On 11/10/20 8:03 AM, 172060045 wrote:
Hi,
Recently, I was interested in GCC AutoFDO optimization, which works by sampling
specific PMU event on production machines and using those profiles to guide
optimization. In this way, information such as cache miss can also be obtained
through sampling, so can we implement feedback-directed cache optimization
according to this idea?
Hello.
AutoFDO support in GCC is currently is quite bad shape. The following tool [1]
is supposed to generate a reasonable .gcda file
that GCC can consume.
I know Bin spent some time working on that. Hopefully, he can tell what was his
conclusion autoFDO?
ARMv8.2 provides SPE features, which can obtain accurate LLC miss, TLB miss,
branch miss and remote access information through perf, it may be helpful to
the idea.
Is any one doing relevant work?It would be grateful if someone could offer any
advices, thx!
I can guide you with generation of the .gcda profile files, but you will need
to understand and transform a perf profile
to a GCC gcda profile.
Martin
[1] https://github.com/google/autofdo