Some small tweaks. Simplified hash32 function, makes it faster even if distribution is worse (good enough).

3 files changed, 17 insertions, 18 deletions

diff --git a/benchmark.c b/benchmark.c
index 9bf3722d..2e5d2b34 100644
--- a/benchmark.c
+++ b/benchmark.c
@@ -13,8 +13,7 @@
 // Visual Studio: compile with -TP to force C++:  cl -TP -EHsc -O2 benchmark.c

 

 declare_CMap(ii, int64_t, int64_t, c_noDestroy, c_lowbias32Hash);

-declare_CVector_string(s);

-declare_CMap(ix, short, short); // sizeof(CMapBucket_ix) = 6 bytes only!

+declare_CMap(ix, short, short); // sizeof(CMapEntry_ix) = 6 bytes only!

 

 const size_t seed = 123; // time(NULL);

 const double maxLoadFactor = 0.77;

@@ -64,6 +63,7 @@ const double maxLoadFactor = 0.77;
 const size_t N1 = 7000000;

 const size_t N2 = 10000000;

 #define      RR   24

+int rr = RR;

 

 #define MAP_TEST1(M, tag) \

 { \

@@ -72,8 +72,8 @@ const size_t N2 = 10000000;
     srand(seed); \

     clock_t difference, before = clock(); \

     for (size_t i = 0; i < N1; ++i) { \

-        checksum += ++M##_PUT(tag, RAND(RR), i); \

-        erased += M##_DEL(tag, RAND(RR)); \

+        checksum += ++M##_PUT(tag, RAND(rr), i); \

+        erased += M##_DEL(tag, RAND(rr)); \

     } \

     difference = clock() - before; \

     printf(#M "(" #tag "): sz: %llu, bucks: %llu, time: %.02f, sum: %llu, erase: %llu\n", M##_SIZE(tag), M##_BUCKETS(tag), (float) difference / CLOCKS_PER_SEC, checksum, erased); \

@@ -96,9 +96,10 @@ const size_t N2 = 10000000;
 }

 

 

-int main()

+int main(int argc, char* argv[])

 {

-    printf("\nmap<uint64_t, uint64_t>: Insert + remove %llu random keys in range 0 - 2^%d:\n", N1, RR); \

+    rr = argc == 2 ? atoi(argv[1]) : RR;

+    printf("\nmap<uint64_t, uint64_t>: Insert + remove %llu random keys in range 0 - 2^%d:\n", N1, rr);

     MAP_TEST1(CMAP, ii)

 #ifdef __cplusplus

     MAP_TEST1(UMAP, ii)

@@ -107,7 +108,7 @@ int main()
     MAP_TEST1(RMAP, ii)

 #endif

 

-    printf("\nmap<uint64_t, uint64_t>: Insert %llu keys, THEN remove them in same order:\n", N2); \

+    printf("\nmap<uint64_t, uint64_t>: Insert %llu keys, THEN remove them in same order:\n", N2);

     MAP_TEST2(CMAP, ii)

 #ifdef __cplusplus

     MAP_TEST2(UMAP, ii)

@@ -116,11 +117,4 @@ int main()
     MAP_TEST2(RMAP, ii)

 #endif

 

-    CMap_ix small = cmap_init;

-    cmap_ix_put(&small, 80, 800);

-    cmap_ix_put(&small, 10, 100);

-    cmap_ix_put(&small, 30, 300);

-    

-    c_foreach (i, cmap_ix, small)

-        printf("%d: %d\n", i.item->key, i.item->value);

 }

diff --git a/stc/cdefs.h b/stc/cdefs.h
index c1635fa5..1c75abcc 100644
--- a/stc/cdefs.h
+++ b/stc/cdefs.h
@@ -81,9 +81,9 @@ static inline uint32_t c_lowbias32Hash(const void *data, size_t len) {
     do {

         x ^= *key++ >> 16;

         x *= UINT32_C(0x7feb352d);

-        x ^= x >> 15;

-        x *= UINT32_C(0x846ca68b);

-        x ^= x >> 16;

+        //x ^= x >> 15;

+        //x *= UINT32_C(0x846ca68b);

+        //x ^= x >> 16;

     } while (len -= 4);

     return x;

 }

diff --git a/stc/cvector.h b/stc/cvector.h
index 0e64fd86..86bdfb13 100644
--- a/stc/cvector.h
+++ b/stc/cvector.h
@@ -184,7 +184,7 @@ cvector_##tag##_sortCompare(const void* x, const void* y) { \
     ValueRaw ry = valueGetRaw((const Value *) y); \

     return valueCompareRaw(&rx, &ry); \

 } \

-extern void qsort(void *base, size_t nitems, size_t size, int (*compar)(const void *, const void*)); \

+STC_EXTERN_IMPORT void qsort(void *base, size_t nitems, size_t size, int (*compar)(const void *, const void*)); \

 STC_API void \

 cvector_##tag##_sort(CVector_##tag* self) { \

     size_t len = cvector_size(*self); \

@@ -202,6 +202,11 @@ cvector_##tag##_begin(CVector_##tag* vec) { \
 #define implement_CVector_6(tag, Value, valueDestroy, ValueRaw, valueCompareRaw, valueGetRaw)

 #endif

 

+#if defined(_WIN32) && defined(_DLL)

+#define STC_EXTERN_IMPORT extern __declspec(dllimport)

+#else

+#define STC_EXTERN_IMPORT extern

+#endif

 

 #define _cvector_size(cv) ((size_t *)(cv).data)[-2]

 #define _cvector_capacity(cv) ((size_t *)(cv).data)[-1]