Skip to content

Scale the number of physical cores requested with the expected memory usage #252

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

Merged
dnerini merged 1 commit into from
Aug 18, 2026
+22 −6
Merged

Scale the number of physical cores requested with the expected memory usage #252

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
16 changes: 16 additions & 0 deletions workflow/rules/common.smk
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -469,6 +469,22 @@ PARAMS_WITHOUT_STEP_ZERO_VALUE = {
}


def balfrin_cpus(mem_mb, actual_cpus=1, node_mem_mb=512_000, node_cpus=256):
"""Minimum cores to request on Balfrin based on memory usage.

Balfrin CPU nodes have no per-job memory control: any job can consume the
full node memory (512 GB / 256 cores). To avoid starving other jobs,
reserve cores proportional to expected memory so the scheduler implicitly
limits total memory usage:

cores = max(actual_cpus, ceil(mem_mb / node_mem_mb * node_cpus))
"""
import math

mem_cpus = math.ceil(mem_mb / node_mem_mb * node_cpus)
return max(actual_cpus, mem_cpus)


def resolve_leadtimes(steps_spec, requested="all", param=None):
"""Lead times to compute for a single participant.

Expand Down
12 changes: 6 additions & 6 deletions workflow/rules/verification.smk
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ rule verification_metrics_baseline:
log:
OUT_ROOT / "logs/verification_metrics_baseline/{baseline_id}-{init_time}.log",
resources:
cpus_per_task=24,
cpus_per_task=balfrin_cpus(80_000, actual_cpus=24),
mem_mb=80_000,
runtime="120m",
params:
Expand Down Expand Up @@ -75,7 +75,7 @@ rule verification_metrics:
log:
OUT_ROOT / "logs/verification_metrics/{run_id}-{init_time}.log",
resources:
cpus_per_task=24,
cpus_per_task=balfrin_cpus(80_000, actual_cpus=24),
mem_mb=80_000,
runtime="60m",
# wildcard_constraints:
Expand Down Expand Up @@ -134,7 +134,7 @@ rule verification_metrics_aggregation:
log:
OUT_ROOT / "logs/verification_metrics_aggregation/{run_id}.log",
resources:
cpus_per_task=24,
cpus_per_task=balfrin_cpus(250_000, actual_cpus=24),
mem_mb=250_000,
runtime="2h",
shell:
Expand Down Expand Up @@ -170,7 +170,7 @@ rule verification_metrics_plot:
log:
OUT_ROOT / "logs/verification_metrics_plot/{experiment}.log",
resources:
cpus_per_task=16,
cpus_per_task=balfrin_cpus(50_000, actual_cpus=16),
mem_mb=50_000,
runtime="20m",
params:
Expand Down Expand Up @@ -210,7 +210,7 @@ rule verification_scoremaps:
OUT_ROOT
/ f"logs/verification_scoremaps/{{run_id}}-{TRUTH_HASH}-{{param}}-{{leadtime}}.log",
resources:
cpus_per_task=2,
cpus_per_task=balfrin_cpus(50_000, actual_cpus=2),
mem_mb=50_000,
runtime="60m",
# wildcard_constraints:
Expand Down Expand Up @@ -250,7 +250,7 @@ rule verification_scoremaps_baseline:
OUT_ROOT
/ f"logs/verification_scoremaps_baseline/{{baseline_id}}-{TRUTH_HASH}-{{param}}-{{leadtime}}.log",
resources:
cpus_per_task=24,
cpus_per_task=balfrin_cpus(50_000, actual_cpus=24),
mem_mb=50_000,
runtime="60m",
params:
Expand Down
Loading