Add PROBE to macro-benchmarks (#43)

charmoniumQ · Dec 7, 2024 · b13c3e9 · b13c3e9
1 parent b4b1d85
commit b13c3e9
Show file tree

Hide file tree

Showing 16 changed files with 4,961 additions and 335 deletions.
diff --git a/benchmark/experiment.py b/benchmark/experiment.py
@@ -137,27 +137,27 @@ def run_experiments(
     # Extracting a minimal set of fields
     # So we don't have all ExperimentStats in memory at once
     for i, (prov_collector, workload, stats) in enumerate(result_list):
-        counter = collections.Counter(
-            op.type for op in stats.operations
-        ) if stats is not None else None
-        record = {
-            "collector": prov_collector.name,
-            "collector_method": prov_collector.method,
-            "collector_submethod": prov_collector.submethod,
-            "workload": workload.name,
-            "workload_kind": workload.kind,
-            "cputime": stats.cputime,
-            "walltime": stats.walltime,
-            "memory": stats.memory,
-            "storage": stats.provenance_size,
-            "n_ops": len(stats.operations),
-            # "n_unique_files": n_unique(itertools.chain(
-            #     (op.target0 for op in stats.operations),
-            #     (op.target1 for op in stats.operations),
-            # )),
-            "op_type_counts": counter,
-        } if stats is not None else None
-        records.append(record)
+        if stats is not None:
+            counter = collections.Counter(
+                op.type for op in stats.operations
+            )
+            records[i] = {
+                "collector": prov_collector.name,
+                "collector_method": prov_collector.method,
+                "collector_submethod": prov_collector.submethod,
+                "workload": workload.name,
+                "workload_kind": workload.kind,
+                "cputime": stats.cputime,
+                "walltime": stats.walltime,
+                "memory": stats.memory,
+                "storage": stats.provenance_size,
+                "n_ops": len(stats.operations),
+                # "n_unique_files": n_unique(itertools.chain(
+                #     (op.target0 for op in stats.operations),
+                #     (op.target1 for op in stats.operations),
+                # )),
+                "op_type_counts": counter,
+            }
     results_df = (
         pandas.DataFrame.from_records(list(filter(bool, records)))
         .assign(**{

diff --git a/benchmark/flake.lock b/benchmark/flake.lock
diff --git a/benchmark/flake.nix b/benchmark/flake.nix
@@ -3,10 +3,13 @@
   # Use newer Nix to apply this bugfix
   # https://github.com/NixOS/nix/pull/10467
   inputs.new-nixos.url = "github:NixOS/nixpkgs";
+
+  inputs.probe.url = "../";
   outputs = {
     self,
     nixpkgs,
     flake-utils,
+    probe,
     ...
   } @ inputs:
     flake-utils.lib.eachDefaultSystem (
@@ -113,6 +116,7 @@
                 pypkgs.scikit-learn
                 pypkgs.nbconvert
                 jupyter-contrib-nbextensions
+                pypkgs.xgboost
               ]))
             ];
           };
@@ -220,7 +224,6 @@
               rev = "master";
               hash = "sha256-HFgqYEHanlwA0FA/7kOSsmcPzcb8BLJ3lG74DV5RtBA=";
             };
-            a = "hi";
             patches = [./splash-3.diff];
             nativeBuildInputs = [pkgs.m4 pkgs.binutils];
             sourceRoot = "source/codes";
@@ -786,6 +789,7 @@
             # proot provides tests with `make -C test` however they do not run in the sandbox
             doCheck = false;
           };
+          probe-pkg = probe.packages.${system}.probe-bundled;
           lftp = pkgs.lftp;
           wget = pkgs.wget;
           libseccomp = pkgs.libseccomp.lib;