checkasm: Serialize read_time() calls on x86

Improves the accuracy of benchmarks, especially in short functions.

To quote the Intel 64 and IA-32 Architectures Software Developer's Manual:
"The RDTSC instruction is not a serializing instruction. It does not necessarily
wait until all previous instructions have been executed before reading the counter.
Similarly, subsequent instructions may begin execution before the read operation
is performed. If software requires RDTSC to be executed only after all previous
instructions have completed locally, it can either use RDTSCP (if the processor
supports that instruction) or execute the sequence LFENCE;RDTSC."

RDTSCP would accomplish the same task, but it's only available since Nehalem.

This change makes SSE2 a requirement to run checkasm.

diff --git a/tools/checkasm.c b/tools/checkasm.c
index f4a854773f7fdefeac040bc502a3da7d0c595c5b..a0e8810cda5635e57bc4be35c09cac3d864eef08 100644 (file)
--- a/tools/checkasm.c
+++ b/tools/checkasm.c
@@ -90,7 +90,9 @@ static inline uint32_t read_time(void)
 {
     uint32_t a = 0;
 #if HAVE_X86_INLINE_ASM
-    asm volatile( "rdtsc" : "=a"(a) :: "edx", "memory" );
+    asm volatile( "lfence \n"
+                  "rdtsc  \n"
+                  : "=a"(a) :: "edx", "memory" );
 #elif ARCH_PPC
     asm volatile( "mftb %0" : "=r"(a) :: "memory" );
 #elif ARCH_ARM     // ARMv7 only