[llvm] [RFC][LV] VPlan-based cost model (PR #67647)

[Kolya Panchenko via llvm-commits]

nikolaypanchenko wrote:

> > > General idea seems sound, but this needs tests to convince me that LV behavior doesn't change with VPlan-based costing.
> > 
> > 
> > Indeed. Similar comment on the original Phab thread. Needs existing tests with flag to not change behaviour based on new cost model.
> 
> The cost model has moved inside the VPlan infrastructure and is no longer called from the same place in the same way. Please provide evidence that, even without the vplan-based cost model flag, the behaviour of the vectorizer hasn't changed.
> 
> Performance numbers on some benchmarks, traces on the test-suite showing what gets vectorized and what doesn't, etc.
> 
> If this is just about the migration, then it should behave in the same way as before without the optional flag.

Is there a list of benchmarks we should care about for the community ? Spec2k6, tsvc ?
To me to enable vplan-based cost model in one shot is not a trivial work, not only because of benchmarks, but also because of multiple targets. We can enable it by targets only and take a responsibility to do that for RISC-V first.

https://github.com/llvm/llvm-project/pull/67647