我正在运行此代码(此处为完整代码:http://codepad.org/5OJBLqIA),用于预先重复daxpy函数调用,无论是否事先从缓存中刷新操作数:
#define KB 1024
int main()
{
int cache_size = 32*KB;
double alpha = 42.5;
int operand_size = cache_size/(sizeof(double)*2);
double* X = new double[operand_size];
double* Y = new double[operand_size];
//95% confidence interval
double max_risk = 0.05;
//Interval half width
double w;
int n_iterations = 50000;
students_t dist(n_iterations-1);
double T = boost::math::quantile(complement(dist,max_risk/2));
accumulator_set<double, stats<tag::mean,tag::variance> > unflushed_acc;
for(int i = 0; i < n_iterations; ++i)
{
fill(X,operand_size);
fill(Y,operand_size);
double seconds = wall_time();
daxpy(alpha,X,Y,operand_size);
seconds = wall_time() - seconds;
unflushed_acc(seconds);
}
w = T*sqrt(variance(unflushed_acc))/sqrt(count(unflushed_acc));
printf("Without flush: time=%g +/- %g ns\n",mean(unflushed_acc)*1e9,w*1e9);
//Using clflush instruction
//We need to put the operands back in cache
accumulator_set<double, stats<tag::mean,tag::variance> > clflush_acc;
for(int i = 0; i < n_iterations; ++i)
{
fill(X,operand_size);
fill(Y,operand_size);
flush_array(X,operand_size);
flush_array(Y,operand_size);
double seconds = wall_time();
daxpy(alpha,X,Y,operand_size);
seconds = wall_time() - seconds;
clflush_acc(seconds);
}
w = T*sqrt(variance(clflush_acc))/sqrt(count(clflush_acc));
printf("With clflush: time=%g +/- %g ns\n",mean(clflush_acc)*1e9,w*1e9);
return 0;
}
此代码测量给定迭代次数的平均速率和不确定度。对大量迭代进行平均可以最大限度地减少因各种内核对内存访问的争用所引起的差异(在我之前的问题here中讨论过),但是这样获得的平均值在巨大之间变化很大单独调用相同的可执行文件:
$ ./variance
Without flush: time=3107.76 +/- 0.268198 ns
With clflush: time=5862.33 +/- 9.84313 ns
$ ./variance
Without flush: time=3105.71 +/- 0.237823 ns
With clflush: time=7802.66 +/- 12.3163 ns
这些是彼此紧接着的。为什么刷新案例的时间(但不是未刷新的案例)在流程之间变化如此之大,但在给定流程中却如此之少?
附录
代码在Intel Xeon 5650上的Mac OS X 10.8上运行。