Re: [PATCH v2 1/3] lib: find_*_bit reimplementation

[Yury]

On 02.02.2015 06:17, George Spelvin wrote:
> Yury Norov <y.norov@xxxxxxxxxxx> wrote:
>> New implementations takes less space in source file (see diffstat)
>> and in object. For me it's 710 vs 453 bytes of text.
>>
>> Patch was boot-tested on x86_64 and MIPS (big-endian) machines.
>> Performance tests were ran on userspace with code like this:
>>
>> 	/* addr[] is filled from /dev/urandom */
>> 	start = clock();
>> 	while (ret < nbits)
>> 		ret = find_next_bit(addr, nbits, ret + 1);
>>
>> 	end = clock();
>> 	printf("%ld\t", (unsigned long) end - start);
>> On Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz rezults are next:
>> (for find_next_bit, nbits is 8M, for find_first_bit - 80K)
>>
>> 	find_next_bit:		find_first_bit:
>> 	new	current		new	current
>> 	26932	43151		14777	14925
>> 	26947	43182		14521	15423
> I'll look at this more carefully, but one immediate thought is that this
> is an unrealistic benchmark.  It will amost never need to look at more
> than one word of the array, but real arrays have long runs of zero
> bits to skip over.
>
> So the code size is appreciated, but the time benefits may be the result
> of you optimizing for the wrong thing.
>
> I'd try filling the array with mostly-identical bits, flipping with odds
> of 1/256 or so.
>
> For full generality, I'd test different 1->0 and 0->1 transition
> probabilities.  (But powers of two are probably enough for benchmarking.)
>
I think, test with random values represents at least one situation: well-fragmented memory
after long time work. (This is what I really have in my project.) In other hand, if long zero runs
is a typical behavior for one's system, it's a good opportunity for improvements, I think.
Anyway, the idea of testing find_bit on a long runs is good. Thank you.
--
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/