Re: [PATCH] fs/fat: calculate checksum in a loop instead of directly calculating

[Heinrich Schuchardt]

On 18.02.2015 21:32, OGAWA Hirofumi wrote:
> Alexander Kuleshov <kuleshovmail@xxxxxxxxx> writes:
> 
>> time ./test_with_loop
>>
>> real    0m0.001s
>> user    0m0.000s
>> sys    0m0.001s
>>
>> And
>>
>> time ./test_direct_calculation:
>>
>> real    0m0.002s
>> user    0m0.000s
>> sys    0m0.001s
> 
> Hm,
> 
> #include <stdio.h>
> #include <stdlib.h>
> #include <unistd.h>
> 
> typedef unsigned char	__u8;
> typedef unsigned char	u8;
> 
> #if 1
> static inline unsigned char fat_checksum(const __u8 *name)
> {
> 	unsigned char s = name[0];
> 	s = (s<<7) + (s>>1) + name[1];	s = (s<<7) + (s>>1) + name[2];
> 	s = (s<<7) + (s>>1) + name[3];	s = (s<<7) + (s>>1) + name[4];
> 	s = (s<<7) + (s>>1) + name[5];	s = (s<<7) + (s>>1) + name[6];
> 	s = (s<<7) + (s>>1) + name[7];	s = (s<<7) + (s>>1) + name[8];
> 	s = (s<<7) + (s>>1) + name[9];	s = (s<<7) + (s>>1) + name[10];
> 	return s;
> }
> #else
> static inline unsigned char fat_checksum(const __u8 *name)
> {
> 	unsigned char s = name[0];
> 	u8 i;
> 	for (i = 1; i < 11; i++)
> 		s = (s << 7) + (s >> 1) + name[i];
> 	return s;
> }
> #endif
> 
> static __attribute__ ((noinline)) int test(unsigned char *name)

You have to put
__attribute__((optimize("unroll-loops")))
here to unroll the loop inside the function:


static __attribute__ ((noinline))
__attribute__((optimize("unroll-loops")))
int test(unsigned char *name)

> {
> 	long i;
> 	for (i = 0; i < 100000000L; i++)
> 		name[i % 11] = fat_checksum(name);
> 	return name[0];
> }
> 
> int main(int argc, char *argv[])
> {
> 	printf("%u\n", test((unsigned char *)argv[1]));
> 	return 0;
> }
> 
> 
>   $ gcc --version
>   gcc (Debian 4.9.1-19) 4.9.1
>   Copyright (C) 2014 Free Software Foundation, Inc.
>   This is free software; see the source for copying conditions.  There is NO
>   warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
> 
>   $ gcc -O2 -o c.inline c.c
>   # change #if 1 => #if 0
>   $ gcc -O2 -o c.loop c.c
> 
>   $ time ./c.inline aaaaaaaaaaa
>   14
> 
>   real	0m0.550s
>   user	0m0.548s
>   sys	0m0.000s
>   $ time ./c.loop aaaaaaaaaaa
>   14
> 
>   real	0m0.901s
>   user	0m0.896s
>   sys	0m0.004s
> 
> This is my environment only? (gcc (Debian 4.9.1-19) 4.9.1)
> 

$ time ./c.inline aaaaaaaaaaa
14

real    0m0.743s
user    0m0.740s
sys     0m0.000s

Without __attribute__((optimize("unroll-loops"))) :
$ time ./c.loop aaaaaaaaaaa
14

real    0m1.482s
user    0m1.472s
sys     0m0.004s

With __attribute__((optimize("unroll-loops"))) :

$ time ./c.loop aaaaaaaaaaa
14

real    0m0.742s
user    0m0.740s
sys     0m0.000s

Best regards

Heinrich Schuchardt
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/