[Intel-gfx] [PATCH igt 2/3] benchmarks/gem_syslatency: Apply vmpressure, measure page allocation

Chris Wilson chris at chris-wilson.co.uk
Wed Sep 27 11:07:29 UTC 2017


Quoting Joonas Lahtinen (2017-09-27 11:02:08)
> On Wed, 2017-09-27 at 10:41 +0100, Chris Wilson wrote:
> > Quoting Joonas Lahtinen (2017-09-27 09:52:42)
> > > On Mon, 2017-09-25 at 21:26 +0100, Chris Wilson wrote:
> > > > +static void *sys_thp_alloc(void *arg)
> > > > +{
> > > > +     struct sys_wait *w = arg;
> > > > +     struct timespec now;
> > > > +
> > > > +     clock_gettime(CLOCK_MONOTONIC, &now);
> > > > +     while (!done) {
> > > > +             const size_t sz = 2 << 20;
> > > > +             const struct timespec start = now;
> > > > +             void *ptr;
> > > > +
> > > > +             ptr = mmap(NULL, sz,
> > > > +                        PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS,
> > > > +                        -1, 0);
> > > > +             assert(ptr != MAP_FAILED);
> > > > +             madvise(ptr, sz, MADV_HUGEPAGE);
> > > > +             for (int page = 0; page < 2 << 20 >> 12; page++)
> > > > +                     *((volatile uint32_t *)ptr + (page << 12 >> 2)) = 0;
> > > 
> > > But what's the point in this iteration, we iterate from 0 to 512 page
> > > index (sz/PAGE_SIZE would be so much easier) and then write to to not
> > > each page but interleave four page writes per page and 3/4 of pages
> > > never get written? If this is intentional, please drop a comment.
> > 
> > -               for (int page = 0; page < 2 << 20 >> 12; page++)
> > -                       *((volatile uint32_t *)ptr + (page << 12 >> 2)) = 0;
> 
> Yes, why not write /sizeof() like civilized people do :P
> 
> > +               for (size_t page = 0; page < sz; page += PAGE_SIZE)
> > +                       *(volatile uint32_t *)(ptr + page) = 0;
> 
> Thats much more clear.
> 
> Reviewed-by: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
> 
> It seems program usage info is not a hot feature for benchmarks.

Indeed. The intention is that you never run these directly but through
an ezbench wrapper. Paging Martin.
-Chris


More information about the Intel-gfx mailing list