//! Galois Field New Instructions (GFNI) //! //! The intrinsics here correspond to those in the `immintrin.h` C header. //! //! The reference is [Intel 64 and IA-32 Architectures Software Developer's //! Manual Volume 2: Instruction Set Reference, A-Z][intel64_ref]. //! //! [intel64_ref]: http://www.intel.de/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf use crate::core_arch::simd::i8x16; use crate::core_arch::simd::i8x32; use crate::core_arch::simd::i8x64; use crate::core_arch::simd_llvm::simd_select_bitmask; use crate::core_arch::x86::__m128i; use crate::core_arch::x86::__m256i; use crate::core_arch::x86::__m512i; use crate::core_arch::x86::__mmask16; use crate::core_arch::x86::__mmask32; use crate::core_arch::x86::__mmask64; use crate::core_arch::x86::_mm256_setzero_si256; use crate::core_arch::x86::_mm512_setzero_si512; use crate::core_arch::x86::_mm_setzero_si128; use crate::core_arch::x86::m128iExt; use crate::core_arch::x86::m256iExt; use crate::core_arch::x86::m512iExt; use crate::mem::transmute; #[cfg(test)] use stdarch_test::assert_instr; #[allow(improper_ctypes)] extern "C" { #[link_name = "llvm.x86.vgf2p8affineinvqb.512"] fn vgf2p8affineinvqb_512(x: i8x64, a: i8x64, imm8: u8) -> i8x64; #[link_name = "llvm.x86.vgf2p8affineinvqb.256"] fn vgf2p8affineinvqb_256(x: i8x32, a: i8x32, imm8: u8) -> i8x32; #[link_name = "llvm.x86.vgf2p8affineinvqb.128"] fn vgf2p8affineinvqb_128(x: i8x16, a: i8x16, imm8: u8) -> i8x16; #[link_name = "llvm.x86.vgf2p8affineqb.512"] fn vgf2p8affineqb_512(x: i8x64, a: i8x64, imm8: u8) -> i8x64; #[link_name = "llvm.x86.vgf2p8affineqb.256"] fn vgf2p8affineqb_256(x: i8x32, a: i8x32, imm8: u8) -> i8x32; #[link_name = "llvm.x86.vgf2p8affineqb.128"] fn vgf2p8affineqb_128(x: i8x16, a: i8x16, imm8: u8) -> i8x16; #[link_name = "llvm.x86.vgf2p8mulb.512"] fn vgf2p8mulb_512(a: i8x64, b: i8x64) -> i8x64; #[link_name = "llvm.x86.vgf2p8mulb.256"] fn vgf2p8mulb_256(a: i8x32, b: i8x32) -> i8x32; #[link_name = "llvm.x86.vgf2p8mulb.128"] fn vgf2p8mulb_128(a: i8x16, b: i8x16) -> i8x16; } // LLVM requires AVX512BW for a lot of these instructions, see // https://github.com/llvm/llvm-project/blob/release/9.x/clang/include/clang/Basic/BuiltinsX86.def#L457 // however our tests also require the target feature list to match Intel's // which *doesn't* require AVX512BW but only AVX512F, so we added the redundant AVX512F // requirement (for now) // also see // https://github.com/llvm/llvm-project/blob/release/9.x/clang/lib/Headers/gfniintrin.h // for forcing GFNI, BW and optionally VL extension /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm512_gf2p8mul_epi8(a: __m512i, b: __m512i) -> __m512i { transmute(vgf2p8mulb_512(a.as_i8x64(), b.as_i8x64())) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_mask_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm512_mask_gf2p8mul_epi8( src: __m512i, k: __mmask64, a: __m512i, b: __m512i, ) -> __m512i { transmute(simd_select_bitmask( k, vgf2p8mulb_512(a.as_i8x64(), b.as_i8x64()), src.as_i8x64(), )) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_maskz_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm512_maskz_gf2p8mul_epi8(k: __mmask64, a: __m512i, b: __m512i) -> __m512i { let zero = _mm512_setzero_si512().as_i8x64(); transmute(simd_select_bitmask( k, vgf2p8mulb_512(a.as_i8x64(), b.as_i8x64()), zero, )) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm256_gf2p8mul_epi8(a: __m256i, b: __m256i) -> __m256i { transmute(vgf2p8mulb_256(a.as_i8x32(), b.as_i8x32())) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_mask_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm256_mask_gf2p8mul_epi8( src: __m256i, k: __mmask32, a: __m256i, b: __m256i, ) -> __m256i { transmute(simd_select_bitmask( k, vgf2p8mulb_256(a.as_i8x32(), b.as_i8x32()), src.as_i8x32(), )) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_maskz_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm256_maskz_gf2p8mul_epi8(k: __mmask32, a: __m256i, b: __m256i) -> __m256i { let zero = _mm256_setzero_si256().as_i8x32(); transmute(simd_select_bitmask( k, vgf2p8mulb_256(a.as_i8x32(), b.as_i8x32()), zero, )) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni")] #[cfg_attr(test, assert_instr(gf2p8mulb))] pub unsafe fn _mm_gf2p8mul_epi8(a: __m128i, b: __m128i) -> __m128i { transmute(vgf2p8mulb_128(a.as_i8x16(), b.as_i8x16())) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_mask_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm_mask_gf2p8mul_epi8( src: __m128i, k: __mmask16, a: __m128i, b: __m128i, ) -> __m128i { transmute(simd_select_bitmask( k, vgf2p8mulb_128(a.as_i8x16(), b.as_i8x16()), src.as_i8x16(), )) } /// Performs a multiplication in GF(2^8) on the packed bytes. /// The field is in polynomial representation with the reduction polynomial /// x^8 + x^4 + x^3 + x + 1. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_maskz_gf2p8mul_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8mulb))] pub unsafe fn _mm_maskz_gf2p8mul_epi8(k: __mmask16, a: __m128i, b: __m128i) -> __m128i { let zero = _mm_setzero_si128().as_i8x16(); transmute(simd_select_bitmask( k, vgf2p8mulb_128(a.as_i8x16(), b.as_i8x16()), zero, )) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm512_gf2p8affine_epi64_epi8(x: __m512i, a: __m512i) -> __m512i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineqb_512(x, a, b); transmute(r) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_maskz_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm512_maskz_gf2p8affine_epi64_epi8( k: __mmask64, x: __m512i, a: __m512i, ) -> __m512i { static_assert_imm8!(B); let b = B as u8; let zero = _mm512_setzero_si512().as_i8x64(); let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineqb_512(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_mask_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm512_mask_gf2p8affine_epi64_epi8( src: __m512i, k: __mmask64, x: __m512i, a: __m512i, ) -> __m512i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineqb_512(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x64())) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm256_gf2p8affine_epi64_epi8(x: __m256i, a: __m256i) -> __m256i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineqb_256(x, a, b); transmute(r) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_maskz_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm256_maskz_gf2p8affine_epi64_epi8( k: __mmask32, x: __m256i, a: __m256i, ) -> __m256i { static_assert_imm8!(B); let b = B as u8; let zero = _mm256_setzero_si256().as_i8x32(); let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineqb_256(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_mask_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm256_mask_gf2p8affine_epi64_epi8( src: __m256i, k: __mmask32, x: __m256i, a: __m256i, ) -> __m256i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineqb_256(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x32())) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni")] #[cfg_attr(test, assert_instr(gf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm_gf2p8affine_epi64_epi8(x: __m128i, a: __m128i) -> __m128i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineqb_128(x, a, b); transmute(r) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_maskz_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm_maskz_gf2p8affine_epi64_epi8( k: __mmask16, x: __m128i, a: __m128i, ) -> __m128i { static_assert_imm8!(B); let b = B as u8; let zero = _mm_setzero_si128().as_i8x16(); let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineqb_128(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the packed bytes in x. /// That is computes a*x+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_mask_gf2p8affine_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm_mask_gf2p8affine_epi64_epi8( src: __m128i, k: __mmask16, x: __m128i, a: __m128i, ) -> __m128i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineqb_128(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x16())) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm512_gf2p8affineinv_epi64_epi8(x: __m512i, a: __m512i) -> __m512i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineinvqb_512(x, a, b); transmute(r) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_maskz_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm512_maskz_gf2p8affineinv_epi64_epi8( k: __mmask64, x: __m512i, a: __m512i, ) -> __m512i { static_assert_imm8!(B); let b = B as u8; let zero = _mm512_setzero_si512().as_i8x64(); let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineinvqb_512(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_mask_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512f")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm512_mask_gf2p8affineinv_epi64_epi8( src: __m512i, k: __mmask64, x: __m512i, a: __m512i, ) -> __m512i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x64(); let a = a.as_i8x64(); let r = vgf2p8affineinvqb_512(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x64())) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm256_gf2p8affineinv_epi64_epi8(x: __m256i, a: __m256i) -> __m256i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineinvqb_256(x, a, b); transmute(r) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_maskz_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm256_maskz_gf2p8affineinv_epi64_epi8( k: __mmask32, x: __m256i, a: __m256i, ) -> __m256i { static_assert_imm8!(B); let b = B as u8; let zero = _mm256_setzero_si256().as_i8x32(); let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineinvqb_256(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_mask_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm256_mask_gf2p8affineinv_epi64_epi8( src: __m256i, k: __mmask32, x: __m256i, a: __m256i, ) -> __m256i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x32(); let a = a.as_i8x32(); let r = vgf2p8affineinvqb_256(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x32())) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni")] #[cfg_attr(test, assert_instr(gf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(2)] pub unsafe fn _mm_gf2p8affineinv_epi64_epi8(x: __m128i, a: __m128i) -> __m128i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineinvqb_128(x, a, b); transmute(r) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are zeroed in the result if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_maskz_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(3)] pub unsafe fn _mm_maskz_gf2p8affineinv_epi64_epi8( k: __mmask16, x: __m128i, a: __m128i, ) -> __m128i { static_assert_imm8!(B); let b = B as u8; let zero = _mm_setzero_si128().as_i8x16(); let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineinvqb_128(x, a, b); transmute(simd_select_bitmask(k, r, zero)) } /// Performs an affine transformation on the inverted packed bytes in x. /// That is computes a*inv(x)+b over the Galois Field 2^8 for each packed byte with a being a 8x8 bit matrix /// and b being a constant 8-bit immediate value. /// The inverse of a byte is defined with respect to the reduction polynomial x^8+x^4+x^3+x+1. /// The inverse of 0 is 0. /// Each pack of 8 bytes in x is paired with the 64-bit word at the same position in a. /// /// Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. /// Otherwise the computation result is written into the result. /// /// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_mask_gf2p8affineinv_epi64_epi8) #[inline] #[target_feature(enable = "gfni,avx512bw,avx512vl")] #[cfg_attr(test, assert_instr(vgf2p8affineinvqb, B = 0))] #[rustc_legacy_const_generics(4)] pub unsafe fn _mm_mask_gf2p8affineinv_epi64_epi8( src: __m128i, k: __mmask16, x: __m128i, a: __m128i, ) -> __m128i { static_assert_imm8!(B); let b = B as u8; let x = x.as_i8x16(); let a = a.as_i8x16(); let r = vgf2p8affineinvqb_128(x, a, b); transmute(simd_select_bitmask(k, r, src.as_i8x16())) } #[cfg(test)] mod tests { // The constants in the tests below are just bit patterns. They should not // be interpreted as integers; signedness does not make sense for them, but // __mXXXi happens to be defined in terms of signed integers. #![allow(overflowing_literals)] use core::hint::black_box; use core::intrinsics::size_of; use stdarch_test::simd_test; use crate::core_arch::x86::*; fn mulbyte(left: u8, right: u8) -> u8 { // this implementation follows the description in // https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm512_gf2p8mul_epi8 const REDUCTION_POLYNOMIAL: u16 = 0x11b; let left: u16 = left.into(); let right: u16 = right.into(); let mut carryless_product: u16 = 0; // Carryless multiplication for i in 0..8 { if ((left >> i) & 0x01) != 0 { carryless_product ^= right << i; } } // reduction, adding in "0" where appropriate to clear out high bits // note that REDUCTION_POLYNOMIAL is zero in this context for i in (8..=14).rev() { if ((carryless_product >> i) & 0x01) != 0 { carryless_product ^= REDUCTION_POLYNOMIAL << (i - 8); } } carryless_product as u8 } const NUM_TEST_WORDS_512: usize = 4; const NUM_TEST_WORDS_256: usize = NUM_TEST_WORDS_512 * 2; const NUM_TEST_WORDS_128: usize = NUM_TEST_WORDS_256 * 2; const NUM_TEST_ENTRIES: usize = NUM_TEST_WORDS_512 * 64; const NUM_TEST_WORDS_64: usize = NUM_TEST_WORDS_128 * 2; const NUM_BYTES: usize = 256; const NUM_BYTES_WORDS_128: usize = NUM_BYTES / 16; const NUM_BYTES_WORDS_256: usize = NUM_BYTES_WORDS_128 / 2; const NUM_BYTES_WORDS_512: usize = NUM_BYTES_WORDS_256 / 2; fn parity(input: u8) -> u8 { let mut accumulator = 0; for i in 0..8 { accumulator ^= (input >> i) & 0x01; } accumulator } fn mat_vec_multiply_affine(matrix: u64, x: u8, b: u8) -> u8 { // this implementation follows the description in // https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_gf2p8affine_epi64_epi8 let mut accumulator = 0; for bit in 0..8 { accumulator |= parity(x & matrix.to_le_bytes()[bit]) << (7 - bit); } accumulator ^ b } fn generate_affine_mul_test_data( immediate: u8, ) -> ( [u64; NUM_TEST_WORDS_64], [u8; NUM_TEST_ENTRIES], [u8; NUM_TEST_ENTRIES], ) { let mut left: [u64; NUM_TEST_WORDS_64] = [0; NUM_TEST_WORDS_64]; let mut right: [u8; NUM_TEST_ENTRIES] = [0; NUM_TEST_ENTRIES]; let mut result: [u8; NUM_TEST_ENTRIES] = [0; NUM_TEST_ENTRIES]; for i in 0..NUM_TEST_WORDS_64 { left[i] = (i as u64) * 103 * 101; for j in 0..8 { let j64 = j as u64; right[i * 8 + j] = ((left[i] + j64) % 256) as u8; result[i * 8 + j] = mat_vec_multiply_affine(left[i], right[i * 8 + j], immediate); } } (left, right, result) } fn generate_inv_tests_data() -> ([u8; NUM_BYTES], [u8; NUM_BYTES]) { let mut input: [u8; NUM_BYTES] = [0; NUM_BYTES]; let mut result: [u8; NUM_BYTES] = [0; NUM_BYTES]; for i in 0..NUM_BYTES { input[i] = (i % 256) as u8; result[i] = if i == 0 { 0 } else { 1 }; } (input, result) } const AES_S_BOX: [u8; NUM_BYTES] = [ 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5, 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76, 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0, 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0, 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc, 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15, 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a, 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75, 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0, 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84, 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b, 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf, 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85, 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8, 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5, 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2, 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17, 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73, 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88, 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb, 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c, 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79, 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9, 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08, 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6, 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a, 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e, 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e, 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94, 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf, 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68, 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16, ]; fn generate_byte_mul_test_data() -> ( [u8; NUM_TEST_ENTRIES], [u8; NUM_TEST_ENTRIES], [u8; NUM_TEST_ENTRIES], ) { let mut left: [u8; NUM_TEST_ENTRIES] = [0; NUM_TEST_ENTRIES]; let mut right: [u8; NUM_TEST_ENTRIES] = [0; NUM_TEST_ENTRIES]; let mut result: [u8; NUM_TEST_ENTRIES] = [0; NUM_TEST_ENTRIES]; for i in 0..NUM_TEST_ENTRIES { left[i] = (i % 256) as u8; right[i] = left[i].wrapping_mul(101); result[i] = mulbyte(left[i], right[i]); } (left, right, result) } #[target_feature(enable = "sse2")] unsafe fn load_m128i_word(data: &[T], word_index: usize) -> __m128i { let byte_offset = word_index * 16 / size_of::(); let pointer = data.as_ptr().add(byte_offset) as *const __m128i; _mm_loadu_si128(black_box(pointer)) } #[target_feature(enable = "avx")] unsafe fn load_m256i_word(data: &[T], word_index: usize) -> __m256i { let byte_offset = word_index * 32 / size_of::(); let pointer = data.as_ptr().add(byte_offset) as *const __m256i; _mm256_loadu_si256(black_box(pointer)) } #[target_feature(enable = "avx512f")] unsafe fn load_m512i_word(data: &[T], word_index: usize) -> __m512i { let byte_offset = word_index * 64 / size_of::(); let pointer = data.as_ptr().add(byte_offset) as *const i32; _mm512_loadu_si512(black_box(pointer)) } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_gf2p8mul_epi8() { let (left, right, expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_512 { let left = load_m512i_word(&left, i); let right = load_m512i_word(&right, i); let expected = load_m512i_word(&expected, i); let result = _mm512_gf2p8mul_epi8(left, right); assert_eq_m512i(result, expected); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_maskz_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_512 { let left = load_m512i_word(&left, i); let right = load_m512i_word(&right, i); let result_zero = _mm512_maskz_gf2p8mul_epi8(0, left, right); assert_eq_m512i(result_zero, _mm512_setzero_si512()); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8mul_epi8(left, right); let result_masked = _mm512_maskz_gf2p8mul_epi8(mask_bytes, left, right); let expected_masked = _mm512_mask_blend_epi32(mask_words, _mm512_setzero_si512(), expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_mask_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_512 { let left = load_m512i_word(&left, i); let right = load_m512i_word(&right, i); let result_left = _mm512_mask_gf2p8mul_epi8(left, 0, left, right); assert_eq_m512i(result_left, left); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8mul_epi8(left, right); let result_masked = _mm512_mask_gf2p8mul_epi8(left, mask_bytes, left, right); let expected_masked = _mm512_mask_blend_epi32(mask_words, left, expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_gf2p8mul_epi8() { let (left, right, expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_256 { let left = load_m256i_word(&left, i); let right = load_m256i_word(&right, i); let expected = load_m256i_word(&expected, i); let result = _mm256_gf2p8mul_epi8(left, right); assert_eq_m256i(result, expected); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_maskz_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_256 { let left = load_m256i_word(&left, i); let right = load_m256i_word(&right, i); let result_zero = _mm256_maskz_gf2p8mul_epi8(0, left, right); assert_eq_m256i(result_zero, _mm256_setzero_si256()); let mask_bytes: __mmask32 = 0x0F_F0_FF_00; const MASK_WORDS: i32 = 0b01_10_11_00; let expected_result = _mm256_gf2p8mul_epi8(left, right); let result_masked = _mm256_maskz_gf2p8mul_epi8(mask_bytes, left, right); let expected_masked = _mm256_blend_epi32::(_mm256_setzero_si256(), expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_mask_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_256 { let left = load_m256i_word(&left, i); let right = load_m256i_word(&right, i); let result_left = _mm256_mask_gf2p8mul_epi8(left, 0, left, right); assert_eq_m256i(result_left, left); let mask_bytes: __mmask32 = 0x0F_F0_FF_00; const MASK_WORDS: i32 = 0b01_10_11_00; let expected_result = _mm256_gf2p8mul_epi8(left, right); let result_masked = _mm256_mask_gf2p8mul_epi8(left, mask_bytes, left, right); let expected_masked = _mm256_blend_epi32::(left, expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_gf2p8mul_epi8() { let (left, right, expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_128 { let left = load_m128i_word(&left, i); let right = load_m128i_word(&right, i); let expected = load_m128i_word(&expected, i); let result = _mm_gf2p8mul_epi8(left, right); assert_eq_m128i(result, expected); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_maskz_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_128 { let left = load_m128i_word(&left, i); let right = load_m128i_word(&right, i); let result_zero = _mm_maskz_gf2p8mul_epi8(0, left, right); assert_eq_m128i(result_zero, _mm_setzero_si128()); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8mul_epi8(left, right); let result_masked = _mm_maskz_gf2p8mul_epi8(mask_bytes, left, right); let expected_masked = _mm_blend_epi32::(_mm_setzero_si128(), expected_result); assert_eq_m128i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_mask_gf2p8mul_epi8() { let (left, right, _expected) = generate_byte_mul_test_data(); for i in 0..NUM_TEST_WORDS_128 { let left = load_m128i_word(&left, i); let right = load_m128i_word(&right, i); let result_left = _mm_mask_gf2p8mul_epi8(left, 0, left, right); assert_eq_m128i(result_left, left); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8mul_epi8(left, right); let result_masked = _mm_mask_gf2p8mul_epi8(left, mask_bytes, left, right); let expected_masked = _mm_blend_epi32::(left, expected_result); assert_eq_m128i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_gf2p8affine_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; let constant: i64 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm512_set1_epi64(identity); let constant = _mm512_set1_epi64(constant); let constant_reference = _mm512_set1_epi8(CONSTANT_BYTE as i8); let (bytes, more_bytes, _) = generate_byte_mul_test_data(); let (matrices, vectors, references) = generate_affine_mul_test_data(IDENTITY_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let data = load_m512i_word(&bytes, i); let result = _mm512_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m512i(result, data); let result = _mm512_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m512i(result, constant_reference); let data = load_m512i_word(&more_bytes, i); let result = _mm512_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m512i(result, data); let result = _mm512_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m512i(result, constant_reference); let matrix = load_m512i_word(&matrices, i); let vector = load_m512i_word(&vectors, i); let reference = load_m512i_word(&references, i); let result = _mm512_gf2p8affine_epi64_epi8::(vector, matrix); assert_eq_m512i(result, reference); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_maskz_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let matrix = load_m512i_word(&matrices, i); let vector = load_m512i_word(&vectors, i); let result_zero = _mm512_maskz_gf2p8affine_epi64_epi8::(0, vector, matrix); assert_eq_m512i(result_zero, _mm512_setzero_si512()); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8affine_epi64_epi8::(vector, matrix); let result_masked = _mm512_maskz_gf2p8affine_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm512_mask_blend_epi32(mask_words, _mm512_setzero_si512(), expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_mask_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let left = load_m512i_word(&vectors, i); let right = load_m512i_word(&matrices, i); let result_left = _mm512_mask_gf2p8affine_epi64_epi8::(left, 0, left, right); assert_eq_m512i(result_left, left); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8affine_epi64_epi8::(left, right); let result_masked = _mm512_mask_gf2p8affine_epi64_epi8::(left, mask_bytes, left, right); let expected_masked = _mm512_mask_blend_epi32(mask_words, left, expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_gf2p8affine_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; let constant: i64 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm256_set1_epi64x(identity); let constant = _mm256_set1_epi64x(constant); let constant_reference = _mm256_set1_epi8(CONSTANT_BYTE as i8); let (bytes, more_bytes, _) = generate_byte_mul_test_data(); let (matrices, vectors, references) = generate_affine_mul_test_data(IDENTITY_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let data = load_m256i_word(&bytes, i); let result = _mm256_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m256i(result, data); let result = _mm256_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m256i(result, constant_reference); let data = load_m256i_word(&more_bytes, i); let result = _mm256_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m256i(result, data); let result = _mm256_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m256i(result, constant_reference); let matrix = load_m256i_word(&matrices, i); let vector = load_m256i_word(&vectors, i); let reference = load_m256i_word(&references, i); let result = _mm256_gf2p8affine_epi64_epi8::(vector, matrix); assert_eq_m256i(result, reference); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_maskz_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let matrix = load_m256i_word(&matrices, i); let vector = load_m256i_word(&vectors, i); let result_zero = _mm256_maskz_gf2p8affine_epi64_epi8::(0, vector, matrix); assert_eq_m256i(result_zero, _mm256_setzero_si256()); let mask_bytes: __mmask32 = 0xFF_0F_F0_00; const MASK_WORDS: i32 = 0b11_01_10_00; let expected_result = _mm256_gf2p8affine_epi64_epi8::(vector, matrix); let result_masked = _mm256_maskz_gf2p8affine_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm256_blend_epi32::(_mm256_setzero_si256(), expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_mask_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let left = load_m256i_word(&vectors, i); let right = load_m256i_word(&matrices, i); let result_left = _mm256_mask_gf2p8affine_epi64_epi8::(left, 0, left, right); assert_eq_m256i(result_left, left); let mask_bytes: __mmask32 = 0xFF_0F_F0_00; const MASK_WORDS: i32 = 0b11_01_10_00; let expected_result = _mm256_gf2p8affine_epi64_epi8::(left, right); let result_masked = _mm256_mask_gf2p8affine_epi64_epi8::(left, mask_bytes, left, right); let expected_masked = _mm256_blend_epi32::(left, expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_gf2p8affine_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; let constant: i64 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm_set1_epi64x(identity); let constant = _mm_set1_epi64x(constant); let constant_reference = _mm_set1_epi8(CONSTANT_BYTE as i8); let (bytes, more_bytes, _) = generate_byte_mul_test_data(); let (matrices, vectors, references) = generate_affine_mul_test_data(IDENTITY_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let data = load_m128i_word(&bytes, i); let result = _mm_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m128i(result, data); let result = _mm_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m128i(result, constant_reference); let data = load_m128i_word(&more_bytes, i); let result = _mm_gf2p8affine_epi64_epi8::(data, identity); assert_eq_m128i(result, data); let result = _mm_gf2p8affine_epi64_epi8::(data, constant); assert_eq_m128i(result, constant_reference); let matrix = load_m128i_word(&matrices, i); let vector = load_m128i_word(&vectors, i); let reference = load_m128i_word(&references, i); let result = _mm_gf2p8affine_epi64_epi8::(vector, matrix); assert_eq_m128i(result, reference); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_maskz_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let matrix = load_m128i_word(&matrices, i); let vector = load_m128i_word(&vectors, i); let result_zero = _mm_maskz_gf2p8affine_epi64_epi8::(0, vector, matrix); assert_eq_m128i(result_zero, _mm_setzero_si128()); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8affine_epi64_epi8::(vector, matrix); let result_masked = _mm_maskz_gf2p8affine_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm_blend_epi32::(_mm_setzero_si128(), expected_result); assert_eq_m128i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_mask_gf2p8affine_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let left = load_m128i_word(&vectors, i); let right = load_m128i_word(&matrices, i); let result_left = _mm_mask_gf2p8affine_epi64_epi8::(left, 0, left, right); assert_eq_m128i(result_left, left); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8affine_epi64_epi8::(left, right); let result_masked = _mm_mask_gf2p8affine_epi64_epi8::(left, mask_bytes, left, right); let expected_masked = _mm_blend_epi32::(left, expected_result); assert_eq_m128i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_gf2p8affineinv_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm512_set1_epi64(identity); // validate inversion let (inputs, results) = generate_inv_tests_data(); for i in 0..NUM_BYTES_WORDS_512 { let input = load_m512i_word(&inputs, i); let reference = load_m512i_word(&results, i); let result = _mm512_gf2p8affineinv_epi64_epi8::(input, identity); let remultiplied = _mm512_gf2p8mul_epi8(result, input); assert_eq_m512i(remultiplied, reference); } // validate subsequent affine operation let (matrices, vectors, _affine_expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let vector = load_m512i_word(&vectors, i); let matrix = load_m512i_word(&matrices, i); let inv_vec = _mm512_gf2p8affineinv_epi64_epi8::(vector, identity); let reference = _mm512_gf2p8affine_epi64_epi8::(inv_vec, matrix); let result = _mm512_gf2p8affineinv_epi64_epi8::(vector, matrix); assert_eq_m512i(result, reference); } // validate everything by virtue of checking against the AES SBox const AES_S_BOX_MATRIX: i64 = 0xF1_E3_C7_8F_1F_3E_7C_F8; let sbox_matrix = _mm512_set1_epi64(AES_S_BOX_MATRIX); for i in 0..NUM_BYTES_WORDS_512 { let reference = load_m512i_word(&AES_S_BOX, i); let input = load_m512i_word(&inputs, i); let result = _mm512_gf2p8affineinv_epi64_epi8::(input, sbox_matrix); assert_eq_m512i(result, reference); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_maskz_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let matrix = load_m512i_word(&matrices, i); let vector = load_m512i_word(&vectors, i); let result_zero = _mm512_maskz_gf2p8affineinv_epi64_epi8::(0, vector, matrix); assert_eq_m512i(result_zero, _mm512_setzero_si512()); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8affineinv_epi64_epi8::(vector, matrix); let result_masked = _mm512_maskz_gf2p8affineinv_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm512_mask_blend_epi32(mask_words, _mm512_setzero_si512(), expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw")] unsafe fn test_mm512_mask_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_512 { let left = load_m512i_word(&vectors, i); let right = load_m512i_word(&matrices, i); let result_left = _mm512_mask_gf2p8affineinv_epi64_epi8::(left, 0, left, right); assert_eq_m512i(result_left, left); let mask_bytes: __mmask64 = 0x0F_0F_0F_0F_FF_FF_00_00; let mask_words: __mmask16 = 0b01_01_01_01_11_11_00_00; let expected_result = _mm512_gf2p8affineinv_epi64_epi8::(left, right); let result_masked = _mm512_mask_gf2p8affineinv_epi64_epi8::( left, mask_bytes, left, right, ); let expected_masked = _mm512_mask_blend_epi32(mask_words, left, expected_result); assert_eq_m512i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_gf2p8affineinv_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm256_set1_epi64x(identity); // validate inversion let (inputs, results) = generate_inv_tests_data(); for i in 0..NUM_BYTES_WORDS_256 { let input = load_m256i_word(&inputs, i); let reference = load_m256i_word(&results, i); let result = _mm256_gf2p8affineinv_epi64_epi8::(input, identity); let remultiplied = _mm256_gf2p8mul_epi8(result, input); assert_eq_m256i(remultiplied, reference); } // validate subsequent affine operation let (matrices, vectors, _affine_expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let vector = load_m256i_word(&vectors, i); let matrix = load_m256i_word(&matrices, i); let inv_vec = _mm256_gf2p8affineinv_epi64_epi8::(vector, identity); let reference = _mm256_gf2p8affine_epi64_epi8::(inv_vec, matrix); let result = _mm256_gf2p8affineinv_epi64_epi8::(vector, matrix); assert_eq_m256i(result, reference); } // validate everything by virtue of checking against the AES SBox const AES_S_BOX_MATRIX: i64 = 0xF1_E3_C7_8F_1F_3E_7C_F8; let sbox_matrix = _mm256_set1_epi64x(AES_S_BOX_MATRIX); for i in 0..NUM_BYTES_WORDS_256 { let reference = load_m256i_word(&AES_S_BOX, i); let input = load_m256i_word(&inputs, i); let result = _mm256_gf2p8affineinv_epi64_epi8::(input, sbox_matrix); assert_eq_m256i(result, reference); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_maskz_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let matrix = load_m256i_word(&matrices, i); let vector = load_m256i_word(&vectors, i); let result_zero = _mm256_maskz_gf2p8affineinv_epi64_epi8::(0, vector, matrix); assert_eq_m256i(result_zero, _mm256_setzero_si256()); let mask_bytes: __mmask32 = 0xFF_0F_F0_00; const MASK_WORDS: i32 = 0b11_01_10_00; let expected_result = _mm256_gf2p8affineinv_epi64_epi8::(vector, matrix); let result_masked = _mm256_maskz_gf2p8affineinv_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm256_blend_epi32::(_mm256_setzero_si256(), expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm256_mask_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_256 { let left = load_m256i_word(&vectors, i); let right = load_m256i_word(&matrices, i); let result_left = _mm256_mask_gf2p8affineinv_epi64_epi8::(left, 0, left, right); assert_eq_m256i(result_left, left); let mask_bytes: __mmask32 = 0xFF_0F_F0_00; const MASK_WORDS: i32 = 0b11_01_10_00; let expected_result = _mm256_gf2p8affineinv_epi64_epi8::(left, right); let result_masked = _mm256_mask_gf2p8affineinv_epi64_epi8::( left, mask_bytes, left, right, ); let expected_masked = _mm256_blend_epi32::(left, expected_result); assert_eq_m256i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_gf2p8affineinv_epi64_epi8() { let identity: i64 = 0x01_02_04_08_10_20_40_80; const IDENTITY_BYTE: i32 = 0; const CONSTANT_BYTE: i32 = 0x63; let identity = _mm_set1_epi64x(identity); // validate inversion let (inputs, results) = generate_inv_tests_data(); for i in 0..NUM_BYTES_WORDS_128 { let input = load_m128i_word(&inputs, i); let reference = load_m128i_word(&results, i); let result = _mm_gf2p8affineinv_epi64_epi8::(input, identity); let remultiplied = _mm_gf2p8mul_epi8(result, input); assert_eq_m128i(remultiplied, reference); } // validate subsequent affine operation let (matrices, vectors, _affine_expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let vector = load_m128i_word(&vectors, i); let matrix = load_m128i_word(&matrices, i); let inv_vec = _mm_gf2p8affineinv_epi64_epi8::(vector, identity); let reference = _mm_gf2p8affine_epi64_epi8::(inv_vec, matrix); let result = _mm_gf2p8affineinv_epi64_epi8::(vector, matrix); assert_eq_m128i(result, reference); } // validate everything by virtue of checking against the AES SBox const AES_S_BOX_MATRIX: i64 = 0xF1_E3_C7_8F_1F_3E_7C_F8; let sbox_matrix = _mm_set1_epi64x(AES_S_BOX_MATRIX); for i in 0..NUM_BYTES_WORDS_128 { let reference = load_m128i_word(&AES_S_BOX, i); let input = load_m128i_word(&inputs, i); let result = _mm_gf2p8affineinv_epi64_epi8::(input, sbox_matrix); assert_eq_m128i(result, reference); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_maskz_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let matrix = load_m128i_word(&matrices, i); let vector = load_m128i_word(&vectors, i); let result_zero = _mm_maskz_gf2p8affineinv_epi64_epi8::(0, vector, matrix); assert_eq_m128i(result_zero, _mm_setzero_si128()); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8affineinv_epi64_epi8::(vector, matrix); let result_masked = _mm_maskz_gf2p8affineinv_epi64_epi8::(mask_bytes, vector, matrix); let expected_masked = _mm_blend_epi32::(_mm_setzero_si128(), expected_result); assert_eq_m128i(result_masked, expected_masked); } } #[simd_test(enable = "gfni,avx512bw,avx512vl")] unsafe fn test_mm_mask_gf2p8affineinv_epi64_epi8() { const CONSTANT_BYTE: i32 = 0x63; let (matrices, vectors, _expected) = generate_affine_mul_test_data(CONSTANT_BYTE as u8); for i in 0..NUM_TEST_WORDS_128 { let left = load_m128i_word(&vectors, i); let right = load_m128i_word(&matrices, i); let result_left = _mm_mask_gf2p8affineinv_epi64_epi8::(left, 0, left, right); assert_eq_m128i(result_left, left); let mask_bytes: __mmask16 = 0x0F_F0; const MASK_WORDS: i32 = 0b01_10; let expected_result = _mm_gf2p8affineinv_epi64_epi8::(left, right); let result_masked = _mm_mask_gf2p8affineinv_epi64_epi8::(left, mask_bytes, left, right); let expected_masked = _mm_blend_epi32::(left, expected_result); assert_eq_m128i(result_masked, expected_masked); } } }