From e87de95fcd452e07a6bed4ef595b64980eb61e6e Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Thu, 4 Sep 2025 16:09:50 -0400
Subject: [PATCH 01/21] Add back random output dir

---
 raps/engine.py     |  2 +-
 raps/run_sim.py    |  8 +++++---
 raps/sim_config.py | 15 ++++++++++++++-
 raps/workload.py   |  5 +++--
 4 files changed, 23 insertions(+), 7 deletions(-)

diff --git a/raps/engine.py b/raps/engine.py
index 23a2605..90b84b0 100644
--- a/raps/engine.py
+++ b/raps/engine.py
@@ -161,7 +161,7 @@ class Engine:
         self.flops_manager = flops_manager
         self.debug = sim_config.debug
         self.continuous_workload = continuous_workload
-        self.output = sim_config.output
+        self.output = sim_config.get_output()
         self.replay = sim_config.replay
         self.downscale = sim_config.downscale  # Factor to downscale the 1s timesteps (power of 10)
         self.simulate_network = sim_config.simulate_network
diff --git a/raps/run_sim.py b/raps/run_sim.py
index 5afd6f1..6c39a2f 100644
--- a/raps/run_sim.py
+++ b/raps/run_sim.py
@@ -44,6 +44,7 @@ shortcuts = {
     "numjobs": "n",
     "verbose": "v",
     "output": "o",
+    "random-output": "O",
     "uncertainties": "u",
     "plot": "p",
     "replay": "f",
@@ -79,7 +80,7 @@ def run_sim(sim_config: SimConfig):
 
     engine, workload_data, time_delta = Engine.from_sim_config(sim_config)
 
-    out = sim_config.output
+    out = sim_config.get_output()
     if out:
         out.mkdir(parents=True)
         engine.telemetry.save_snapshot(
@@ -253,10 +254,11 @@ def run_parts_sim(sim_config: SimConfig):
     multi_engine, workload_results, timestep_start, timestep_end, time_delta = \
         MultiPartEngine.from_sim_config(sim_config)
 
-    if sim_config.output:
+    out = sim_config.get_output()
+    if out:
         for part, engine in multi_engine.engines.items():
             engine.telemetry.save_snapshot(
-                dest=str(sim_config.output / part.split('/')[-1]),
+                dest=str(out / part.split('/')[-1]),
                 result=workload_results[part],
                 args=sim_config,
             )
diff --git a/raps/sim_config.py b/raps/sim_config.py
index 5cdd09c..f18c730 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -1,10 +1,11 @@
 import argparse
+from pathlib import Path
 from functools import cached_property
 from datetime import timedelta
 from typing import Literal
 from raps.schedulers.default import PolicyType, BackfillType
 from raps.utils import (
-    parse_time_unit, convert_to_time_unit, infer_time_unit, ExpandedPath, parse_td,
+    parse_time_unit, convert_to_time_unit, infer_time_unit, ExpandedPath, parse_td, create_casename,
 )
 from raps.system_config import SystemConfig, get_partition_configs
 from pydantic import BaseModel, model_validator
@@ -62,8 +63,17 @@ class SimConfig(BaseModel):
 
     seed: int | None = None
     """ Set RNG seed for deterministic simulation """
+
     output: ExpandedPath | None = None
     """ Output power, cooling, and loss models for later analysis. Argument specifies name. """
+    random_output: bool = False
+    """ If True, output to a randomly named directory. Mutually exclusive with output """
+
+    _random_output: Path | None = None
+    def get_output(self):
+        if self.random_output and not self._random_output:
+            self._random_output = Path(create_casename("out-")).resolve()
+        return self._random_output or self.output
 
     debug: bool = False
     """ Enable debug mode and disable rich layout """
@@ -291,6 +301,9 @@ class SimConfig(BaseModel):
                 raise ValueError(f"policy {self.backfill} not implemented by {self.scheduler}. "
                                  f"Valid selections: {sorted(valid_backfilltypes)}")
 
+        if self.random_output and self.output:
+            raise ValueError("--random-output and --output are mutually exclusive")
+
         return self
 
     @property
diff --git a/raps/workload.py b/raps/workload.py
index 6fb3c3b..57b536d 100644
--- a/raps/workload.py
+++ b/raps/workload.py
@@ -987,10 +987,11 @@ def run_workload(sim_config: SimConfig):
                   dist_split=sim_config.multimodal,
                   gantt_nodes=sim_config.gantt_nodes)
 
-    if sim_config.output:
+    out = sim_config.get_output()
+    if out:
         timestep_start = min([x.submit_time for x in jobs])
         timestep_end = math.ceil(max([x.submit_time for x in jobs]) + max([x.expected_run_time for x in jobs]))
-        filename = create_file_indexed('wl', create=False, ending="npz").split(".npz")[0]
+        filename = create_file_indexed('wl', path = str(out), create=False, ending="npz").split(".npz")[0]
         # savez_compressed add npz itself, but create_file_indexed needs to check for .npz to find existing files
         np.savez_compressed(filename, jobs=jobs, timestep_start=timestep_start, timestep_end=timestep_end, args=args)
         print(filename + ".npz")  # To std-out to show which npz was created.
-- 
GitLab


From 5b83b1867a326f65f7ae3576f8a2f8f347bb12ba Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Thu, 4 Sep 2025 16:32:38 -0400
Subject: [PATCH 02/21] Set pydantic config so CLI help includes docstrings

---
 raps/sim_config.py    |  7 ++++---
 raps/system_config.py | 19 ++++++++++---------
 raps/telemetry.py     |  6 +++---
 raps/utils.py         |  9 ++++++++-
 4 files changed, 25 insertions(+), 16 deletions(-)

diff --git a/raps/sim_config.py b/raps/sim_config.py
index f18c730..300ac94 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -3,18 +3,19 @@ from pathlib import Path
 from functools import cached_property
 from datetime import timedelta
 from typing import Literal
+import importlib
 from raps.schedulers.default import PolicyType, BackfillType
 from raps.utils import (
     parse_time_unit, convert_to_time_unit, infer_time_unit, ExpandedPath, parse_td, create_casename,
+    RAPSBaseModel,
 )
 from raps.system_config import SystemConfig, get_partition_configs
-from pydantic import BaseModel, model_validator
-import importlib
+from pydantic import model_validator
 
 Distribution = Literal['uniform', 'weibull', 'normal']
 
 
-class SimConfig(BaseModel):
+class SimConfig(RAPSBaseModel):
     system: str | None = None
     """ System config to use """
     partitions: list[str] = []
diff --git a/raps/system_config.py b/raps/system_config.py
index 726c086..adbf470 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -5,13 +5,14 @@ from typing import Any, Literal
 from pathlib import Path
 from functools import cached_property
 import yaml
-from pydantic import BaseModel, computed_field, model_validator, field_validator
+from pydantic import computed_field, model_validator, field_validator
+from raps.utils import RAPSBaseModel
 from raps.raps_config import raps_config
 
 # Define Pydantic models for the config to handle parsing and validation
 
 
-class SystemSystemConfig(BaseModel):
+class SystemSystemConfig(RAPSBaseModel):
     num_cdus: int
     racks_per_cdu: int
     nodes_per_rack: int
@@ -79,7 +80,7 @@ class SystemSystemConfig(BaseModel):
         return self.total_nodes - len(self.down_nodes)
 
 
-class SystemPowerConfig(BaseModel):
+class SystemPowerConfig(RAPSBaseModel):
     power_gpu_idle: float
     power_gpu_max: float
     power_cpu_idle: float
@@ -100,7 +101,7 @@ class SystemPowerConfig(BaseModel):
     power_cost: float
 
 
-class SystemUqConfig(BaseModel):
+class SystemUqConfig(RAPSBaseModel):
     power_gpu_uncertainty: float
     power_cpu_uncertainty: float
     power_mem_uncertainty: float
@@ -115,7 +116,7 @@ class SystemUqConfig(BaseModel):
 JobEndStates = Literal["COMPLETED", "FAILED", "CANCELLED", "TIMEOUT", "NODE_FAIL"]
 
 
-class SystemSchedulerConfig(BaseModel):
+class SystemSchedulerConfig(RAPSBaseModel):
     job_arrival_time: int
     mtbf: int
     trace_quanta: int
@@ -127,7 +128,7 @@ class SystemSchedulerConfig(BaseModel):
     multitenant: bool = False
 
 
-class SystemCoolingConfig(BaseModel):
+class SystemCoolingConfig(RAPSBaseModel):
     cooling_efficiency: float
     wet_bulb_temp: float
     zip_code: str | None = None
@@ -140,7 +141,7 @@ class SystemCoolingConfig(BaseModel):
     temperature_keys: list[str]
 
 
-class SystemNetworkConfig(BaseModel):
+class SystemNetworkConfig(RAPSBaseModel):
     topology: Literal["capacity", "fat-tree", "dragonfly", "torus3d"]
     network_max_bw: float
     latency: float | None = None
@@ -163,7 +164,7 @@ class SystemNetworkConfig(BaseModel):
     node_coords_csv: str | None = None
 
 
-class SystemConfig(BaseModel):
+class SystemConfig(RAPSBaseModel):
     system_name: str
     """ Name of the system, defaults to the yaml file name """
 
@@ -202,7 +203,7 @@ class SystemConfig(BaseModel):
         return config_dict
 
 
-class MultiPartitionSystemConfig(BaseModel):
+class MultiPartitionSystemConfig(RAPSBaseModel):
     system_name: str
     partitions: list[SystemConfig]
 
diff --git a/raps/telemetry.py b/raps/telemetry.py
index 340b9ae..63ee158 100644
--- a/raps/telemetry.py
+++ b/raps/telemetry.py
@@ -15,7 +15,7 @@ from types import ModuleType
 import importlib
 import numpy as np
 import pandas as pd
-from pydantic import BaseModel, model_validator
+from pydantic import model_validator
 # from rich.progress import track
 
 from raps.sim_config import SimConfig
@@ -28,12 +28,12 @@ from raps.plotting import (
     plot_network_histogram
 )
 from raps.utils import (
-    next_arrival_byconfargs, pydantic_add_args, SubParsers, ExpandedPath, WorkloadData,
+    next_arrival_byconfargs, pydantic_add_args, SubParsers, ExpandedPath, WorkloadData, RAPSBaseModel,
 )
 
 
 # TODO: should reuse this model in SimConfig
-class TelemetryArgs(BaseModel):
+class TelemetryArgs(RAPSBaseModel):
     jid: str = '*'
     """ Replay job id """
     replay: list[ExpandedPath] | None = None
diff --git a/raps/utils.py b/raps/utils.py
index 323ac8a..6042a45 100644
--- a/raps/utils.py
+++ b/raps/utils.py
@@ -639,6 +639,13 @@ SmartTimedelta = A[timedelta, AfterValidator(parse_td)]
 T = TypeVar("T", bound=BaseModel)
 
 
+class RAPSBaseModel(BaseModel):
+    """ Base Pydantic model with shared config """
+    model_config = ConfigDict(
+        use_attribute_docstrings=True,
+    )
+
+
 def pydantic_add_args(
     parser: argparse.ArgumentParser, model_cls: type[T],
     model_config: SettingsConfigDict | None = None,
@@ -711,7 +718,7 @@ def yaml_dump(data):
     )
 
 
-class WorkloadData(BaseModel):
+class WorkloadData(RAPSBaseModel):
     """
     Represents a workload, a list of jobs with some metadata. Returned by dataloaders load_data()
     function, and by Workload.generate_jobs().
-- 
GitLab


From f82ed3a5253a0f4c81f84e7ab4bafb352a673db1 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 10:30:58 -0400
Subject: [PATCH 03/21] Better error reporting

---
 raps/utils.py | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/raps/utils.py b/raps/utils.py
index 6042a45..dbb5ea3 100644
--- a/raps/utils.py
+++ b/raps/utils.py
@@ -21,7 +21,7 @@ import json
 import argparse
 from pathlib import Path
 from typing import Annotated as A, TypeVar, Callable, TypeAlias
-from pydantic import BaseModel, TypeAdapter, AfterValidator, ConfigDict, AwareDatetime
+from pydantic import BaseModel, TypeAdapter, AfterValidator, ConfigDict, AwareDatetime, ValidationError
 from pydantic_settings import BaseSettings, SettingsConfigDict, CliApp, CliSettingsSource
 import yaml
 from raps.job import Job
@@ -662,6 +662,7 @@ def pydantic_add_args(
     model_config_dict = SettingsConfigDict({
         "cli_implicit_flags": True,
         "cli_kebab_case": True,
+        "title": model_cls.__name__,
         **(model_config or {}),
         "cli_parse_args": False,  # Don't automatically parse args
     })
@@ -678,13 +679,17 @@ def pydantic_add_args(
     cli_settings_source = CliSettingsSource(SettingsModel, root_parser=parser)
 
     def model_validate_args(args: argparse.Namespace, data: dict | None = None):
-        model = CliApp.run(SettingsModel,
-                           cli_args=args,
-                           cli_settings_source=cli_settings_source,
-                           **(data or {}),
-                           )
-        # Recreate model so we don't return the SettingsModel subclass
-        return model_cls.model_validate(model.model_dump())
+        try:
+            model = CliApp.run(SettingsModel,
+                            cli_args=args,
+                            cli_settings_source=cli_settings_source,
+                            **(data or {}),
+                            )
+            # Recreate model so we don't return the SettingsModel subclass
+            return model_cls.model_validate(model.model_dump())
+        except ValidationError as err:
+            print(err)
+            sys.exit(1)
     return model_validate_args
 
 
-- 
GitLab


From 623dae8f415a37a259ff33d06b5d536e097d5c44 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 11:10:10 -0400
Subject: [PATCH 04/21] Set system in legacy args

---
 raps/engine.py     | 2 --
 raps/sim_config.py | 1 +
 2 files changed, 1 insertion(+), 2 deletions(-)

diff --git a/raps/engine.py b/raps/engine.py
index 90b84b0..d51905d 100644
--- a/raps/engine.py
+++ b/raps/engine.py
@@ -232,8 +232,6 @@ class Engine:
         sim_config_args = sim_config.get_legacy_args()
         sim_config_dict = sim_config.get_legacy_args_dict()
         sim_config_dict['config'] = system_config_dict
-        if partition:
-            sim_config_dict["system"] = sim_config.system_name
 
         if sim_config.seed:
             random.seed(sim_config.seed)
diff --git a/raps/sim_config.py b/raps/sim_config.py
index 300ac94..f1cb6ea 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -340,6 +340,7 @@ class SimConfig(RAPSBaseModel):
         contains the SimConfig object itself.
         """
         args_dict = self.model_dump(mode="json")
+        args_dict['system'] = self.system_name
         # validate has been renamed to power_scope
         args_dict['validate'] = args_dict["power_scope"] == "node"
         args_dict['downscale'] = self.downscale
-- 
GitLab


From 25b406c821b2620fae2daad9bb293b8b2490feb0 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 11:10:52 -0400
Subject: [PATCH 05/21] Add base to SystemConfig

---
 raps/system_config.py | 24 +++++++++++++++++++++---
 raps/utils.py         | 10 ++++++++++
 2 files changed, 31 insertions(+), 3 deletions(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index adbf470..c85867b 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -6,7 +6,7 @@ from pathlib import Path
 from functools import cached_property
 import yaml
 from pydantic import computed_field, model_validator, field_validator
-from raps.utils import RAPSBaseModel
+from raps.utils import RAPSBaseModel, deep_merge
 from raps.raps_config import raps_config
 
 # Define Pydantic models for the config to handle parsing and validation
@@ -168,6 +168,12 @@ class SystemConfig(RAPSBaseModel):
     system_name: str
     """ Name of the system, defaults to the yaml file name """
 
+    base: str |  None = None
+    """
+    Optional, name or path to another SystemConfig to "inherit" from. Lets you make small modifications
+    to an existing system without having to copy the whole config.
+    """
+
     system: SystemSystemConfig
     power: SystemPowerConfig
     scheduler: SystemSchedulerConfig
@@ -175,6 +181,13 @@ class SystemConfig(RAPSBaseModel):
     cooling: SystemCoolingConfig | None = None
     network: SystemNetworkConfig | None = None
 
+    @model_validator(mode="before")
+    def _load_base(cls, data):
+        if data.get("base"):
+            base = get_system_config(data['base'])
+            data = deep_merge(base.model_dump(mode='json'), data)
+        return data
+
     def get_legacy(self) -> dict[str, Any]:
         """
         Return the system config as a flattened, uppercased dict. This is for backwards
@@ -248,10 +261,12 @@ def get_system_config(system: str) -> SystemConfig:
         "system_name": system_name,  # You can override system_name in the yaml as well
         **yaml.safe_load(config_path.read_text()),
     }
+    if str(config.get('base', '')).endswith(".yaml"):
+        config['base'] = config_path.parent / str(config['base']) # path relative to yaml
     return SystemConfig.model_validate(config)
 
 
-def get_partition_configs(partitions: list[str]) -> MultiPartitionSystemConfig:
+def get_partition_configs(partitions: list[str|SystemConfig]) -> MultiPartitionSystemConfig:
     """
     Resolves multiple partition config files. Can pass globs, or directories to include all yaml
     files under the directory.
@@ -262,7 +277,10 @@ def get_partition_configs(partitions: list[str]) -> MultiPartitionSystemConfig:
 
     parsed_configs: list[SystemConfig] = []
     for pat in partitions:
-        if pat in multi_partition_systems:
+        if isinstance(pat, SystemConfig):
+            parsed_configs.append(pat)
+            combined_system_name.append(pat.system_name)
+        elif pat in multi_partition_systems:
             matched_systems = fnmatch.filter(systems, f"{pat}/*")
             combined_system_name.append(pat)
         elif fnmatch.filter(systems, pat):
diff --git a/raps/utils.py b/raps/utils.py
index dbb5ea3..ef2f7ef 100644
--- a/raps/utils.py
+++ b/raps/utils.py
@@ -27,6 +27,16 @@ import yaml
 from raps.job import Job
 
 
+def deep_merge(a: dict, b: dict):
+    a = {**a}
+    for key in b.keys():
+        if key in a and isinstance(a[key], dict) and isinstance(b[key], dict):
+            a[key] = deep_merge(a[key], b[key])
+        else:
+            a[key] = b[key]
+    return a
+
+
 def sum_values(values):
     return sum(x[1] for x in values) if values else 0
 
-- 
GitLab


From 4d075aa43c171af7ca3d0dc8476b029ded854617 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 12:46:50 -0400
Subject: [PATCH 06/21] Add ability to override system settings

---
 raps/sim_config.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/raps/sim_config.py b/raps/sim_config.py
index f1cb6ea..cfdde03 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -16,9 +16,12 @@ Distribution = Literal['uniform', 'weibull', 'normal']
 
 
 class SimConfig(RAPSBaseModel):
-    system: str | None = None
-    """ System config to use """
-    partitions: list[str] = []
+    system: SystemConfig | str | None = None
+    """
+    System config to use. Either the name of one of the redefined systems, or specify the full
+    system. You can modify a system with system.base, e.g. `--system.base frontier --system.cooling.fmu-path my.fmu`
+    """
+    partitions: list[SystemConfig|str] = []
     """ List of multiple system configurations for a multi-partition run. Can contain wildcards """
 
     cooling: bool = False
-- 
GitLab


From be3cfb520ee694b41b26294d56d6e7c6c9957b46 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 13:09:28 -0400
Subject: [PATCH 07/21] Fix bug

---
 raps/system_config.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index c85867b..b118eef 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -183,7 +183,7 @@ class SystemConfig(RAPSBaseModel):
 
     @model_validator(mode="before")
     def _load_base(cls, data):
-        if data.get("base"):
+        if isinstance(data, dict) and data.get("base"):
             base = get_system_config(data['base'])
             data = deep_merge(base.model_dump(mode='json'), data)
         return data
-- 
GitLab


From 9f601385a7d2315aba4b9d83bcd4758bede7e3d9 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 13:50:02 -0400
Subject: [PATCH 08/21] Don't use computed_field in SystemConfig

computed_field is always serialized which causes some complications when saving SimConfigs and SystemConfigs.
Instead add them to the legacy dict explicitly
---
 raps/system_config.py | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index b118eef..679b3fd 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -5,7 +5,7 @@ from typing import Any, Literal
 from pathlib import Path
 from functools import cached_property
 import yaml
-from pydantic import computed_field, model_validator, field_validator
+from pydantic import model_validator, field_validator
 from raps.utils import RAPSBaseModel, deep_merge
 from raps.raps_config import raps_config
 
@@ -42,27 +42,22 @@ class SystemSystemConfig(RAPSBaseModel):
         self.down_nodes = sorted(set(self.down_nodes))
         return self
 
-    @computed_field
     @cached_property
     def num_racks(self) -> int:
         return self.num_cdus * self.racks_per_cdu - len(self.missing_racks)
 
-    @computed_field
     @cached_property
     def sc_shape(self) -> list[int]:
         return [self.num_cdus, self.racks_per_cdu, self.nodes_per_rack]
 
-    @computed_field
     @cached_property
     def total_nodes(self) -> int:
         return self.num_cdus * self.racks_per_cdu * self.nodes_per_rack
 
-    @computed_field
     @cached_property
     def blades_per_chassis(self) -> int:
         return int(self.nodes_per_rack / self.chassis_per_rack / self.nodes_per_blade)
 
-    @computed_field
     @cached_property
     def power_df_header(self) -> list[str]:
         power_df_header = ["CDU"]
@@ -74,7 +69,6 @@ class SystemSystemConfig(RAPSBaseModel):
         power_df_header.append("Loss")
         return power_df_header
 
-    @computed_field
     @cached_property
     def available_nodes(self) -> int:
         return self.total_nodes - len(self.down_nodes)
@@ -195,6 +189,8 @@ class SystemConfig(RAPSBaseModel):
         gradually. The dict also as a "system_config" key that contains the SystemConfig object
         itself.
         """
+        dump = self.model_dump(mode="json", exclude_none=True)
+
         renames = {  # fields that need to be renamed to something other than just .upper()
             "system_name": "system_name",
             "w_htwps_key": "W_HTWPs_KEY",
@@ -202,7 +198,6 @@ class SystemConfig(RAPSBaseModel):
             "w_cts_key": "W_CTs_KEY",
             "multitenant": "multitenant",
         }
-        dump = self.model_dump(mode="json", exclude_none=True)
 
         config_dict: dict[str, Any] = {}
         for k, v in dump.items():  # flatten
@@ -210,6 +205,13 @@ class SystemConfig(RAPSBaseModel):
                 config_dict.update(v)
             else:
                 config_dict[k] = v
+        config_dict["num_racks"] = self.system.num_racks
+        config_dict["sc_shape"] = self.system.sc_shape
+        config_dict["total_nodes"] = self.system.total_nodes
+        config_dict["blades_per_chassis"] = self.system.blades_per_chassis
+        config_dict["power_df_header"] = self.system.power_df_header
+        config_dict["available_nodes"] = self.system.available_nodes
+
         # rename keys
         config_dict = {renames.get(k, k.upper()): v for k, v in config_dict.items()}
         config_dict['system_config'] = self
-- 
GitLab


From 8745bf7be63786ca647ca28346d1650a072b69f2 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 16:42:36 -0400
Subject: [PATCH 09/21] Only output fields not in base

---
 raps/system_config.py | 14 ++++++++++++++
 raps/utils.py         | 17 +++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/raps/system_config.py b/raps/system_config.py
index 679b3fd..928385f 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -7,6 +7,11 @@ from functools import cached_property
 import yaml
 from pydantic import model_validator, field_validator
 from raps.utils import RAPSBaseModel, deep_merge
+from pydantic import (
+    model_validator, field_validator, model_serializer, SerializationInfo,
+    SerializerFunctionWrapHandler,
+)
+from raps.utils import RAPSBaseModel, deep_merge, deep_subtract_dicts
 from raps.raps_config import raps_config
 
 # Define Pydantic models for the config to handle parsing and validation
@@ -182,6 +187,15 @@ class SystemConfig(RAPSBaseModel):
             data = deep_merge(base.model_dump(mode='json'), data)
         return data
 
+    @model_serializer(mode='wrap')
+    def model_serializer(self, handler: SerializerFunctionWrapHandler, info: SerializationInfo):
+        # don't include the base system data in the output
+        if self.base and (info.exclude_defaults or info.exclude_unset):
+            base = get_system_config(self.base)
+            return deep_subtract_dicts(handler(self), handler(base))
+        else:
+            return handler(self)
+
     def get_legacy(self) -> dict[str, Any]:
         """
         Return the system config as a flattened, uppercased dict. This is for backwards
diff --git a/raps/utils.py b/raps/utils.py
index ef2f7ef..afc76cf 100644
--- a/raps/utils.py
+++ b/raps/utils.py
@@ -37,6 +37,23 @@ def deep_merge(a: dict, b: dict):
     return a
 
 
+def deep_subtract_dicts(a: dict, b: dict):
+    """
+    Remove all fields from a that are already in b, such that
+    deep_merge(deep_subtract_dicts(a, b), b) == a
+    a should contain a superset of b's keys.
+    """
+    a = {**a}
+    for key in b.keys():
+        if key in a:
+            if a[key] == b[key]:
+                a.pop(key)
+            elif isinstance(a[key], dict) and isinstance(b[key], dict):
+                a[key] = deep_subtract_dicts(a[key], b[key])
+            # otherwise keep key in a as is
+    return a
+
+
 def sum_values(values):
     return sum(x[1] for x in values) if values else 0
 
-- 
GitLab


From d79b8a998eb8b8821964fbeccf1eb525c2c41f5e Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 16:46:18 -0400
Subject: [PATCH 10/21] Formatting

---
 raps/dataloaders/mit_supercloud/loader.py | 3 +--
 raps/sim_config.py                        | 3 ++-
 raps/system_config.py                     | 8 +++-----
 raps/utils.py                             | 8 ++++----
 4 files changed, 10 insertions(+), 12 deletions(-)

diff --git a/raps/dataloaders/mit_supercloud/loader.py b/raps/dataloaders/mit_supercloud/loader.py
index 2c8dbc1..ab68eb7 100644
--- a/raps/dataloaders/mit_supercloud/loader.py
+++ b/raps/dataloaders/mit_supercloud/loader.py
@@ -119,7 +119,7 @@ from collections import Counter
 from datetime import datetime, timezone
 
 from raps.job import job_dict, Job
-from raps.utils import summarize_ranges, next_arrival, WorkloadData
+from raps.utils import summarize_ranges, WorkloadData
 from .utils import proc_cpu_series, proc_gpu_series, to_epoch
 from .utils import DEFAULT_START, DEFAULT_END
 
@@ -211,7 +211,6 @@ def load_data(local_dataset_path, **kwargs):
     """
     debug = kwargs.get("debug")
     config = kwargs.get("config")
-    arrival = kwargs.get("arrival")
     NL_PATH = os.path.dirname(__file__)
 
     skip_counts = Counter()
diff --git a/raps/sim_config.py b/raps/sim_config.py
index cfdde03..7be5258 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -21,7 +21,7 @@ class SimConfig(RAPSBaseModel):
     System config to use. Either the name of one of the redefined systems, or specify the full
     system. You can modify a system with system.base, e.g. `--system.base frontier --system.cooling.fmu-path my.fmu`
     """
-    partitions: list[SystemConfig|str] = []
+    partitions: list[SystemConfig | str] = []
     """ List of multiple system configurations for a multi-partition run. Can contain wildcards """
 
     cooling: bool = False
@@ -74,6 +74,7 @@ class SimConfig(RAPSBaseModel):
     """ If True, output to a randomly named directory. Mutually exclusive with output """
 
     _random_output: Path | None = None
+
     def get_output(self):
         if self.random_output and not self._random_output:
             self._random_output = Path(create_casename("out-")).resolve()
diff --git a/raps/system_config.py b/raps/system_config.py
index 928385f..663e789 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -5,8 +5,6 @@ from typing import Any, Literal
 from pathlib import Path
 from functools import cached_property
 import yaml
-from pydantic import model_validator, field_validator
-from raps.utils import RAPSBaseModel, deep_merge
 from pydantic import (
     model_validator, field_validator, model_serializer, SerializationInfo,
     SerializerFunctionWrapHandler,
@@ -167,7 +165,7 @@ class SystemConfig(RAPSBaseModel):
     system_name: str
     """ Name of the system, defaults to the yaml file name """
 
-    base: str |  None = None
+    base: str | None = None
     """
     Optional, name or path to another SystemConfig to "inherit" from. Lets you make small modifications
     to an existing system without having to copy the whole config.
@@ -278,11 +276,11 @@ def get_system_config(system: str) -> SystemConfig:
         **yaml.safe_load(config_path.read_text()),
     }
     if str(config.get('base', '')).endswith(".yaml"):
-        config['base'] = config_path.parent / str(config['base']) # path relative to yaml
+        config['base'] = config_path.parent / str(config['base'])  # path relative to yaml
     return SystemConfig.model_validate(config)
 
 
-def get_partition_configs(partitions: list[str|SystemConfig]) -> MultiPartitionSystemConfig:
+def get_partition_configs(partitions: list[str | SystemConfig]) -> MultiPartitionSystemConfig:
     """
     Resolves multiple partition config files. Can pass globs, or directories to include all yaml
     files under the directory.
diff --git a/raps/utils.py b/raps/utils.py
index afc76cf..ab02a2a 100644
--- a/raps/utils.py
+++ b/raps/utils.py
@@ -708,10 +708,10 @@ def pydantic_add_args(
     def model_validate_args(args: argparse.Namespace, data: dict | None = None):
         try:
             model = CliApp.run(SettingsModel,
-                            cli_args=args,
-                            cli_settings_source=cli_settings_source,
-                            **(data or {}),
-                            )
+                               cli_args=args,
+                               cli_settings_source=cli_settings_source,
+                               **(data or {}),
+                               )
             # Recreate model so we don't return the SettingsModel subclass
             return model_cls.model_validate(model.model_dump())
         except ValidationError as err:
-- 
GitLab


From e45d9666d53c6e2b41314e5318e185648438ef90 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 16:51:24 -0400
Subject: [PATCH 11/21] Output files by default

---
 raps/engine.py     |  1 -
 raps/run_sim.py    |  1 -
 raps/sim_config.py | 26 +++++++++++++++-----------
 3 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/raps/engine.py b/raps/engine.py
index d51905d..3c83c84 100644
--- a/raps/engine.py
+++ b/raps/engine.py
@@ -161,7 +161,6 @@ class Engine:
         self.flops_manager = flops_manager
         self.debug = sim_config.debug
         self.continuous_workload = continuous_workload
-        self.output = sim_config.get_output()
         self.replay = sim_config.replay
         self.downscale = sim_config.downscale  # Factor to downscale the 1s timesteps (power of 10)
         self.simulate_network = sim_config.simulate_network
diff --git a/raps/run_sim.py b/raps/run_sim.py
index 6c39a2f..9846eed 100644
--- a/raps/run_sim.py
+++ b/raps/run_sim.py
@@ -44,7 +44,6 @@ shortcuts = {
     "numjobs": "n",
     "verbose": "v",
     "output": "o",
-    "random-output": "O",
     "uncertainties": "u",
     "plot": "p",
     "replay": "f",
diff --git a/raps/sim_config.py b/raps/sim_config.py
index 7be5258..2a31724 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -68,17 +68,24 @@ class SimConfig(RAPSBaseModel):
     seed: int | None = None
     """ Set RNG seed for deterministic simulation """
 
-    output: ExpandedPath | None = None
-    """ Output power, cooling, and loss models for later analysis. Argument specifies name. """
-    random_output: bool = False
-    """ If True, output to a randomly named directory. Mutually exclusive with output """
+    output: ExpandedPath | Literal['none'] | None = None
+    """
+    Where to output power, cooling, and loss models for later analysis.
+    If omitted it will output to raps-output-<id> by default.
+    Set to "none" to disable file output entirely.
+    """
 
     _random_output: Path | None = None
 
-    def get_output(self):
-        if self.random_output and not self._random_output:
-            self._random_output = Path(create_casename("out-")).resolve()
-        return self._random_output or self.output
+    def get_output(self) -> Path | None:
+        if self.output is None:  # by default, output to a random directory
+            if not self._random_output:
+                self._random_output = Path(create_casename("raps-output-")).resolve()
+            return self._random_output
+        elif self.output == "none":  # allow explicitly disabling output with "none"
+            return None
+        else:
+            return self.output  # return user defined output path
 
     debug: bool = False
     """ Enable debug mode and disable rich layout """
@@ -306,9 +313,6 @@ class SimConfig(RAPSBaseModel):
                 raise ValueError(f"policy {self.backfill} not implemented by {self.scheduler}. "
                                  f"Valid selections: {sorted(valid_backfilltypes)}")
 
-        if self.random_output and self.output:
-            raise ValueError("--random-output and --output are mutually exclusive")
-
         return self
 
     @property
-- 
GitLab


From 7a567ae9813a7372bba72ee908bc47de5d4aa946 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 17:32:44 -0400
Subject: [PATCH 12/21] Output sim config

---
 raps/run_sim.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/raps/run_sim.py b/raps/run_sim.py
index 9846eed..93a2cf8 100644
--- a/raps/run_sim.py
+++ b/raps/run_sim.py
@@ -87,6 +87,9 @@ def run_sim(sim_config: SimConfig):
             result=workload_data,
             args=sim_config,
         )
+        config_yaml = yaml_dump(sim_config.model_dump(mode="json", exclude_defaults=True))
+        (out / 'sim_config.yaml').write_text(config_yaml)
+
     jobs = workload_data.jobs
     timestep_start, timestep_end = workload_data.telemetry_start, workload_data.telemetry_end
     total_timesteps = timestep_end - timestep_start
@@ -243,7 +246,6 @@ def run_parts_sim_add_parser(subparsers: SubParsers):
 
 
 def run_parts_sim(sim_config: SimConfig):
-
     if len(sim_config.system_configs) == 1:
         warnings.warn(
             "run_parts_sim is usually for multiple partitions. Did you mean to run with one?",
@@ -261,6 +263,9 @@ def run_parts_sim(sim_config: SimConfig):
                 result=workload_results[part],
                 args=sim_config,
             )
+        config_yaml = yaml_dump(sim_config.model_dump(mode="json", exclude_defaults=True))
+        (out / 'sim_config.yaml').write_text(config_yaml)
+
     jobs = {p: w.jobs for p, w in workload_results.items()}
 
     ui_update_freq = sim_config.system_configs[0].scheduler.ui_update_freq
-- 
GitLab


From 5473162c6622cb0083aa2c6aff7c7e76ee2c5b58 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 17:47:45 -0400
Subject: [PATCH 13/21] Pass SystemConfig through in get_system_config

---
 raps/system_config.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index 663e789..e621d4b 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -255,13 +255,15 @@ def list_systems() -> list[str]:
     ])
 
 
-@functools.cache
-def get_system_config(system: str) -> SystemConfig:
+def get_system_config(system: str|SystemConfig) -> SystemConfig:
     """
     Returns the system config as a Pydantic object.
     system can either be a path to a custom .yaml file, or the name of one of the pre-configured
     systems defined in RAPS_SYSTEM_CONFIG_DIR.
     """
+    if isinstance(system, SystemConfig): # Just pass system through if its already parsed
+        return system
+
     if system in list_systems():
         config_path = raps_config.system_config_dir / f"{system}.yaml"
         system_name = system
-- 
GitLab


From 4a80ebcba640a88b2da21036a662d08263c69dbf Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 17:56:46 -0400
Subject: [PATCH 14/21] Separate configs for single and multi run

---
 raps/engine.py            |  6 ++--
 raps/multi_part_engine.py |  4 +--
 raps/run_sim.py           | 16 +++++------
 raps/sim_config.py        | 59 ++++++++++++++++++++++++---------------
 raps/workload.py          |  6 ++--
 5 files changed, 52 insertions(+), 39 deletions(-)

diff --git a/raps/engine.py b/raps/engine.py
index 3c83c84..3d27848 100644
--- a/raps/engine.py
+++ b/raps/engine.py
@@ -38,7 +38,7 @@ from raps.workload import Workload, continuous_job_generation
 from raps.account import Accounts
 from raps.downtime import Downtime
 from raps.weather import Weather
-from raps.sim_config import SimConfig
+from raps.sim_config import SingleSimConfig
 from raps.system_config import SystemConfig
 from bisect import bisect_right
 
@@ -135,7 +135,7 @@ class Engine:
                  # Workload class to generate from for continuous generation
                  continuous_workload: Workload | None = None,
                  accounts=None,
-                 sim_config: SimConfig,
+                 sim_config: SingleSimConfig,
                  system_config: SystemConfig,
                  ):
         self.config = system_config.get_legacy()
@@ -212,7 +212,7 @@ class Engine:
             self.network_model = None
 
     @staticmethod
-    def from_sim_config(sim_config: SimConfig, partition: str | None = None):
+    def from_sim_config(sim_config: SingleSimConfig, partition: str | None = None):
         if partition:
             system_config_by_name = {s.system_name: s for s in sim_config.system_configs}
             system_config = system_config_by_name.get(partition)
diff --git a/raps/multi_part_engine.py b/raps/multi_part_engine.py
index 944ced9..57e3e27 100644
--- a/raps/multi_part_engine.py
+++ b/raps/multi_part_engine.py
@@ -1,6 +1,6 @@
 from collections.abc import Iterable
 from raps.engine import Engine, TickData
-from raps.sim_config import SimConfig
+from raps.sim_config import MultiPartSimConfig
 from raps.utils import WorkloadData
 
 
@@ -11,7 +11,7 @@ class MultiPartEngine:
         self.jobs = jobs
 
     @staticmethod
-    def from_sim_config(sim_config: SimConfig):
+    def from_sim_config(sim_config: MultiPartSimConfig):
         if sim_config.replay:
             root_systems = set(s.system_name.split("/")[0] for s in sim_config.system_configs)
             # TODO should consider how to pass separate replay values for separate systems
diff --git a/raps/run_sim.py b/raps/run_sim.py
index 93a2cf8..06af0b4 100644
--- a/raps/run_sim.py
+++ b/raps/run_sim.py
@@ -22,7 +22,7 @@ from raps.stats import (
     print_formatted_report
 )
 
-from raps.sim_config import SimConfig
+from raps.sim_config import SingleSimConfig, MultiPartSimConfig
 
 
 def read_yaml(config_file: str):
@@ -62,7 +62,7 @@ def run_sim_add_parser(subparsers: SubParsers):
         YAML sim config file, can be used to configure an experiment instead of using CLI
         flags. Pass "-" to read from stdin.
     """)
-    model_validate = pydantic_add_args(parser, SimConfig, model_config={
+    model_validate = pydantic_add_args(parser, SingleSimConfig, model_config={
         "cli_shortcuts": shortcuts,
     })
     parser.set_defaults(
@@ -70,9 +70,9 @@ def run_sim_add_parser(subparsers: SubParsers):
     )
 
 
-def run_sim(sim_config: SimConfig):
+def run_sim(sim_config: SingleSimConfig):
     if sim_config.verbose or sim_config.debug:
-        print(f"SimConfig: {sim_config.model_dump_json(indent=4)}")
+        print(f"SingleSimConfig: {sim_config.model_dump_json(indent=4)}")
     if len(sim_config.system_configs) > 1:
         print("Use run-parts to run multi-partition simulations")
         sys.exit(1)
@@ -237,7 +237,7 @@ def run_parts_sim_add_parser(subparsers: SubParsers):
         YAML sim config file, can be used to configure an experiment instead of using CLI
         flags. Pass "-" to read from stdin.
     """)
-    model_validate = pydantic_add_args(parser, SimConfig, model_config={
+    model_validate = pydantic_add_args(parser, MultiPartSimConfig, model_config={
         "cli_shortcuts": shortcuts,
     })
     parser.set_defaults(
@@ -245,7 +245,7 @@ def run_parts_sim_add_parser(subparsers: SubParsers):
     )
 
 
-def run_parts_sim(sim_config: SimConfig):
+def run_parts_sim(sim_config: MultiPartSimConfig):
     if len(sim_config.system_configs) == 1:
         warnings.warn(
             "run_parts_sim is usually for multiple partitions. Did you mean to run with one?",
@@ -323,7 +323,7 @@ def show_add_parser(subparsers: SubParsers):
     parser.add_argument("--show-defaults", default=False, help="""
         If true, include defaults in the output YAML
     """)
-    model_validate = pydantic_add_args(parser, SimConfig, model_config={
+    model_validate = pydantic_add_args(parser, SingleSimConfig, model_config={
         "cli_shortcuts": shortcuts,
     })
 
@@ -334,6 +334,6 @@ def show_add_parser(subparsers: SubParsers):
     parser.set_defaults(impl=impl)
 
 
-def show(sim_config: SimConfig, show_defaults=False):
+def show(sim_config: SingleSimConfig, show_defaults=False):
     data = sim_config.model_dump(mode="json", exclude_defaults=not show_defaults)
     print(yaml_dump(data), end="")
diff --git a/raps/sim_config.py b/raps/sim_config.py
index 2a31724..6aec463 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -1,4 +1,5 @@
 import argparse
+import abc
 from pathlib import Path
 from functools import cached_property
 from datetime import timedelta
@@ -9,21 +10,13 @@ from raps.utils import (
     parse_time_unit, convert_to_time_unit, infer_time_unit, ExpandedPath, parse_td, create_casename,
     RAPSBaseModel,
 )
-from raps.system_config import SystemConfig, get_partition_configs
+from raps.system_config import SystemConfig, get_partition_configs, get_system_config
 from pydantic import model_validator
 
 Distribution = Literal['uniform', 'weibull', 'normal']
 
 
-class SimConfig(RAPSBaseModel):
-    system: SystemConfig | str | None = None
-    """
-    System config to use. Either the name of one of the redefined systems, or specify the full
-    system. You can modify a system with system.base, e.g. `--system.base frontier --system.cooling.fmu-path my.fmu`
-    """
-    partitions: list[SystemConfig | str] = []
-    """ List of multiple system configurations for a multi-partition run. Can contain wildcards """
-
+class SimConfig(RAPSBaseModel, abc.ABC):
     cooling: bool = False
     """ Include the FMU cooling model """
     simulate_network: bool = False
@@ -264,12 +257,6 @@ class SimConfig(RAPSBaseModel):
 
     @model_validator(mode="after")
     def _validate_after(self):
-        # This is called after Pydantic has parsed everything into the model
-        if self.system and self.partitions:
-            raise ValueError("system and partitions are mutually exclusive")
-        elif not self.system and not self.partitions:
-            self.system = "frontier"
-
         if not self.replay and not self.workload:
             self.workload = "random"
 
@@ -316,24 +303,22 @@ class SimConfig(RAPSBaseModel):
         return self
 
     @property
+    @abc.abstractmethod
     def system_name(self) -> str:
         """
         Name of the system.
-        Note, this is different than system, as system can be a file or None if partition is set.
+        Note, this is different than system, as system can be a file, or there can be multiple systems
         """
-        return self._multi_partition_system_config.system_name
+        pass
 
     @property
+    @abc.abstractmethod
     def system_configs(self) -> list[SystemConfig]:
         """
         Return the SystemConfigs for the selected systems.
         Will be a single element array unless multiple `partitions` are selected.
         """
-        return self._multi_partition_system_config.partitions
-
-    @cached_property
-    def _multi_partition_system_config(self):
-        return get_partition_configs(self.partitions if self.partitions else [self.system])
+        pass
 
     def get_legacy_args(self):
         """
@@ -363,3 +348,31 @@ class SimConfig(RAPSBaseModel):
 
         args_dict['sim_config'] = self
         return args_dict
+
+
+class SingleSimConfig(SimConfig, abc.ABC):
+    system: SystemConfig | str = "frontier"
+
+    @property
+    def system_name(self) -> str:
+        return self.system_configs[0].system_name
+
+    @cached_property
+    def system_configs(self) -> list[SystemConfig]:
+        return [get_system_config(self.system)]
+
+
+class MultiPartSimConfig(SimConfig):
+    partitions: list[SystemConfig | str]
+
+    @property
+    def system_name(self) -> str:
+        return self._multi_partition_system_config.system_name
+
+    @property
+    def system_configs(self) -> list[SystemConfig]:
+        return self._multi_partition_system_config.partitions
+
+    @cached_property
+    def _multi_partition_system_config(self):
+        return get_partition_configs(self.partitions)
diff --git a/raps/workload.py b/raps/workload.py
index 57b536d..78d8d2a 100644
--- a/raps/workload.py
+++ b/raps/workload.py
@@ -40,7 +40,7 @@ import matplotlib.pyplot as plt
 from raps.telemetry import Telemetry
 from raps.job import job_dict, Job
 from raps.utils import create_file_indexed, SubParsers, pydantic_add_args
-from raps.sim_config import SimConfig
+from raps.sim_config import SingleSimConfig
 
 
 JOB_NAMES = ["LAMMPS", "GROMACS", "VASP", "Quantum ESPRESSO", "NAMD",
@@ -965,13 +965,13 @@ def run_workload_add_parser(subparsers: SubParsers):
         YAML sim config file, can be used to configure an experiment instead of using CLI
         flags. Pass "-" to read from stdin.
     """)
-    model_validate = pydantic_add_args(parser, SimConfig, model_config={
+    model_validate = pydantic_add_args(parser, SingleSimConfig, model_config={
         "cli_shortcuts": shortcuts,
     })
     parser.set_defaults(impl=lambda args: run_workload(model_validate(args, {})))
 
 
-def run_workload(sim_config: SimConfig):
+def run_workload(sim_config: SingleSimConfig):
     args = sim_config.get_legacy_args()
     args_dict = sim_config.get_legacy_args()
     config = sim_config.system_configs[0].get_legacy()
-- 
GitLab


From a9fade70cf5deb32e425e78214abb9bbb4b99b3f Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 18:31:01 -0400
Subject: [PATCH 15/21] Fix tests

---
 raps/run_sim.py                                      |  3 ++-
 tests/smoke.py                                       |  2 +-
 tests/systems/test_engine.py                         |  4 ++--
 tests/systems/test_main_basic_run.py                 |  2 +-
 tests/systems/test_main_time_delta_sub_second_run.py |  8 +-------
 tests/systems/test_multi_part_sim_basic_run.py       |  3 ++-
 tests/systems/test_multi_part_sim_withdata_run.py    |  3 ++-
 tests/test_main.py                                   | 11 ++++++-----
 8 files changed, 17 insertions(+), 19 deletions(-)

diff --git a/raps/run_sim.py b/raps/run_sim.py
index 06af0b4..ce89529 100644
--- a/raps/run_sim.py
+++ b/raps/run_sim.py
@@ -83,7 +83,7 @@ def run_sim(sim_config: SingleSimConfig):
     if out:
         out.mkdir(parents=True)
         engine.telemetry.save_snapshot(
-            dest=str(out),
+            dest=str(out / 'snapshot.npz'),
             result=workload_data,
             args=sim_config,
         )
@@ -257,6 +257,7 @@ def run_parts_sim(sim_config: MultiPartSimConfig):
 
     out = sim_config.get_output()
     if out:
+        out.mkdir(parents=True)
         for part, engine in multi_engine.engines.items():
             engine.telemetry.save_snapshot(
                 dest=str(out / part.split('/')[-1]),
diff --git a/tests/smoke.py b/tests/smoke.py
index 946f6db..a2ea598 100644
--- a/tests/smoke.py
+++ b/tests/smoke.py
@@ -32,7 +32,7 @@ def run_command(command):
 def build_command(system, file_paths, additional_args=""):
     """Build the command string for the given system and file paths."""
     full_paths = " ".join([os.path.join(DATAPATH, path) for path in file_paths.split()])
-    return f"python main.py run --system {system} -f {full_paths} -t {DEFAULT_TIME} {additional_args}".strip()
+    return f"python main.py run --system {system} -f {full_paths} -t {DEFAULT_TIME} -o none {additional_args}".strip()
 
 
 def execute_system_tests(systems):
diff --git a/tests/systems/test_engine.py b/tests/systems/test_engine.py
index e483b18..0404e89 100644
--- a/tests/systems/test_engine.py
+++ b/tests/systems/test_engine.py
@@ -1,6 +1,6 @@
 import pytest
 from raps.engine import Engine
-from raps.sim_config import SimConfig
+from raps.sim_config import SingleSimConfig
 from raps.stats import (
     get_engine_stats,
     # get_job_stats,
@@ -18,7 +18,7 @@ def test_engine(system, system_config, sim_output):
     if not system_config.get("main", False):
         pytest.skip(f"{system} does not support basic main run.")
 
-    sim_config = SimConfig.model_validate({
+    sim_config = SingleSimConfig.model_validate({
         "system": system,
         "time": "2m",
     })
diff --git a/tests/systems/test_main_basic_run.py b/tests/systems/test_main_basic_run.py
index 0cc9b69..37661f3 100644
--- a/tests/systems/test_main_basic_run.py
+++ b/tests/systems/test_main_basic_run.py
@@ -19,6 +19,6 @@ def test_main_basic_run(system, system_config, sim_output):
         "python", "main.py", "run",
         "--time", "1m",
         "--system", system,
-        "-o", sim_output
+        "-o", sim_output,
     ], capture_output=True, text=True, stdin=subprocess.DEVNULL)
     assert result.returncode == 0, f"Failed on {system}: {result.stderr}"
diff --git a/tests/systems/test_main_time_delta_sub_second_run.py b/tests/systems/test_main_time_delta_sub_second_run.py
index 55c0e3c..db80105 100644
--- a/tests/systems/test_main_time_delta_sub_second_run.py
+++ b/tests/systems/test_main_time_delta_sub_second_run.py
@@ -34,17 +34,11 @@ def test_main_time_delta_sub_second_run(system, system_config, time_arg, tdelta_
         "--time-delta", tdelta_arg,
         "--system", system,
         "--noui",
-        "-o", sim_output
+        "-o", sim_output,
     ], capture_output=True, text=True, stdin=subprocess.DEVNULL)
     assert result.returncode == 0, f"Failed on {system}: {result.stderr}"
     time = parse_td(time_arg).seconds
     assert f"Time Simulated: {convert_seconds_to_hhmmss(time)}" in result.stdout
 
-    subprocess.run(
-        f"rm {sim_output}.npz && rm -fr simulation_results/{sim_output}",
-        shell=True,
-        check=True
-    )
-
     del result
     gc.collect()
diff --git a/tests/systems/test_multi_part_sim_basic_run.py b/tests/systems/test_multi_part_sim_basic_run.py
index 9351fd6..0edcc90 100644
--- a/tests/systems/test_multi_part_sim_basic_run.py
+++ b/tests/systems/test_multi_part_sim_basic_run.py
@@ -11,7 +11,7 @@ pytestmark = [
 ]
 
 
-def test_multi_part_sim_basic_run(system, system_config):
+def test_multi_part_sim_basic_run(system, system_config, sim_output):
 
     if not system_config.get("multi-part-sim", False):
         pytest.skip(f"{system} does not support basic multi-part-sim run.")
@@ -21,6 +21,7 @@ def test_multi_part_sim_basic_run(system, system_config):
         "python", "main.py", "run-parts",
         "--time", "1h",
         "-x", f"{system}/*",
+        "-o", sim_output,
     ], capture_output=True, text=True, stdin=subprocess.DEVNULL)
     assert result.returncode == 0, f"Failed on {system}: {result.stderr}"
     del result
diff --git a/tests/systems/test_multi_part_sim_withdata_run.py b/tests/systems/test_multi_part_sim_withdata_run.py
index f38cf8e..ab62f93 100644
--- a/tests/systems/test_multi_part_sim_withdata_run.py
+++ b/tests/systems/test_multi_part_sim_withdata_run.py
@@ -12,7 +12,7 @@ pytestmark = [
 ]
 
 
-def test_multi_part_sim_withdata_run(system, system_config, system_files):
+def test_multi_part_sim_withdata_run(system, system_config, system_files, sim_output):
     if not system_config.get("multi-part-sim", False):
         pytest.skip(f"{system} does not support basic multi-part-sim run even without data.")
     if not system_config.get("withdata", False):
@@ -24,5 +24,6 @@ def test_multi_part_sim_withdata_run(system, system_config, system_files):
         "--time", "1h",
         "-x", f"{system}/*",
         "-f", ','.join(system_files),
+        "-o", sim_output,
     ], capture_output=True, text=True, stdin=subprocess.DEVNULL)
     assert result.returncode == 0, f"Failed on {system}: {result.stderr}"
diff --git a/tests/test_main.py b/tests/test_main.py
index 5c08182..0c31dd0 100644
--- a/tests/test_main.py
+++ b/tests/test_main.py
@@ -10,11 +10,12 @@ PROJECT_ROOT = Path(__file__).resolve().parent.parent  # adjust if needed
 
 
 @pytest.mark.order(1)
-def test_main_withui():
+def test_main_withui(sim_output):
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
                             "python", "main.py", "run",
                             "--time", "1h",
+                            "-o", sim_output,
                             ], capture_output=True,
                             text=True
                             )
@@ -22,12 +23,12 @@ def test_main_withui():
 
 
 @pytest.mark.order(2)
-def test_main_noui():
+def test_main_noui(sim_output):
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
                             "python", "main.py", "run",
                             "--time", "1h",
-                            "--noui"
+                            "--noui", "-o", sim_output,
                             ], capture_output=True,
                             text=True
                             )
@@ -36,10 +37,10 @@ def test_main_noui():
 
 @pytest.mark.long
 @pytest.mark.order(3)
-def test_main_long():
+def test_main_long(sim_output):
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
-                            "python", "main.py", "run",
+                            "python", "main.py", "run", "-o", sim_output,
                             ], capture_output=True,
                             text=True
                             )
-- 
GitLab


From 44c044eab39db223bec84be8ce38f9f67b57d866 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 20:35:29 -0400
Subject: [PATCH 16/21] Formatting

---
 raps/system_config.py                             | 4 ++--
 raps/workload.py                                  | 2 +-
 tests/systems/test_main_network_withdata_run.py   | 2 +-
 tests/systems/test_main_withdata_run.py           | 2 +-
 tests/systems/test_multi_part_sim_withdata_run.py | 3 +--
 tests/systems/test_telemetry_withdata_run.py      | 2 +-
 6 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index e621d4b..198f250 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -255,13 +255,13 @@ def list_systems() -> list[str]:
     ])
 
 
-def get_system_config(system: str|SystemConfig) -> SystemConfig:
+def get_system_config(system: str | SystemConfig) -> SystemConfig:
     """
     Returns the system config as a Pydantic object.
     system can either be a path to a custom .yaml file, or the name of one of the pre-configured
     systems defined in RAPS_SYSTEM_CONFIG_DIR.
     """
-    if isinstance(system, SystemConfig): # Just pass system through if its already parsed
+    if isinstance(system, SystemConfig):  # Just pass system through if its already parsed
         return system
 
     if system in list_systems():
diff --git a/raps/workload.py b/raps/workload.py
index 78d8d2a..2a630b2 100644
--- a/raps/workload.py
+++ b/raps/workload.py
@@ -991,7 +991,7 @@ def run_workload(sim_config: SingleSimConfig):
     if out:
         timestep_start = min([x.submit_time for x in jobs])
         timestep_end = math.ceil(max([x.submit_time for x in jobs]) + max([x.expected_run_time for x in jobs]))
-        filename = create_file_indexed('wl', path = str(out), create=False, ending="npz").split(".npz")[0]
+        filename = create_file_indexed('wl', path=str(out), create=False, ending="npz").split(".npz")[0]
         # savez_compressed add npz itself, but create_file_indexed needs to check for .npz to find existing files
         np.savez_compressed(filename, jobs=jobs, timestep_start=timestep_start, timestep_end=timestep_end, args=args)
         print(filename + ".npz")  # To std-out to show which npz was created.
diff --git a/tests/systems/test_main_network_withdata_run.py b/tests/systems/test_main_network_withdata_run.py
index 58d14f9..1cbeae8 100644
--- a/tests/systems/test_main_network_withdata_run.py
+++ b/tests/systems/test_main_network_withdata_run.py
@@ -1,7 +1,7 @@
 import os
 import subprocess
 import pytest
-from tests.util import PROJECT_ROOT, DATA_PATH
+from tests.util import PROJECT_ROOT
 
 
 pytestmark = [
diff --git a/tests/systems/test_main_withdata_run.py b/tests/systems/test_main_withdata_run.py
index eb996a3..3539db9 100644
--- a/tests/systems/test_main_withdata_run.py
+++ b/tests/systems/test_main_withdata_run.py
@@ -1,7 +1,7 @@
 import os
 import subprocess
 import pytest
-from tests.util import PROJECT_ROOT, DATA_PATH
+from tests.util import PROJECT_ROOT
 
 
 pytestmark = [
diff --git a/tests/systems/test_multi_part_sim_withdata_run.py b/tests/systems/test_multi_part_sim_withdata_run.py
index ab62f93..538726c 100644
--- a/tests/systems/test_multi_part_sim_withdata_run.py
+++ b/tests/systems/test_multi_part_sim_withdata_run.py
@@ -1,8 +1,7 @@
 import os
 import subprocess
-import gc
 import pytest
-from tests.util import PROJECT_ROOT, DATA_PATH
+from tests.util import PROJECT_ROOT
 
 
 pytestmark = [
diff --git a/tests/systems/test_telemetry_withdata_run.py b/tests/systems/test_telemetry_withdata_run.py
index 2729c7c..43a218b 100644
--- a/tests/systems/test_telemetry_withdata_run.py
+++ b/tests/systems/test_telemetry_withdata_run.py
@@ -1,7 +1,7 @@
 import os
 import subprocess
 import pytest
-from tests.util import PROJECT_ROOT, DATA_PATH
+from tests.util import PROJECT_ROOT
 
 
 pytestmark = [
-- 
GitLab


From c11a7088687b15cf359cd2c37664d9247484f087 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Fri, 5 Sep 2025 20:41:03 -0400
Subject: [PATCH 17/21] More test fixes

---
 tests/test_main.py | 32 +++++++++++++-------------------
 1 file changed, 13 insertions(+), 19 deletions(-)

diff --git a/tests/test_main.py b/tests/test_main.py
index 0c31dd0..4b09fa0 100644
--- a/tests/test_main.py
+++ b/tests/test_main.py
@@ -10,38 +10,32 @@ PROJECT_ROOT = Path(__file__).resolve().parent.parent  # adjust if needed
 
 
 @pytest.mark.order(1)
-def test_main_withui(sim_output):
+def test_main_withui():
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
-                            "python", "main.py", "run",
-                            "--time", "1h",
-                            "-o", sim_output,
-                            ], capture_output=True,
-                            text=True
-                            )
+        "python", "main.py", "run",
+        "--time", "1h",
+        "-o", 'none',
+    ], capture_output=True, text=True)
     assert result.returncode == 0
 
 
 @pytest.mark.order(2)
-def test_main_noui(sim_output):
+def test_main_noui():
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
-                            "python", "main.py", "run",
-                            "--time", "1h",
-                            "--noui", "-o", sim_output,
-                            ], capture_output=True,
-                            text=True
-                            )
+        "python", "main.py", "run",
+        "--time", "1h",
+        "--noui", "-o", 'none',
+    ], capture_output=True, text=True)
     assert result.returncode == 0
 
 
 @pytest.mark.long
 @pytest.mark.order(3)
-def test_main_long(sim_output):
+def test_main_long():
     os.chdir(PROJECT_ROOT)
     result = subprocess.run([
-                            "python", "main.py", "run", "-o", sim_output,
-                            ], capture_output=True,
-                            text=True
-                            )
+        "python", "main.py", "run", "-o", 'none',
+    ], capture_output=True, text=True)
     assert result.returncode == 0
-- 
GitLab


From 0ec59e9e29e0c4c78820b3c5199ebb7503b47987 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Sat, 6 Sep 2025 11:56:12 -0400
Subject: [PATCH 18/21] Fix to SystemConfig base path resolution

---
 raps/system_config.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/raps/system_config.py b/raps/system_config.py
index 198f250..bd405be 100644
--- a/raps/system_config.py
+++ b/raps/system_config.py
@@ -277,8 +277,9 @@ def get_system_config(system: str | SystemConfig) -> SystemConfig:
         "system_name": system_name,  # You can override system_name in the yaml as well
         **yaml.safe_load(config_path.read_text()),
     }
-    if str(config.get('base', '')).endswith(".yaml"):
-        config['base'] = config_path.parent / str(config['base'])  # path relative to yaml
+    base = str(config.get('base', ''))
+    if base.endswith(".yaml"):
+        config['base'] = str(config_path.parent / base)  # path relative to yaml
     return SystemConfig.model_validate(config)
 
 
-- 
GitLab


From 7415b66910d5a14efc8e387938d74374cf5ad168 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Mon, 8 Sep 2025 13:44:24 -0400
Subject: [PATCH 19/21] Fix lassen dataloader

---
 raps/dataloaders/lassen.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/raps/dataloaders/lassen.py b/raps/dataloaders/lassen.py
index fd0e364..bc57a7c 100644
--- a/raps/dataloaders/lassen.py
+++ b/raps/dataloaders/lassen.py
@@ -245,7 +245,8 @@ def load_data_from_df(allocation_df, node_df, step_df, **kwargs):
     return WorkloadData(
         jobs=job_list,
         telemetry_start=telemetry_start_time, telemetry_end=telemetry_end_time,
-        start_date=telemetry_start_timestamp,
+        # TODO: Confirm whether lassen timestamps are UTC or PDT
+        start_date=telemetry_start_timestamp.tz_localize("UTC"),
     )
 
 
-- 
GitLab


From 2b3625ab6634fc5e7f0dfbf795aa8a90e78cd051 Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Tue, 9 Sep 2025 10:49:22 -0400
Subject: [PATCH 20/21] Use base SimConfig in Engine

---
 raps/engine.py     | 11 ++++-------
 raps/sim_config.py |  6 ++++++
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/raps/engine.py b/raps/engine.py
index 3d27848..c569bd0 100644
--- a/raps/engine.py
+++ b/raps/engine.py
@@ -38,7 +38,7 @@ from raps.workload import Workload, continuous_job_generation
 from raps.account import Accounts
 from raps.downtime import Downtime
 from raps.weather import Weather
-from raps.sim_config import SingleSimConfig
+from raps.sim_config import SimConfig
 from raps.system_config import SystemConfig
 from bisect import bisect_right
 
@@ -135,7 +135,7 @@ class Engine:
                  # Workload class to generate from for continuous generation
                  continuous_workload: Workload | None = None,
                  accounts=None,
-                 sim_config: SingleSimConfig,
+                 sim_config: SimConfig,
                  system_config: SystemConfig,
                  ):
         self.config = system_config.get_legacy()
@@ -212,12 +212,9 @@ class Engine:
             self.network_model = None
 
     @staticmethod
-    def from_sim_config(sim_config: SingleSimConfig, partition: str | None = None):
+    def from_sim_config(sim_config: SimConfig, partition: str | None = None):
         if partition:
-            system_config_by_name = {s.system_name: s for s in sim_config.system_configs}
-            system_config = system_config_by_name.get(partition)
-            if not system_config:
-                raise ValueError(f"Partition {partition} isn't in SimConfig")
+            system_config = sim_config.get_system_config_by_name(partition)
         elif len(sim_config.system_configs) > 1:
             raise ValueError(
                 "Engine can only run single-partition simulations. Use MultiPartEngine for " +
diff --git a/raps/sim_config.py b/raps/sim_config.py
index 6aec463..0cbe587 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -320,6 +320,12 @@ class SimConfig(RAPSBaseModel, abc.ABC):
         """
         pass
 
+    def get_system_config_by_name(self, name: str) -> SystemConfig:
+        for s in self.system_configs:
+            if s.system_name == name:
+                return s
+        raise ValueError(f"Partition {name} isn't in SimConfig")
+
     def get_legacy_args(self):
         """
         Return as an argparse.Namespace object for backwards compatability
-- 
GitLab


From 49539856ad4b19de5e3270ee955d9d91eb5ba08c Mon Sep 17 00:00:00 2001
From: Jesse Hines <hinesjr@ornl.gov>
Date: Tue, 9 Sep 2025 11:09:24 -0400
Subject: [PATCH 21/21] Improve help strings

---
 raps/sim_config.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/raps/sim_config.py b/raps/sim_config.py
index 0cbe587..a73cd3e 100644
--- a/raps/sim_config.py
+++ b/raps/sim_config.py
@@ -358,6 +358,11 @@ class SimConfig(RAPSBaseModel, abc.ABC):
 
 class SingleSimConfig(SimConfig, abc.ABC):
     system: SystemConfig | str = "frontier"
+    """
+    Name of the system to simulate, e.g "frontier". Can also be a path to a yaml file containing
+    the SystemConfig. You can also make modificiations to the SystemConfig on the CLI using
+    `--system.base`, e.g.  `--system.base frontier --system.cooling.fmu-path path/to/my.fmu`
+    """
 
     @property
     def system_name(self) -> str:
@@ -370,6 +375,10 @@ class SingleSimConfig(SimConfig, abc.ABC):
 
 class MultiPartSimConfig(SimConfig):
     partitions: list[SystemConfig | str]
+    """
+    List of multiple systems/partitions to run. Can be names of preconfigured systems, or paths
+    to custom SystemConfig yaml files.
+    """
 
     @property
     def system_name(self) -> str:
-- 
GitLab