This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] aarch64: Optimized memset for Kunpeng processor.


Hi Derek,

> Here the setting inerval if 64..127 bytes rather than 64..96 bytes, so if no branch,
> the 64..80 bytes setting will beyond the border using dstend. And the interval
> became longer just can benefit 96..127 bytes.

For the 64..128 case it is always safe to copy 64 bytes from the start and 64 bytes
from the end - the tail overlap means you never can go outside the bounds.
Generally it's faster that way due to avoiding unnecessary branches which may mispredict.

Wilco

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]