diff --git a/Makefile b/Makefile
index 264e606a..c060bb63 100644
--- a/Makefile
+++ b/Makefile
@@ -531,8 +531,8 @@ test-list list-tests: test-runner
 test-marks: SUMMARYFLAGS+=-Si
 test-marks: $(TEST_CSV)
 	$(strip ./scripts/csv.py $^ \
+		-Fi='min(i)' \
 		-bsuite \
-		-Fi='min(enumerate())' \
 		-fpassed=test_passed \
 		-fruntime=test_runtime \
 		$(SUMMARYFLAGS))
@@ -545,8 +545,8 @@ test-marks-csv: $(BUILDDIR)/lfs3.test.csv
 .PHONY: test-marks-diff
 test-marks-diff: $(TEST_CSV)
 	$(strip ./scripts/csv.py $^ \
+		-Fi='min(i)' \
 		-bsuite \
-		-Fi='min(enumerate())' \
 		-fpassed=test_passed \
 		-fruntime=test_runtime \
 		$(SUMMARYFLAGS) -d $(BUILDDIR)/lfs3.test.csv)
@@ -556,8 +556,8 @@ test-marks-diff: $(TEST_CSV)
 test-bottlenecks: SUMMARYFLAGS+=-Sruntime
 test-bottlenecks: $(TEST_CSV)
 	$(strip ./scripts/csv.py $^ \
+		-Fi='min(i)' \
 		-bcase \
-		-Fi='min(enumerate())' \
 		-fpassed=test_passed \
 		-fruntime=test_runtime \
 		$(SUMMARYFLAGS))
@@ -595,8 +595,8 @@ bench-marks: $(BENCH_CSV)
 	$(strip ./scripts/csv.py \
 		<(./scripts/csv.py $^ \
 			-Uprobe=stack,heap,usage \
+			-Fi='min(i)' \
 			-bprobe='%(case)s+%(probe)s' \
-			-Fi='min(enumerate())' \
 			-fn='max(n)' \
 			-ft='max(float(bench_simtime)/1.0e9)' \
 			-o-) \
@@ -617,15 +617,15 @@ bench-marks-diff: $(BENCH_CSV)
 	$(strip ./scripts/csv.py \
 		<(./scripts/csv.py $^ \
 			-Uprobe=stack,heap,usage \
+			-Fi='min(i)' \
 			-bprobe='%(case)s+%(probe)s' \
-			-Fi='min(enumerate())' \
 			-fn='max(n)' \
 			-ft='max(float(bench_simtime)/1.0e9)' \
 			-o-) \
 		-d <(./scripts/csv.py $(BUILDDIR)/lfs3.bench.csv \
 			-Uprobe=stack,heap,usage \
+			-Fi='min(i)' \
 			-bprobe='%(case)s+%(probe)s' \
-			-Fi='min(enumerate())' \
 			-fn='max(n)' \
 			-ft='max(float(bench_simtime)/1.0e9)' \
 			-o-) \
@@ -641,8 +641,8 @@ bench-bottlenecks: $(BENCH_CSV)
 	$(strip ./scripts/csv.py \
 		<(./scripts/csv.py $^ \
 			-Uprobe=stack,heap,usage \
+			-Fi='min(i)' \
 			-bprobe='%(case)s+%(probe)s' \
-			-Fi='min(enumerate())' \
 			-fn='max(n)' \
 			-ft='max(float(bench_simtime)/1.0e9)' \
 			-fruntime='max(bench_runtime)' \
@@ -660,9 +660,9 @@ bench-ops: SUMMARYFLAGS+=-Si
 bench-ops: $(BENCH_CSV)
 	$(strip ./scripts/csv.py $^ \
 		-Uprobe=stack,heap,usage \
+		-Fi='min(i)' \
 		-bprobe='%(case)s+%(probe)s' \
 		-Hprobe=bench+probe \
-		-Fi='min(enumerate())' \
 		-freads='bench_reads' \
 		-fprogs='bench_progs' \
 		-ferases='bench_erases' \
@@ -677,9 +677,9 @@ bench-widths: SUMMARYFLAGS+=-Si
 bench-widths: $(BENCH_CSV)
 	$(strip ./scripts/csv.py $^ \
 		-Uprobe=stack,heap,usage \
+		-Fi='min(i)' \
 		-bprobe='%(case)s+%(probe)s' \
 		-Hprobe=bench+probe \
-		-Fi='min(enumerate())' \
 		-freaded="avg(saturate(ffrac( \
 			float(bench_readed)/float(bench_reads), \
 			max(1, $$( \
@@ -707,20 +707,20 @@ bench-ram bench-usage: $(BENCH_CSV)
 	$(strip ./scripts/csv.py \
 		<(./scripts/csv.py $^ \
 			-Dprobe=stack \
+			-Fi='min(i)' \
 			-bcase \
-			-Fi='min(enumerate())' \
 			-fstack='max(bench_simtime)' \
 			-o-) \
 		<(./scripts/csv.py $^ \
 			-Dprobe=heap \
+			-Fi='min(i)' \
 			-bcase \
-			-Fi='min(enumerate())' \
 			-fheap='max(bench_simtime)' \
 			-o-) \
 		<(./scripts/csv.py $^ \
 			-Dprobe=usage \
+			-Fi='min(i)' \
 			-bcase \
-			-Fi='min(enumerate())' \
 			-fdisk='max(bench_simtime)' \
 			-o-) \
 		-bcase \
diff --git a/scripts/bench.py b/scripts/bench.py
index b9b89e5f..69b5a22c 100755
--- a/scripts/bench.py
+++ b/scripts/bench.py
@@ -1147,7 +1147,8 @@ class BenchFailure(Exception):
         self.assert_ = assert_
 
 
-def run_stage(name, runner, bench_ids, stdout_, trace_, output_, **args):
+def run_stage(offset, name, runner, bench_ids,
+        stdout_, trace_, output_, **args):
     # get expected suite/case/perm counts
     (case_suites,
             expected_suite_perms,
@@ -1301,6 +1302,9 @@ def run_stage(name, runner, bench_ids, stdout_, trace_, output_, **args):
                             # write measurements immediately, this allows
                             # analysis of partial results
                             output_.writerow({
+                                    'i': offset
+                                        + locals.start
+                                        + locals.seen_perms*locals.step,
                                     'suite': last_suite,
                                     'case': last_case,
                                     **last_defines,
@@ -1338,14 +1342,16 @@ def run_stage(name, runner, bench_ids, stdout_, trace_, output_, **args):
         nonlocal killed
         nonlocal locals
 
-        start = start or 0
-        step = step or 1
-        while start < total_perms:
+        locals.start = start or 0
+        locals.step = step or 1
+        while locals.start < total_perms:
             runner_ = find_runner(runner, main=main, **args)
             if args.get('isolate') or args.get('valgrind'):
-                runner_.append('--step=%s,%s,%s' % (start, start+step, step))
-            elif start != 0 or step != 1:
-                runner_.append('--step=%s,,%s' % (start, step))
+                runner_.append('--step=%s,%s,%s' % (
+                        locals.start, locals.start+locals.step, locals.step))
+            elif locals.start != 0 or locals.step != 1:
+                runner_.append('--step=%s,,%s' % (
+                        locals.start, locals.step))
 
             runner_.extend(bench_ids)
 
@@ -1354,7 +1360,7 @@ def run_stage(name, runner, bench_ids, stdout_, trace_, output_, **args):
                 locals.seen_perms = 0
                 run_runner(runner_)
                 assert locals.seen_perms > 0
-                start += locals.seen_perms*step
+                locals.start += locals.seen_perms*locals.step
 
             except BenchFailure as failure:
                 # race condition for multiple failures?
@@ -1371,7 +1377,7 @@ def run_stage(name, runner, bench_ids, stdout_, trace_, output_, **args):
                 if args.get('keep_going') and not killed:
                     # resume after failed bench
                     assert locals.seen_perms > 0
-                    start += locals.seen_perms*step
+                    locals.start += locals.seen_perms*locals.step
                     continue
                 else:
                     # stop other benches
@@ -1489,9 +1495,10 @@ def run(runner, bench_ids=[], **args):
     output = None
     if args.get('output'):
         output = BenchOutput(args['output'],
-                ['suite', 'case'],
+                ['i', 'suite', 'case'],
                 # defines go here
-                ['probe', 'n',
+                ['probe',
+                    'n',
                     'bench_reads',
                     'bench_progs',
                     'bench_erases',
@@ -1522,6 +1529,7 @@ def run(runner, bench_ids=[], **args):
                 erased_,
                 failures_,
                 killed) = run_stage(
+                    expected,
                     by or 'benches',
                     runner,
                     [by] if by is not None else [],
diff --git a/scripts/test.py b/scripts/test.py
index e403d739..e2c0a7f8 100755
--- a/scripts/test.py
+++ b/scripts/test.py
@@ -1134,7 +1134,8 @@ class TestFailure(Exception):
         self.stdout = stdout
         self.assert_ = assert_
 
-def run_stage(name, runner, test_ids, stdout_, trace_, output_, **args):
+def run_stage(offset, name, runner, test_ids,
+        stdout_, trace_, output_, **args):
     # get expected suite/case/perm counts
     (case_suites,
             expected_suite_perms,
@@ -1228,6 +1229,9 @@ def run_stage(name, runner, test_ids, stdout_, trace_, output_, **args):
                             defines = find_defines(
                                     runner, m.group('id'), **args)
                             output_.writerow({
+                                    'i': offset
+                                        + locals.start
+                                        + locals.seen_perms*locals.step,
                                     'suite': suite,
                                     'case': case,
                                     **defines,
@@ -1265,14 +1269,16 @@ def run_stage(name, runner, test_ids, stdout_, trace_, output_, **args):
         nonlocal killed
         nonlocal locals
 
-        start = start or 0
-        step = step or 1
-        while start < total_perms:
+        locals.start = start or 0
+        locals.step = step or 1
+        while locals.start < total_perms:
             runner_ = find_runner(runner, main=main, **args)
             if args.get('isolate') or args.get('valgrind'):
-                runner_.append('--step=%s,%s,%s' % (start, start+step, step))
-            elif start != 0 or step != 1:
-                runner_.append('--step=%s,,%s' % (start, step))
+                runner_.append('--step=%s,%s,%s' % (
+                        locals.start, locals.start+locals.step, locals.step))
+            elif locals.start != 0 or locals.step != 1:
+                runner_.append('--step=%s,,%s' % (
+                        locals.start, locals.step))
 
             runner_.extend(test_ids)
 
@@ -1281,7 +1287,7 @@ def run_stage(name, runner, test_ids, stdout_, trace_, output_, **args):
                 locals.seen_perms = 0
                 run_runner(runner_)
                 assert locals.seen_perms > 0
-                start += locals.seen_perms*step
+                locals.start += locals.seen_perms*locals.step
 
             except TestFailure as failure:
                 # keep track of failures
@@ -1310,7 +1316,7 @@ def run_stage(name, runner, test_ids, stdout_, trace_, output_, **args):
                 if args.get('keep_going') and not killed:
                     # resume after failed test
                     assert locals.seen_perms > 0
-                    start += locals.seen_perms*step
+                    locals.start += locals.seen_perms*locals.step
                     continue
                 else:
                     # stop other tests
@@ -1428,7 +1434,7 @@ def run(runner, test_ids=[], **args):
     output = None
     if args.get('output'):
         output = TestOutput(args['output'],
-                ['suite', 'case'],
+                ['i', 'suite', 'case'],
                 # defines go here
                 ['test_passed', 'test_runtime'])
 
@@ -1449,6 +1455,7 @@ def run(runner, test_ids=[], **args):
                 powerlosses_,
                 failures_,
                 killed) = run_stage(
+                    expected,
                     by or 'tests',
                     runner,
                     [by] if by is not None else [],