On my MacBook, with Raku 2020.01 built on MoarVM version 2020.01.1
changing:
$ed.data[^$ed.data_size]
to:
$ed.data.head($ed.data_size)
cut the time in half.
I cannot speak to what might be happening with the Promises.
Test code:
class Foo {
has @.data = 0 xx (1024 ** 2);
has $.data_size = @!data.elems;
}
my Foo $ed .= new;
sub time_it ( Int $type ) {
for ^3 {
my Instant $init3 = DateTime.now().Instant;
for ^10 {
if $type == 1 {
my Blob $bindata =
Blob[uint8].new($ed.data[^$ed.data_size]);
}
else {
my Blob $bindata =
Blob[uint8].new($ed.data.head($ed.data_size));
}
}
say "Bindata in {DateTime.now.Instant - $init3}";
}
}
time_it(1);
say '';
time_it(2);
Output:
Bindata in 8.207297
Bindata in 8.12964
Bindata in 8.0798136
Bindata in 3.434386
Bindata in 3.4062148
Bindata in 3.35743893
—
Hope this helps,
Bruce Gray (Util of PerlMonks)
> On Jun 16, 2020, at 5:17 PM, David Santiago <[email protected]> wrote:
>
> Thanks for the answer.
>
> There's a slight performance improvement, but It still takes more than 1
> second:
>
> Code:
>
> my Instant $init3 = DateTime.now().Instant;
> #my Blob $bindata = Blob[uint8].new(@data);
> my Blob $bindata = Blob[uint8].new($ed.data[^$ed.data_size]);
> say "Bindata in {DateTime.now.Instant - $init3}";
>
>
> Bindata in 1.1250962
>
>
> :-(
>
> Curt Tilmes <[email protected]> escreveu no dia terça, 16/06/2020 à(s) 21:40:
>>
>> On Tue, Jun 16, 2020 at 5:18 PM David Santiago <[email protected]> wrote:
>>> my uint8 @data = $ed.data[0..$ed.data_size-1].Array;
>>> my Blob $bindata = Blob[uint8].new(@data);
>>
>> Not absolutely sure, but it seems like you are copying the data twice.
>> Try just
>>
>> my $bindata = Blob.new($ed.data[^$ed.data_size]);
