Re: [PATCH v3 1/n] perf/core: addressing 4x slowdown during per-process profiling of STREAM benchmark on Intel Xeon Phi

From: Mark Rutland
Date: Fri Jun 16 2017 - 05:10:31 EST

On Fri, Jun 16, 2017 at 01:10:10AM +0300, Alexey Budankov wrote:
> On 15.06.2017 22:56, Mark Rutland wrote:
> >On Thu, Jun 15, 2017 at 08:41:42PM +0300, Alexey Budankov wrote:
> >>This series of patches continues v2 and addresses captured comments.

> >>Specifically this patch replaces pinned_groups and flexible_groups
> >>lists of perf_event_context by red-black cpu indexed trees avoiding
> >>data structures duplication and introducing possibility to iterate
> >>event groups for a specific CPU only.
> >
> >If you use --per-thread, I take it the overhead is significantly
> >lowered?
> Please ask more.

IIUC, you're seeing the slowdown when using perf record, correct?

There's a --per-thread option to ask perf record to not duplicate the
event per-cpu.

If you use that, what amount of slowdown do you see?

It might be preferable to not open task-bound per-cpu events on systems
with large cpu counts, and it would be good to know what the trade-off
looks like for this case.

> >>+static void
> >>+perf_cpu_tree_insert(struct rb_root *tree, struct perf_event *event)
> >>+{
> >>+ struct rb_node **node;
> >>+ struct rb_node *parent;
> >>+
> >>+ WARN_ON_ONCE(!tree || !event);
> >>+
> >>+ node = &tree->rb_node;
> >>+ parent = *node;
> >
> >The first iteration of the loop handles this, so it can go.
> If tree is empty parent will be uninitialized what is harmful.

Sorry; my bad.