Skip to content

batching redis pushes on worker into pipeline #35

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

batching redis pushes on worker into pipeline #35

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 30 additions & 2 deletions rocket_learn/rollout_generator/redis/redis_rollout_worker.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,10 @@ def __init__(self, redis: Redis, name: str, match: Match,
send_obs=True, scoreboard=None, pretrained_agents=None,
human_agent=None, force_paging=False, auto_minimize=True,
local_cache_name=None,
gamemode_weights=None,):
gamemode_weights=None,
batch_mode=False,
step_size=100_000,
):
# TODO model or config+params so workers can recreate just from redis connection?
self.redis = redis
self.name = name
Expand Down Expand Up @@ -87,6 +90,11 @@ def __init__(self, redis: Redis, name: str, match: Match,
self.uuid = str(uuid4())
self.redis.rpush(WORKER_IDS, self.uuid)

self.batch_mode = batch_mode
self.step_size_limit = min(step_size / 20, 25_000)
if self.batch_mode:
self.red_pipe = self.redis.pipeline()
self.step_last_send = 0

# currently doesn't rebuild, if the old is there, reuse it.
if self.local_cache_name:
Expand Down Expand Up @@ -318,7 +326,7 @@ def run(self): # Mimics Thread
if not self.streamer_mode:
print(post_stats)

if not self.streamer_mode:
if not self.streamer_mode and not self.batch_mode:
rollout_data = encode_buffers(rollouts,
return_obs=self.send_obs,
return_states=self.send_gamestates,
Expand All @@ -345,6 +353,26 @@ def send():
# t.start()
# time.sleep(0.01)

elif not self.streamer_mode and self.batch_mode:

rollout_data = encode_buffers(rollouts,
return_obs=self.send_obs,
return_states=self.send_gamestates,
return_rewards=True)
rollout_bytes = _serialize((rollout_data, versions, self.uuid, self.name, result,
self.send_obs, self.send_gamestates, True))

self.red_pipe.rpush(ROLLOUTS, rollout_bytes)

# def send():
if (self.total_steps_generated - self.step_last_send) > self.step_size_limit or \
len(self.red_pipe) > 100:
n_items = self.red_pipe.execute()
if n_items[-1] >= 10000:
print("Had to limit rollouts. Learner may have have crashed, or is overloaded")
self.redis.ltrim(ROLLOUTS, -100, -1)
self.step_last_send = self.total_steps_generated

def _generate_matchup(self, n_agents, latest_version, pretrained_choice):
n_old = 0
if n_agents > 1:
Expand Down