# APPO/IMPALA Logging

**URL:** <https://discuss.ray.io/t/appo-impala-logging/1270>\
**Category:** RLlib\
**Created:** [March 15, 2021, 11:21pm UTC](https://discuss.ray.io/t/appo-impala-logging/1270 "2021-03-15T23:21:03Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![jsuarez5341](https://avatars.discourse-cdn.com/v4/letter/j/50afbb/32.png) [@jsuarez5341](https://discuss.ray.io/u/jsuarez5341)\
**Post date:** [March 15, 2021, 11:21pm UTC](https://discuss.ray.io/t/appo-impala-logging/1270/1 "2021-03-15T23:21:03Z")

</div>

It appears that APPO and PPO return quite different stats upon every trainer.train() step. Logging times/throughput is quite easy with PPO, but some associated fields are missing in APPO, specifically learn\_throughput and learn\_time\_ms.

Additionally, the values returned by sample\_time\_ms are far below wall-clock time. I’ve also had to dramatically reduce the batch size, as APPO seems to be consuming far more GPU memory than PPO

Questions:

1. Where are the throughput values for APPO?
2. Why are sample times low?
3. Why does APPO consume so much memory?