Code

Feature/add arm support#286

Code

Closed

markos wants to merge 53 commits intointel:masterintel/hyperscan:masterfrom VectorCamp:feature/add-arm-supportVectorCamp/vectorscan:feature/add-arm-supportCopy head branch name to clipboard

+3,594-1,952

Commits

Commits on Sep 17, 2020

• move x86 arch and SIMD types to x86 arch folder

  markos

  committed
  2d89df4

  Copy full SHA for 2d89df4

• move cpuid stuff to util/arch/x86

  markos

  committed
  6a40793

  Copy full SHA for 6a40793

Commits on Sep 18, 2020

• move crc32 SSE42 implementation to util/arch/x86

  markos

  committed
  ea721c9

  Copy full SHA for ea721c9

• move masked_move* AVX2 implementation to util/arch/x86

  markos

  committed
  956b001

  Copy full SHA for 956b001

• fix include paths for masked_move

  markos

  committed
  8ed5f4a

  Copy full SHA for 8ed5f4a

Commits on Sep 22, 2020

• move x86 bitutils.h implementations to util/arch/x86/bitutils.h

  markos

  committed
  aac1f0f

  Copy full SHA for aac1f0f

• move x86 popcount.h implementations to util/arch/x86/popcount.h

  markos

  committed
  6581aae

  Copy full SHA for 6581aae

• move andn helper function to bitutils.h

  markos

  committed
  9f3ad89

  Copy full SHA for 9f3ad89

• no need to check for WIN32*

  markos

  committed
  e915d84

  Copy full SHA for e915d84

• move x86 implementations of simd_utils.h to util/arch/x86/

  markos

  committed
  e8e188a

  Copy full SHA for e8e188a

Commits on Sep 23, 2020

• add some set*() functions, harmonize names, rename setAxB to set1_AxB when using mm_set1_* internally

  markos

  committed
  f7a6b89

  Copy full SHA for f7a6b89

• fix names, use own intrinsic instead of explicit _mm* ones

  markos

  committed
  5333467

  Copy full SHA for 5333467

Commits on Sep 24, 2020

• Revert "move x86 popcount.h implementations to util/arch/x86/popcount.h"

  Show description for 04fbf24
  markos

  committed
  04fbf24

  Copy full SHA for 04fbf24

• Revert "Revert "move x86 popcount.h implementations to util/arch/x86/popcount.h""

  Show description for f0e70bc
  markos

  committed
  f0e70bc

  Copy full SHA for f0e70bc

Commits on Oct 6, 2020

• add arm checks in platform.cmake

  markos

  committed
  b1170bc

  Copy full SHA for b1170bc

• add necessary modifications to CMake system to enable building on ARM, add arm_neon.h intrinsic header to intrinsics.h

  markos

  committed
  5952c64

  Copy full SHA for 5952c64

• use right intrinsic

  markos

  committed
  e91082d

  Copy full SHA for e91082d

Commits on Oct 7, 2020

• minor fix

  markos

  committed
  9a04942

  Copy full SHA for 9a04942

• add arm architecture basic defines

  markos

  committed
  4c924cc

  Copy full SHA for 4c924cc

• use C implementation of popcount for arm

  markos

  committed
  5d773dd

  Copy full SHA for 5d773dd

Commits on Oct 8, 2020

• move cpuid_flags.h header to common

  markos

  committed
  d2cf1a7

  Copy full SHA for d2cf1a7

• add C implementation of pdep64()

  markos

  committed
  1c2c73b

  Copy full SHA for 1c2c73b

• add arm bitutils.h header

  markos

  committed
  a921217

  Copy full SHA for a921217

Commits on Oct 13, 2020

• add ARM version of simd_utils.h

  markos

  committed
  31ac671

  Copy full SHA for 31ac671

Commits on Oct 15, 2020

• add arm simple cpuid_flags

  markos

  committed
  5b425bd

  Copy full SHA for 5b425bd

• add ARM simd_utils vectorized functions for 128-bit vectors

  markos

  committed
  c5a7f4b

  Copy full SHA for c5a7f4b

• add scalar versions of the vectorized functions for architectures that don't support 256-bit/512-bit SIMD vectors such as ARM

  markos

  committed
  45bfed9

  Copy full SHA for 45bfed9

Commits on Oct 16, 2020

• fix compilation paths for cpuid_flags for x86

  markos

  committed
  e7e1308

  Copy full SHA for e7e1308

• split arch-agnostic simd_utils.h functions into the common file

  markos

  committed
  83977db

  Copy full SHA for 83977db

• Revert "move x86 popcount.h implementations to util/arch/x86/popcount.h"

  Show description for 4bce012
  markos

  committed
  4bce012

  Copy full SHA for 4bce012

• scalar implementations of diffrich256 and diffrich384

  markos

  committed
  c4db636

  Copy full SHA for c4db636

• don't redefine function on x86

  markos

  committed
  149ea93

  Copy full SHA for 149ea93

Commits on Oct 30, 2020

• don't use SSE directly in the tests

  markos

  committed
  0bef151

  Copy full SHA for 0bef151

• fix ARM implementations

  markos

  committed
  5482429

  Copy full SHA for 5482429

• small optimization in storecompress*()

  markos

  committed
  547f79b

  Copy full SHA for 547f79b

• needed for ARM vector type conversions

  markos

  committed
  592b190

  Copy full SHA for 592b190

Commits on Nov 5, 2020

• fix 32-bit/64-bit detection

  markos

  committed
  18296ee

  Copy full SHA for 18296ee

• add extra instructions (currently arm-only), fix order of elements in set4x32/set2x64

  markos

  committed
  7b8cf97

  Copy full SHA for 7b8cf97

• add compress128 function and implementation

  markos

  committed
  3390418

  Copy full SHA for 3390418

• add some debug info

  markos

  committed
  501f60e

  Copy full SHA for 501f60e

• add some debug and minor optimizations in unit test

  markos

  committed
  62fed20

  Copy full SHA for 62fed20

• remove debug from functions

  markos

  committed
  c4f1372

  Copy full SHA for c4f1372

Commits on Nov 24, 2020

• fix compiler flag testcase

  markos

  committed
  606c53a

  Copy full SHA for 606c53a

• when building in debug mode, vgetq_lane_*() and vextq_*() need immediate operands, and we have to use switch()'ed versions

  markos

  committed
  1c26f04

  Copy full SHA for 1c26f04

• helper functions to print a m128 vector in debug mode

  markos

  committed
  d763652

  Copy full SHA for d763652

• small optimization that was for some reason failing in ARM, should be faster anyway

  markos

  committed
  17ab42d

  Copy full SHA for 17ab42d

Commits on Dec 3, 2020

• define debug vector print functions to NULL in non-debug mode

  markos

  committed
  259c257

  Copy full SHA for 259c257

• fix movq and load_m128_from_u64a and resp. test for NEON

  markos

  committed
  38477b0

  Copy full SHA for 38477b0

• add ARM platform

  markos

  committed
  c38722a

  Copy full SHA for c38722a

• clear zones array

  markos

  committed
  39945b7

  Copy full SHA for 39945b7

Commits on Dec 7, 2020

• optimize *shiftbyte_m128() functions to use palign instead of variable_byte_shift_m128()

  markos

  committed
  773dc6f

  Copy full SHA for 773dc6f

• remove forgotten printf

  markos

  committed
  e088c6a

  Copy full SHA for e088c6a

Commits on Dec 8, 2020

• fix x86 compilation

  markos

  committed
  61b963a

  Copy full SHA for 61b963a