I am also working on a different patch that uses the "pcmpestri" SSE4 processor instruction, it looks like this for now.

While at it I realized there is (maybe) another potential speedup: avoiding the ucs4lib_find_max_char we do for each chunk of the string ( that entails scanning the string in memory one more time)... anyways that's another (much longer) story, probably for another issue?


```
diff --git a/Modules/_json.c b/Modules/_json.c
index 38beb6f50d..25b1cf4a99 100644
--- a/Modules/_json.c
+++ b/Modules/_json.c
@@ -400,6 +400,38 @@ _build_rval_index_tuple(PyObject *rval, Py_ssize_t idx) {
         Py_CLEAR(chunk); \
     }
 
+
+inline unsigned int
+_fast_search(const void *needle, unsigned int needle_len, const void *haystack, unsigned int haystack_len)
+{
+  unsigned int pos;
+  __asm__ __volatile__("movq (%1), %%xmm1;\n"
+                       "mov %2, %%eax;\n"
+                       "movq %3, %%r8;\n"
+                       "mov %4, %%edx;\n"
+                       ".intel_syntax noprefix;\n"
+                       "loop: pcmpestri xmm1, [r8], 0;\n" /* 0 = equal any */
+                       /* "pcmpestri %%mm1, (%%r8), $0;\n" /\* 0 = equal any *\/ */
+                       ".att_syntax prefix;\n"
+                       "cmp $15, %%ecx;\n"
+                       "jbe found;\n"
+                       "sub $16, %%edx;\n"
+                       "jnge notfound;\n"
+                       "add $16, %%r8;\n"
+                       "jmp loop;\n"
+                       "notfound: movl %4, %%ecx;\n"
+                       "jmp exit;\n"
+                       "found: mov %4, %%eax;\n"
+                       "sub %%edx, %%eax;\n"
+                       "add %%eax, %%ecx;\n"
+                       "exit: mov %%ecx, %0;\n"
+                       :"=m"(pos)
+                       :"r"(needle), "r"(needle_len), "r"(haystack), "r"(haystack_len)
+                       :"%eax", "%edx", "%ecx", "%r8", "%xmm1");
+  return pos;
+}
+
+
 static PyObject *
 scanstring_unicode(PyObject *pystr, Py_ssize_t end, int strict, Py_ssize_t *next_end_ptr)
 {
@@ -431,17 +463,26 @@ scanstring_unicode(PyObject *pystr, Py_ssize_t end, int strict, Py_ssize_t *next
         PyErr_SetString(PyExc_ValueError, "end is out of bounds");
         goto bail;
     }
+    char needle[2];
+    needle[0] = '"';
+    needle[1] = '\\';
     while (1) {
         /* Find the end of the string or the next escape */
         Py_UCS4 c = 0;
-        for (next = end; next < len; next++) {
+        if (kind == PyUnicode_1BYTE_KIND) {
+          next = _fast_search(needle, 2, buf+end, len-end) + end;
+          if (next < len)
             c = PyUnicode_READ(kind, buf, next);
-            if (c == '"' || c == '\\') {
-                break;
-            }
-            else if (strict && c <= 0x1f) {
-                raise_errmsg("Invalid control character at", pystr, next);
-                goto bail;
+        } else {
+            for (next = end; next < len; next++) {
+                c = PyUnicode_READ(kind, buf, next);
+                if (c == '"' || c == '\\') {
+                    break;
+                }
+                else if (strict && c <= 0x1f) {
+                    raise_errmsg("Invalid control character at", pystr, next);
+                    goto bail;
+                }
             }
         }
         if (!(c == '"' || c == '\\')) {
```