/external/chromium_org/third_party/yasm/source/patched-yasm/modules/arch/x86/tests/ |
clmul.asm | 3 pclmulqdq xmm1, xmm2, 5 label 4 pclmulqdq xmm1, [rax], byte 5 label 5 pclmulqdq xmm1, dqword [rax], 5 label
|
aes.asm | 60 pclmulqdq xmm1, xmm2, 5 label 61 pclmulqdq xmm1, [rax], byte 5 label 62 pclmulqdq xmm1, dqword [rax], 5 label 64 ; pclmulqdq variants
|
/external/chromium_org/third_party/boringssl/src/crypto/modes/asm/ |
ghash-x86_64.pl | 36 # Add PCLMULQDQ version performing at 2.02 cycles per processed byte. 50 # consists of 3 independent pclmulqdq instructions, Karatsuba post- 52 # multiplication operations, triplets of independent pclmulqdq's. As 53 # issue rate for pclmulqdq is limited, it makes lesser sense to 59 # longer. A CPU with higher pclmulqdq issue rate would also benefit 408 # PCLMULQDQ version. 435 pclmulqdq \$0x00,$Hkey,$Xi ####### 436 pclmulqdq \$0x11,$Hkey,$Xhi ####### 437 pclmulqdq \$0x00,$HK,$T1 ####### 588 pclmulqdq \$0x00,$Xi,$T1 # ?(0xE1<<1 [all...] |
ghash-x86.pl | 50 # Add PCLMULQDQ version performing at 2.10 cycles per processed byte. 51 # The question is how close is it to theoretical limit? The pclmulqdq 111 # on same XMM register, PCLMULQDQ subroutine was measured to process 829 # PCLMULQDQ version. 853 &pclmulqdq ($Xi,$Hkey,0x00); ####### 854 &pclmulqdq ($Xhi,$Hkey,0x11); ####### 855 &pclmulqdq ($T1,$HK,0x00); ####### 876 &pclmulqdq ($Xi,$Hkey,0x00); ####### 877 &pclmulqdq ($Xhi,$Hkey,0x11); ####### 882 &pclmulqdq ($T2,$T3,0x00); ###### [all...] |
/external/openssl/crypto/bn/asm/ |
x86_64-gf2m.pl | 15 # for any x86_64 CPU and PCLMULQDQ one suitable for Westmere and 18 # [not very impressive], while PCLMULQDQ - whole 85%-160% better on 187 pclmulqdq \$0,%xmm1,%xmm0 # a1·b1 190 pclmulqdq \$0,%xmm3,%xmm2 # a0·b0 191 pclmulqdq \$0,%xmm5,%xmm4 # (a0+a1)·(b0+b1)
|
x86-gf2m.pl | 16 # and PCLMULQDQ suitable for Westmere and later. Improvement varies 26 # Westmere 53%-121%(PCLMULQDQ)/20%-32%(MMX) 27 # Sandy Bridge 72%-127%(PCLMULQDQ)/27%-23%(MMX) 211 &test ("edx",1<<1); # check PCLMULQDQ bit 216 &pclmulqdq ("xmm0","xmm0",1);
|
/external/openssl/crypto/modes/asm/ |
ghash-x86_64.pl | 34 # Add PCLMULQDQ version performing at 2.02 cycles per processed byte. 349 # PCLMULQDQ version. 368 pclmulqdq \$0x00,$Hkey,$Xi ####### 369 pclmulqdq \$0x11,$Hkey,$Xhi ####### 370 pclmulqdq \$0x00,$T2,$T1 ####### 553 pclmulqdq \$0x00,$Hkey,$Xn ####### 561 pclmulqdq \$0x11,$Hkey,$Xhn ####### 571 pclmulqdq \$0x00,$T2n,$T1n #######
|
ghash-x86.pl | 48 # Add PCLMULQDQ version performing at 2.10 cycles per processed byte. 49 # The question is how close is it to theoretical limit? The pclmulqdq 109 # on same XMM register, PCLMULQDQ subroutine was measured to process 817 # PCLMULQDQ version. 840 &pclmulqdq ($Xi,$Hkey,0x00); ####### 841 &pclmulqdq ($Xhi,$Hkey,0x11); ####### 842 &pclmulqdq ($T1,$T2,0x00); ####### 863 &pclmulqdq ($Xi,$Hkey,0x00); ####### 864 &pclmulqdq ($Xhi,$Hkey,0x11); ####### 869 &pclmulqdq ($T2,$T3,0x00); ###### [all...] |
/external/openssl/crypto/perlasm/ |
x86asm.pl | 118 sub ::pclmulqdq 123 { &::generic("pclmulqdq",@_); }
|
x86_64-xlate.pl | 749 my $pclmulqdq = sub { [all...] |
/external/chromium_org/third_party/boringssl/src/crypto/perlasm/ |
x86asm.pl | 118 sub ::pclmulqdq 123 { &::generic("pclmulqdq",@_); }
|
x86_64-xlate.pl | 766 my $pclmulqdq = sub { [all...] |
/external/llvm/test/MC/X86/ |
x86-64.s | 1231 // CHECK: pclmulqdq $17, %xmm0, %xmm1 1235 // CHECK: pclmulqdq $1, %xmm0, %xmm1 1237 pclmulqdq $1, %xmm0, %xmm1 label 1239 // CHECK: pclmulqdq $16, (%rdi), %xmm1 1243 // CHECK: pclmulqdq $0, (%rdi), %xmm1 1245 pclmulqdq $0, (%rdi), %xmm1 label
|
/external/chromium_org/third_party/yasm/source/patched-yasm/modules/arch/x86/ |
x86cpu.gperf | 371 pclmulqdq, x86_cpu_set, CPU_CLMUL
|
/external/qemu/target-i386/ |
ops_sse_header.h | 347 DEF_HELPER_4(glue(pclmulqdq, SUFFIX), void, env, Reg, Reg, i32)
|
/external/chromium_org/third_party/yasm/source/patched-yasm/ |
x86insn_nasm.gperf | 483 pclmulqdq, pclmulqdq_insn, 2, SUF_Z, 0x3A, 0x44, 0, 0, CPU_AVX, 0, 0 [all...] |
x86insn_gas.gperf | [all...] |