Re: [PATCH v4 00/10] riscv: Allow userspace to directly access perf counters

[Alexandre Ghiti]

Hi Atish, Rémi,


On 19/07/2023 19:14, Atish Patra wrote:
> On Wed, Jul 19, 2023 at 7:46 AM Rémi Denis-Courmont <remi@xxxxxxxxxx> wrote:
> > Le keskiviikkona 19. heinäkuuta 2023, 1.48.49 EEST Atish Patra a écrit :
> > > > Isn't RDTIM susceptible to interference from power management and CPU
> > > > frequency scaling? I suppose that RDCYCLE may behave differently depending
> > > > on PM in *some* designs, but that would still be way better than RDTIME
> > > > for the purpose.
> > > Yes. But that's what it is probably using for other ISAs ?
> > At least on AArch64, it is using either Linux perf cycle counter, or if that
> > is disabled at build time, the raw PMU cycle counter - which obviously leads
> > to SIGILL on Linux, just like this MR would do with RDCYCLE.
> Good to know. Thanks for the clarification.
>
> > Again, I do not *personally* have objections to disabling RDCYCLE for
> > userspace (somebody else does, but that's neither my nor your problem). I do
> > have objections to the wording of some of the commit messages though.
> Completely agreed. We will update the commit text with more clarification in v5.


Thanks for all your comments and sorry for being slow here. I will 
improve the commit logs in the next version, that's an oversight, sorry 
about that.

Thanks again,

Alex


> > > My point was it should just do whatever it does for other ISA. RISC-V is no
> > > special in that regard.
> > Sure. My point is that RDTIME may be great for, so to say, system-level
> > benchmarks. For FFmpeg that could something like how long it takes to
> > transcode a video. But it doesn't seem to make much sense for
> > microbenchmarking of single threaded tightly optimised loops, as opposed to
> > RDCYCLE (or a wrapper for RDCYCLE).
> >
> > --
> > Rémi Denis-Courmont
> > http://www.remlab.net/