Re: [Perfctr-devel] [patch] Performance Counters for Linux, v4

[Andi Kleen]

> To make life easier for developers not interested in exact measuring, but in
> cross-platform development, PAPI has introduced "generic" events. These are
> documented not to be interpretable in a cross-platform manner - and
> typically consist of aggregation of a number of low-level architecture
> specific events. Instead a measure as PERF_COUNT_CACHE_REFERENCES tries to
> create an event set that is a "best effort" for the platform. Typically it
> would try to count read and write events to L1 and L2, but depending on the

The point was that L1 and L2/L3 behave so differently in practice
due to their latency/bandwidth differences (typically L1 is more like a 
slower register file while L3 is more like faster DRAM) that it's unclear
that any generic event not defining what it means is any useful.

I could see the point of generic FLC/LLC cache events perhaps,
but not of a single one lumping these two together.

But although with multi core it looks like 3 level hierarchies are
becoming more and more common and also more important to know
about.

-Andi

-- 
ak@xxxxxxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/