsum.d | |
| VFSUM | __vr _ve_vfsums_vv(__vr vy) | vfsum.s | |
| VFSUM | __vr _ve_vfsums_vvm(__vr vy, __vm256 vm) | vfsum.s | |
| VMAXS | __vr _ve_vrmaxswfstsx_vv(__vr vy) | vrmaxs.w.fst.sx | |
| VMAXS | __vr _ve_vrmaxswlstsx_vv(__vr vy) | vrmaxs.w.lst.sx | |
| VMAXS | __vr _ve_vrmaxswfstzx_vv(__vr vy) | vrmaxs.w.fst.zx | |
| VMAXS | __vr _ve_vrmaxswlstzx_vv(__vr vy) | vrmaxs.w.lst.zx | |
| VMAXS | __vr _ve_vrminswfstsx_vv(__vr vy) | vrmins.w.fst.sx | |
| VMAXS | __vr _ve_vrminswlstsx_vv(__vr vy) | vrmins.w.lst.sx | |
| VMAXS | __vr _ve_vrminswfstzx_vv(__vr vy) | vrmins.w.fst.zx | |
| VMAXS | __vr _ve_vrminswlstzx_vv(__vr vy) | vrmins.w.lst.zx | |
| VMAXX | __vr _ve_vrmaxslfst_vv(__vr vy) | vrmaxs.l.fst | |
| VMAXX | __vr _ve_vrmaxsllst_vv(__vr vy) | vrmaxs.l.lst | |
| VMAXX | __vr _ve_vrminslfst_vv(__vr vy) | vrmins.l.fst | |
| VMAXX | __vr _ve_vrminsllst_vv(__vr vy) | vrmins.l.lst | |
| VFMAX | __vr _ve_vfrmaxdfst_vv(__vr vy) | vfrmax.d.fst | |
| VFMAX | __vr _ve_vfrmaxdlst_vv(__vr vy) | vfrmax.d.lst | |
| VFMAX | __vr _ve_vfrmaxsfst_vv(__vr vy) | vfrmax.s.fst | |
| VFMAX | __vr _ve_vfrmaxslst_vv(__vr vy) | vfrmax.s.lst | |
| VFMAX | __vr _ve_vfrmindfst_vv(__vr vy) | vfrmin.d.fst | |
| VFMAX | __vr _ve_vfrmindlst_vv(__vr vy) | vfrmin.d.lst | |
| VFMAX | __vr _ve_vfrminsfst_vv(__vr vy) | vfrmin.s.fst | |
| VFMAX | __vr _ve_vfrminslst_vv(__vr vy) | vfrmin.s.lst | |
| VRAND | not yet implemented | ||
| VROR | not yet implemented | ||
| VRXOR | not yet implemented | ||
| VFIA | not yet implemented | ||
| VFIS | not yet implemented | ||
| VFIM | not yet implemented | ||
| VFIAM | not yet implemented | ||
| VFISM | not yet implemented | ||
| VFIMA | not yet implemented | ||
| VFIMS | not yet implemented |
7. Vector Gathering/Scattering Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| VGT | __vr _ve_vgt_vv(__vr vy) | vgt | |
| VGT | __vr _ve_vgt_vvm(__vr vy, __vm256 vm) | vgt | |
| VGTU | __vr _ve_vgtu_vv(__vr vy) | vgtu | |
| VGTU | __vr _ve_vgtu_vvm(__vr vy, __vm256 vm) | vgtu | |
| VGTL | __vr _ve_vgtlsx_vv(__vr vy) | vgtl.sx | |
| VGTL | __vr _ve_vgtlsx_vvm(__vr vy, __vm256 vm) | vgtl.sx | |
| VGTL | __vr _ve_vgtlzx_vv(__vr vy) | vgtl.zx | |
| VGTL | __vr _ve_vgtlzx_vvm(__vr vy, __vm256 vm) | vgtl.zx | |
| VSC | void _ve_vsc_vv(__vr vx, __vr vy) | vsc | |
| VSC | void _ve_vsc_vvm(__vr vx, __vr vy, __vm256 vm) | vsc | |
| VSC | void _ve_vscot_vv(__vr vx, __vr vy) | vsc.ot | |
| VSC | void _ve_vscot_vvm(__vr vx, __vr vy, __vm256 vm) | vsc.ot | |
| VSCU | void _ve_vscu_vv(__vr vx, __vr vy) | vscu | |
| VSCU | void _ve_vscu_vvm(__vr vx, __vr vy, __vm256 vm) | vscu | |
| VSCU | void _ve_vscuot_vv(__vr vx, __vr vy) | vscu.ot | |
| VSCU | void _ve_vscuot_vvm(__vr vx, __vr vy, __vm256 vm) | vscu.ot | |
| VSCL | void _ve_vscl_vv(__vr vx, __vr vy) | vscl | |
| VSCL | void _ve_vscl_vvm(__vr vx, __vr vy, __vm256 vm) | vscl | |
| VSCL | void _ve_vsclot_vv(__vr vx, __vr vy) | vscl.ot | |
| VSCL | void _ve_vsclot_vvm(__vr vx, __vr vy, __vm256 vm) | vscl.ot |
8. Vector Mask Register Instructions
| Instruction | Function | asm | Description | |
|---|---|---|---|---|
| ANDM | __vm256 _ve_andm_mmm(__vm256 vmy, __vm256 vmz) | andm | vmx[:] = vmy[:] ? vmy[:] & vmz[:] : vmz[:] | |
| ANDM | __vm512 _ve_andm_MMM(__vm512 vmy, __vm512 vmz) | andm | vmx[:] = vmy[:] ? vmy[:] & vmz[:] : vmz[:] | |
| ORM | __vm256 _ve_orm_mmm(__vm256 vmy, __vm256 vmz) | orm | vmx[:] = vmy[:] ? vmy[:] | vmz[:] : vmz[:] |
| ORM | __vm512 _ve_orm_MMM(__vm512 vmy, __vm512 vmz) | orm | vmx[:] = vmy[:] ? vmy[:] | vmz[:] : vmz[:] |
| XORM | __vm256 _ve_xorm_mmm(__vm256 vmy, __vm256 vmz) | xorm | vmx[:] = vmy[:] ? vmy[:] ^ vmz[:] : vmz[:] | |
| XORM | __vm512 _ve_xorm_MMM(__vm512 vmy, __vm512 vmz) | xorm | vmx[:] = vmy[:] ? vmy[:] ^ vmz[:] : vmz[:] | |
| EQVM | __vm256 _ve_eqvm_mmm(__vm256 vmy, __vm256 vmz) | eqvm | vmx[:] = vmy[:] ? ~(vmy[:] ^ vmz[:]) : vmz[:] | |
| EQVM | __vm512 _ve_eqvm_MMM(__vm512 vmy, __vm512 vmz) | eqvm | vmx[:] = vmy[:] ? ~(vmy[:] ^ vmz[:]) : vmz[:] | |
| NNDM | __vm256 _ve_nndm_mmm(__vm256 vmy, __vm256 vmz) | nndm | vmx[:] = vmy[:] ? (~vmy[:]) & vmz[:] : vmz[:] | |
| NNDM | __vm512 _ve_nndm_MMM(__vm512 vmy, __vm512 vmz) | nndm | vmx[:] = vmy[:] ? (~vmy[:]) & vmz[:] : vmz[:] | |
| NEGM | __vm256 _ve_negm_mm(__vm256 vmy) | negm | vmx[:] = vmx[:] ? ~vmy[:] : vmy[:] | |
| NEGM | __vm512 _ve_negm_MM(__vm512 vmy) | negm | vmx[:] = vmx[:] ? ~vmy[:] : vmy[:] | |
| PCVM | unsigned long int _ve_pcvm_sm(__vm256 vmy) | pcvm | ||
| LZVM | unsigned long int _ve_lzvm_sm(__vm256 vmy) | lzvm | ||
| TOVM | unsigned long int _ve_tovm_sm(__vm256 vmy) | tovm |
9. Vector Control Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| LVL | void _ve_lvl(int vl) | lvl | |
| SVL | not yet implemented | ||
| SMVL | not yet implemented | ||
| LVIX | not yet implemented |
10. Control Instructions
| Instruction | Function | asm | Description |
|---|---|---|---|
| SVOB | void _ve_svob(void) | svob |
11. Others
| Instruction | Function | asm | Description |
|---|---|---|---|
| unsigned long int _ve_pack_f32p(float const* p0, float const* p1) | ldu,ldl,or | ||
| unsigned long int _ve_pack_f32a(float const* p) | load and mul | ||
| unsigned long int _ve_pack_i32(int a, int b) | sll,add,or | ||
| __vr _ve_vec_expf_vv(__vr vy) | vx[:] = expf(vy[:]) | ||
| __vr _ve_vec_exp_vv(__vr vy) | vx[:] = exp(vy[:]) | ||
| __vm _ve_extract_vm512u(__vm512 vm) | |||
| __vm _ve_extract_vm512l(__vm512 vm) | |||
| __vm512 _ve_insert_vm512u(__vm512 vmx, __vm vmy) | |||
| __vm512 _ve_insert_vm512l(__vm512 vmx, __vm vmy) |