Re: [PATCH v3 00/16] Introduce and use generic parity16/32/64 helper

[Kuan-Wei Chiu <visitorckw@gmail.com>]

Hi Jeremy,

On Fri, Apr 04, 2025 at 10:51:55AM +0800, Jeremy Kerr wrote:
> Hi Yuri & Kuan-Wei:
> 
> > Thank you for sharing your opinion on this fixed parity(). Your
> > arguments may or may not be important, depending on what existing
> > users actually need. Unfortunately, Kuan-Wei didn't collect
> > performance numbers and opinions from those proposed users.
> 
> For the fsi-i2c side: this isn't a performance-critical path, and any
> reasonable common approach would likely perform better that the current
> per-bit implementation.
> 
> Our common targets for this driver would be arm and powerpc64le. In case
> it's useful as a reference, using the kernel compilers I have to hand, a
> __builtin_parity() is a library call on the former, and a two-instruction
> sequence for the latter.
> 
Thanks for your feedback.

IIUC, from the fsi-i2c perspective, parity efficiency isn't a major
concern, but you still prefer optimizing with methods like
__builtin_parity(). I'm just unsure if this aligns with Yury's point
about needing "evidence that performance and/or code generation is
important here."

Regards,
Kuna-Wei