8 Feb
2022
8 Feb
'22
8:50 p.m.
On Sun, Jan 30, 2022 at 01:33:35AM +0200, Nir Soffer wrote:
Using Slice() we can use builtin copy() instead of a manual loop, which is 4.6 times faster with a typical 256k buffer:
Again, the benefits of the underlying libraries being able to vectorize over larger chunks of memory ;)
Before: BenchmarkFromBytes-12 9806 111474 ns/op
After: BenchmarkFromBytes-12 48193 24106 ns/op
Signed-off-by: Nir Soffer <nsoffer@redhat.com> --- golang/aio_buffer.go | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-)
ACK. -- Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org