wjones127 opened a new issue, #34058:
URL: https://github.com/apache/arrow/issues/34058
### Describe the enhancement requested
#15231 enabled memory tracking in benchmarks, but only for components that
always used the memory pool passed. Unfortunately, many components will make at
least some allocations with the default memory pool, so before we can add
memory metrics to their benchmarks we need to fix the implementations first.
#15231 also added a new metric on `MemoryPool`: `num_allocations`. This
provides a way to assert that an implementation only uses the passed memory
pool. First, wrap the default memory pool in a `ProxyMemoryPool` and record the
current `num_allocations` on the global pool. Pass the proxy pool to the
function. Then once the function returns, assert that the `num_allocations` on
the proxy pool equals the number of allocations on the global pool minus the
starting value.
Components:
* [ ] Parquet read and write
* [ ] CSV read and write
* [ ] Acero queries
* [ ] Dataset scans
* [ ] Compute functions
### Component(s)
Benchmarking, C++
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
