Hi Remy,
It turns out that you are at least partially on the right track :-) I
changed the way Barrett is written, so as to use as few temporary variables
as possible. In particular, the Rsh doesn't use any at all anymore ... the
60% of the Rsh time that was being used in runtime.mallocgc disappeared,
resulting in a bit more than 10% overall performance improvement in the
entire code. Now I understand better the comments about reuse of arguments
in the math/big documentation! Thanks for your suggestion.
On Tuesday, July 18, 2017 at 9:09:54 AM UTC+2, Jeff Templon wrote:
>
> Hi Remy,
>
> On Monday, July 17, 2017 at 10:39:56 PM UTC+2, Rémy Oudompheng wrote:
>>
>> 2017-07-17 15:56 GMT+02:00 Jeff Templon <jeff.temp...@gmail.com>:
>>
>
>
>> > it turns out that 77% of the time of the program is spent in
>> > runtime.mallocgc :-) I think the reason why is stuff like this:
>> >
>> > func (z nat) shr(x nat, s uint) nat {
>> > m := len(x)
>> > n := m - int(s/_W)
>> > if n <= 0 {
>> > return z[:0]
>> > }
>> > // n > 0
>> >
>> > z = z.make(n)
>> > shrVU(z, x[m-n:], s%_W)
>> >
>> > return z.norm()
>> >
>> > }
>> >
>>
>> How are you writing your code ? Wisely using variables should only
>> cause very minimal memory allocation.
>>
>>
> It's not my code that is taking all the time ... it's math/big itself.
> The way I understand it, for example in the above code, x[m-n:] creates a
> three-byte slice object. In the tightest part of the loop of my code, this
> function is called twice, for a million loop iterations this is already 3
> MB that will need to be garbage collected, and this is not the only
> function being called.
>
> I include the profile for a short run, you can see most of the time is
> spent in routine "Barrett", of which almost all that time is spent in
> functions in math/big, tracing through the graph you can see that
> ultimately most of that time is spent inside runtime.mallocgc ... is there
> a way to call math/big functions that does not result in so much garbage
> collection?
>
>
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
