=== modified file 'deflate.c'
--- deflate.c	2011-03-14 01:01:37 +0000
+++ deflate.c	2011-05-08 14:30:48 +0000
@@ -1060,6 +1060,54 @@
 /* For 80x86 and 680x0, an optimized version will be provided in match.asm or
  * match.S. The code will be functionally equivalent.
  */
+#if defined(__x86_64__) && defined(__GNUC__)
+#  define LM_LEN_REMAINDER (scan < strend-16)
+#  define LM_INNER_CHECK \
+    unsigned int t; \
+    asm ("movdqu (%0), %%xmm0\n\t" \
+        "movdqu (%1), %%xmm1\n\t" \
+        "pcmpeqb %%xmm0, %%xmm1\n\t" \
+        "pmovmskb %%xmm1, %2\n\t" \
+        "xor $0x0000ffff, %2\n\t" \
+        "jnz 2f\n" \
+        "1:\n\t" \
+        "add $16, %0\n\t" \
+        "add $16, %1\n\t" \
+        "cmp %3, %0\n\t" \
+        "ja 2f\n\t" \
+        "movdqu (%0), %%xmm0\n\t" \
+        "movdqu (%1), %%xmm1\n\t" \
+        "pcmpeqb %%xmm0, %%xmm1\n\t" \
+        "pmovmskb %%xmm1, %2\n\t" \
+        "xor $0x0000ffff, %2\n\t" \
+        "jz 1b\n\t" \
+        "2:" \
+        : /* %0 */ "=r" (scan), \
+          /* %1 */ "=r" (match), \
+          /* %2 */ "=&r" (t) \
+        : /* %3 */ "r" (strend-16), \
+          "0" (scan), \
+          "1" (match) \
+    ); \
+    if (t) { \
+        asm ("bsf %1, %0" : "=r" (t) : "r" (t)); \
+        scan += t; \
+        match += t; \
+    } else if (strend - scan) { \
+        if (strend - scan >= 8) \
+            goto CHECK_REST; \
+        else do { \
+            if (*++scan == *++match) \
+                break; \
+        } while (scan < strend); \
+    }
+#endif
+
+#ifndef LM_LEN_REMAINDER
+#  define LM_LEN_REMAINDER 0
+#  define LM_INNER_CHECK goto CHECK_REST;
+#endif
+
 local uInt longest_match(s, cur_match)
     deflate_state *s;
     IPos cur_match;                             /* current match */
@@ -1168,15 +1216,20 @@
         scan += 2, match++;
         Assert(*scan == *match, "match[2]?");
 
-        /* We check for insufficient lookahead only every 8th comparison;
-         * the 256th check will be made at strstart+258.
-         */
-        do {
-        } while (*++scan == *++match && *++scan == *++match &&
-                 *++scan == *++match && *++scan == *++match &&
-                 *++scan == *++match && *++scan == *++match &&
-                 *++scan == *++match && *++scan == *++match &&
-                 scan < strend);
+        if (LM_LEN_REMAINDER) {
+            LM_INNER_CHECK
+        } else {
+CHECK_REST:
+            /* We check for insufficient lookahead only every 8th comparison;
+             * the 256th check will be made at strstart+258.
+             */
+            do {
+            } while (*++scan == *++match && *++scan == *++match &&
+                     *++scan == *++match && *++scan == *++match &&
+                     *++scan == *++match && *++scan == *++match &&
+                     *++scan == *++match && *++scan == *++match &&
+                     scan < strend);
+        }
 
         Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");