Hello Hironobu-san,
Here is a v15 which is a rebase, plus a large simplification of the modmul
function if an int128 type is available, which is probably always…
Could you have a look and possibly revalidate?
> Sorry for the late reply. I reviewed this patch.
>
> Function nbits(), which was previously discussed, has been simplified by
> using the function pg_popcount64().
>
> By adding the mathematical explanation, it has been easier to understand the
> operation of this function.
>
> I believe that these improvements will have a positive impact on maintenance.
>
> The patch could be applied successfully and the tests passed without
> problems.
>
> So, I think the latest patch is fine.
>
>
> Best regards,
>
>
>
>> On 3/3/19 12:55 PM, Fabien COELHO wrote:
>>>
>>>> Indeed, the patch needs a rebase & conflit resolution. I'll do it. Later.
>>>
>>> Here is an update:
>>>
>>> - take advantage of pg_bitutils (although I noted that the "slow"
>>> popcount there could be speeded-up and shorten with a bitwise operator
>>> implementation that I just removed from pgbench).
>>>
>>> - add comments about the bijective transformations in the code.
>>>
>>> As already stated, this function makes sense for people who want to test
>>> performance with pgbench using non uniform rands. If you do not want to do
>>> that, you will probably find the function pretty useless. I can't help it.
>>>
>>> Also, non uniform rands is also a way to test pg lock contention behavior.
>>
>> You have signed up as a reviewer for this patch. Do you know when you'll
>> have time to do the review?
>>
>> Regards,
>
>
--
Fabien Coelho - CRI, MINES ParisTech