last commit

src/Common/memchrSmall.h

@@ -0,0 +1,46 @@
+#pragma once
+
+#include <bit>
+#include <Common/MemorySanitizer.h>
+#include <base/types.h>
+
+#if defined(__SSE2__)
+#    include <emmintrin.h>
+
+namespace detail
+{
+inline const char * memchrSmallAllowOverflow15Impl(const char * s, int c, ssize_t n)
+{
+    __msan_unpoison_overflow_15(s, n);
+
+    __m128i c16 = _mm_set1_epi8(c);
+    while (n > 0)
+    {
+        __m128i block = _mm_loadu_si128(reinterpret_cast<const __m128i *>(s));
+        UInt16 mask = _mm_movemask_epi8(_mm_cmpeq_epi8(block, c16));
+        if (mask)
+        {
+            auto offset = std::countl_zero(mask);
+            return offset < n ? s + offset : nullptr;
+        }
+
+        s += 16;
+        n -= 16;
+    }
+
+    return nullptr;
+}
+}
+
+/// Works under assumption, that it's possible to read up to 15 excessive bytes after end of 's' region
+inline const void * memchrSmallAllowOverflow15(const void * s, int c, size_t n)
+{
+    return detail::memchrSmallAllowOverflow15Impl(reinterpret_cast<const char *>(s), c, n);
+}
+
+#else
+inline const void * memchrSmallAllowOverflow15(const void * s, int c, size_t n)
+{
+    return memchr(s, c, n);
+}
+#endif

src/Functions/splitByChar.cpp

@@ -4,6 +4,7 @@
 #include <Functions/FunctionFactory.h>
 #include <Common/StringUtils/StringUtils.h>
 #include <Common/assert_cast.h>
+#include <Common/memchrSmall.h>
 
 
 namespace DB
@@ -93,7 +94,7 @@ public:
 
     bool get(Pos & token_begin, Pos & token_end)
     {
-        if (!pos)
+        if (!pos) [[unlikely]]
             return false;
 
         token_begin = pos;
@@ -114,7 +115,7 @@ public:
                    return false;
         }
 
-        pos = reinterpret_cast<Pos>(memchr(pos, separator, end - pos));
+        pos = reinterpret_cast<Pos>(memchrSmallAllowOverflow15(pos, separator, end - pos));
         if (pos)
         {
             token_end = pos;

tests/performance/function_tokens.xml

@@ -1,4 +1,5 @@
 <test>
     <query>with 'Many years later as he faced the firing squad, Colonel Aureliano Buendia was to remember that distant afternoon when his father took him to discover ice.' as s select splitByChar(' ', materialize(s)) as w from numbers(1000000)</query>
     <query>with 'Many years later as he faced the firing squad, Colonel Aureliano Buendia was to remember that distant afternoon when his father took him to discover ice.' as s select splitByRegexp(' ', materialize(s)) as w from numbers(1000000)</query>
+    <query>with 'Many years later as he faced the firing squad, Colonel Aureliano Buendia was to remember that distant afternoon when his father took him to discover ice.' as s select splitByRegexp('\s+', materialize(s)) as w from numbers(100000)</query>
 </test>