Skip to content

SparkBench 3x: change default number of shuffle partitions to 3x NPROC (#692) #692

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
ahmadelyoussef wants to merge 1 commit into
base: v2-beta
Choose a base branch
from
Open

SparkBench 3x: change default number of shuffle partitions to 3x NPROC (#692) #692

Show file tree
Hide file tree
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
2 changes: 1 addition & 1 deletion benchpress/config/jobs.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1042,7 +1042,7 @@

- benchmark: spark_standalone
name: spark_standalone_remote_3x
description: Spark standalone using remote SSDs with 3x query load (2 more joins) and auto-scaled shuffle partitions
description: Spark standalone using remote SSDs with 3x query load (2 more joins) and auto-scaled shuffle partitions (3 * CPU cores)
args:
- 'run'
- '--dataset-path /flash23/'
Expand Down
8 changes: 4 additions & 4 deletions packages/spark_standalone/templates/proj_root/scripts/run_perf_common.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,11 +93,11 @@ def write_sql_create_tables(args) -> List[str]:
# Set spark.sql.shuffle.partitions if specified
if hasattr(args, "shuffle_partitions") and args.shuffle_partitions is not None:
shuffle_partitions = args.shuffle_partitions
# Auto-scale to 4 * CPU cores if set to 0 or negative
# Auto-scale to 3 * CPU cores if set to 0 or negative
if shuffle_partitions <= 0:
shuffle_partitions = 4 * os.cpu_count()
shuffle_partitions = 3 * os.cpu_count()
print(
f"Auto-scaling shuffle partitions to {shuffle_partitions} (4 * {os.cpu_count()} CPU cores)"
f"Auto-scaling shuffle partitions to {shuffle_partitions} (3 * {os.cpu_count()} CPU cores)"
)
fp.write(f"""SET spark.sql.shuffle.partitions = {shuffle_partitions};\n""")
fp.write(f"""USE {args.database};""")
Expand Down Expand Up @@ -631,7 +631,7 @@ def init_parser():
type=int,
default=None,
help="Set the number of partitions for Spark SQL shuffle operations. "
"If set to 0 or negative, automatically scales to 4 * number of available CPU cores.",
"If set to 0 or negative, automatically scales to 3 * number of available CPU cores.",
)
# custom query option (for run_parser and exp_parser since they run queries)
for x in [run_parser, exp_parser]:
Expand Down
2 changes: 1 addition & 1 deletion packages/spark_standalone/templates/runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -427,7 +427,7 @@ def init_parser():
type=int,
default=None,
help="Set the number of partitions for Spark SQL shuffle operations. "
"If set to 0 or negative, automatically scales to 4 * number of available CPU cores.",
"If set to 0 or negative, automatically scales to 3 * number of available CPU cores.",
)
run_parser.add_argument(
"--query",
Expand Down
Loading