Use xxhash for byte hashing

haskell-unordered-containers · Jun 8, 2024 · 2a12cf5 · 2a12cf5
1 parent 7837d1d
commit 2a12cf5
Show file tree

Hide file tree

Showing 18 changed files with 7,297 additions and 200 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -3,6 +3,9 @@ See also https://pvp.haskell.org/faq
 ## Version 1.4.5.0
 
   * Drop support for GHCs prior 8.6.5
+  * Use xxhash for hashing bytestrings and bytearrays.
+    Note: when compiling binaries for distribution, you may need to disable
+    `arch-native` flag.
 
 ## Version 1.4.4.0
 

diff --git a/cbits/fnv.c b/cbits/fnv.c
diff --git a/hashable-bench/hashable-bench.cabal b/hashable-bench/hashable-bench.cabal
@@ -34,6 +34,8 @@ tested-with:
 
 extra-source-files:
   include/HsHashable.h
+  include/HsXXHash.h
+  xxHash-0.8.2/xxhash.h
 
 flag integer-gmp
   description:
@@ -50,13 +52,19 @@ library
 
   other-modules:
     Data.Hashable.Class
+    Data.Hashable.FFI
     Data.Hashable.Generic.Instances
     Data.Hashable.Imports
     Data.Hashable.LowLevel
     Data.Hashable.Mix
+    Data.Hashable.XXH3
+
+  include-dirs:     include xxHash-0.8.2/
+  includes:
+    HsHashable.h
+    HsXXHash.h
+    xxhash.h
 
-  c-sources:        cbits/fnv.c
-  include-dirs:     include
   hs-source-dirs:   src
   build-depends:
       base        >=4.10.1.0 && <4.21
@@ -116,6 +124,7 @@ library
     UnliftedFFITypes
 
   ghc-options:      -Wall -fwarn-tabs
+  ghc-options:      -optc=-march=native -optc-mtune=native
 
 benchmark hashable-benchmark
   -- We cannot depend on the hashable library directly as that creates
@@ -143,7 +152,7 @@ benchmark hashable-benchmark
   if impl(ghc >=7.2.1)
     cpp-options: -DGENERICS
 
-  ghc-options:      -Wall -O2
+  ghc-options:      -Wall
   default-language: Haskell2010
 
 source-repository head

diff --git a/hashable-bench/xxHash-0.8.2 b/hashable-bench/xxHash-0.8.2
@@ -0,0 +1 @@
+../xxHash-0.8.2
diff --git a/hashable.cabal b/hashable.cabal
@@ -47,7 +47,9 @@ tested-with:
 extra-source-files:
   CHANGES.md
   include/HsHashable.h
+  include/HsXXHash.h
   README.md
+  xxHash-0.8.2/xxhash.h
 
 flag integer-gmp
   description:
@@ -56,6 +58,14 @@ flag integer-gmp
   manual:      False
   default:     True
 
+flag arch-native
+  description:
+    Use @-march=native@ when compiling C sources.
+    You may need to disable this flag if you are building distributable binaries
+
+  manual:      True
+  default:     True
+
 flag random-initial-seed
   description:
     Randomly initialize the initial seed on each final executable invocation
@@ -74,13 +84,19 @@ library
 
   other-modules:
     Data.Hashable.Class
+    Data.Hashable.FFI
     Data.Hashable.Generic.Instances
     Data.Hashable.Imports
     Data.Hashable.LowLevel
     Data.Hashable.Mix
+    Data.Hashable.XXH3
+
+  include-dirs:     include xxHash-0.8.2
+  includes:
+    HsHashable.h
+    HsXXHash.h
+    xxhash.h
 
-  c-sources:        cbits/fnv.c
-  include-dirs:     include
   hs-source-dirs:   src
   build-depends:
     , base        >=4.12.0.0 && <4.21
@@ -149,7 +165,12 @@ library
     TypeOperators
     UnliftedFFITypes
 
-  ghc-options:      -Wall -fwarn-tabs
+  ghc-options:      -Wall
+
+  if flag(arch-native)
+    -- Cabal doesn't pass cc-options to "ordinary" Haskell source compilation
+    -- https://github.com/haskell/cabal/issues/9801
+    ghc-options: -optc=-march=native -optc-mtune=native
 
   if impl(ghc >=9.0)
     -- these flags may abort compilation with GHC-8.10
@@ -190,6 +211,31 @@ test-suite hashable-tests
   ghc-options:      -Wall -fno-warn-orphans
   default-language: Haskell2010
 
+test-suite xxhash-tests
+  type:             exitcode-stdio-1.0
+  hs-source-dirs:   tests src
+  main-is:          xxhash-tests.hs
+  other-modules:
+    Data.Hashable.FFI
+    Data.Hashable.XXH3
+
+  default-language: Haskell2010
+  build-depends:
+    , base
+    , bytestring
+    , primitive         ^>=0.9.0.0
+    , tasty             ^>=1.5
+    , tasty-hunit       ^>=0.10.1
+    , tasty-quickcheck  ^>=0.10.3
+
+  include-dirs:     include xxHash-0.8.2
+  includes:
+    HsXXHash.h
+    xxhash.h
+
+  if !impl(ghc >=9.4)
+    build-depends: data-array-byte >=0.1.0.1 && <0.2
+
 test-suite hashable-examples
   type:             exitcode-stdio-1.0
   build-depends:

diff --git a/include/HsHashable.h b/include/HsHashable.h
@@ -1,22 +1,8 @@
 #ifndef HS_HASHABLE_H
 #define HS_HASHABLE_H
 
-#include "MachDeps.h"
 #include <stdint.h>
 
-#if WORD_SIZE_IN_BITS == 64
-#define FNV_PRIME 1099511628211
-#define FNV_SIGNED int64_t
-#define FNV_UNSIGNED uint64_t
-#else
-#define FNV_PRIME 16777619
-#define FNV_SIGNED int32_t
-#define FNV_UNSIGNED uint32_t
-#endif
-
 uint64_t hs_hashable_init();
 
-FNV_UNSIGNED hashable_fnv_hash(const unsigned char* str, FNV_SIGNED len, FNV_UNSIGNED salt);
-FNV_UNSIGNED hashable_fnv_hash_offset(const unsigned char* str, FNV_SIGNED offset, FNV_SIGNED len, FNV_UNSIGNED salt);
-
 #endif
diff --git a/include/HsXXHash.h b/include/HsXXHash.h
@@ -0,0 +1,35 @@
+#ifndef HS_XXHASH_H
+#define HS_XXHASH_H
+
+#include <stdint.h>
+
+#define XXH_INLINE_ALL
+#include "xxhash.h"
+
+#define hs_XXH3_sizeof_state_s sizeof(struct XXH3_state_s)
+
+static inline uint64_t hs_XXH3_64bits_withSeed_offset(const uint8_t *ptr, size_t off, size_t len, uint64_t seed) {
+    return XXH3_64bits_withSeed(ptr + off, len, seed);
+}
+
+static inline uint64_t hs_XXH3_64bits_withSeed_u64(uint64_t val, uint64_t seed) {
+    return XXH3_64bits_withSeed(&val, sizeof(val), seed);
+}
+
+static inline uint64_t hs_XXH3_64bits_withSeed_u32(uint32_t val, uint64_t seed) {
+    return XXH3_64bits_withSeed(&val, sizeof(val), seed);
+}
+
+static inline void hs_XXH3_64bits_update_offset(XXH3_state_t *statePtr, const uint8_t *ptr, size_t off, size_t len) {
+    XXH3_64bits_update(statePtr, ptr + off, len);
+}
+
+static inline void hs_XXH3_64bits_update_u64(XXH3_state_t *statePtr, uint64_t val) {
+    XXH3_64bits_update(statePtr, &val, sizeof(val));
+}
+
+static inline void hs_XXH3_64bits_update_u32(XXH3_state_t *statePtr, uint32_t val) {
+    XXH3_64bits_update(statePtr, &val, sizeof(val));
+}
+
+#endif /* HS_XXHASH_H */