On Thu, 19 Jan 2017 19:17:26 +0100
Thomas Mitterfellner <thomas.mitterfellner@xxxxxxxxx> wrote:
Not tried it on this occasion. Previously it slowed yoshi down to a crawl
and
even with everything set at 1024 frames I got masses of Xruns :(
To make matters worse I had a lot of difficulty trying to work out exactly
what
all the reams of data were telling me. I guess I'm doing it wrong {sigh}
For starters, you could look for invalid reads and writes, and
conditional jumps depending on uninitialized values and figure out
whether you could do anything about them. I usually use it with the
--track-origins=yes to get information about the origins of the
uninitialized values.
Of course running something through valgrind thrashes performance. I'd
suggest you do only short runs where you do stuff which triggered the
segfault previously. You don't necessarily need to trigger the segfault
in order to find possible problems in a valgrind run.
Greetings,
Thomas