Skip to content

Fix SCHEMATIC silently shipping uninstrumented binaries #71

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
byeongjee merged 10 commits into from
Aug 17, 2026
Merged

Fix SCHEMATIC silently shipping uninstrumented binaries #71

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
15823d1
Collect SCHEMATIC traces with the caller's device-debug setting
byeongjee Aug 16, 2026
f20941e
Make SCHEMATIC trace-loading failures fatal
byeongjee Aug 16, 2026
715dff5
Drop the mutatedIR out-parameter from SchematicPass
byeongjee Aug 16, 2026
a17b180
Commit the benchmark result to NVM in non-debug builds too
byeongjee Aug 16, 2026
42e42da
Anchor trace scaling on the 99th percentile
byeongjee Aug 16, 2026
42b42c8
Ignore bench_commit_result in assembly energy estimation
byeongjee Aug 17, 2026
0912542
Size dijkstra, qsort, and bitcount to fit a harvesting trace
byeongjee Aug 17, 2026
5d43907
Retry a bench run that fails its capture or reports a violation
byeongjee Aug 17, 2026
00880ca
Re-measure continuous-power results with the current energy parameters
byeongjee Aug 17, 2026
45c11f1
Use full benchmark sizes for measurement, shrink for verify/intermittent
byeongjee Aug 17, 2026
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
3 changes: 1 addition & 2 deletions benchmarks/assembly_params.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,10 +5,9 @@
"timing_gpio_stop",
"_timing_delay_cycles",
"bench_halt",
"bench_commit_done",
"debug_init",
"debug_exit",
"debug_exit_commit"
"bench_commit_result"
],
"parameters": {
"add_autoincrement_register": 0.8291945408090596,
Expand Down
10 changes: 5 additions & 5 deletions benchmarks/intermittent/bitcount.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,11 +14,11 @@
#define NUM_FUNCS 8U
#define RNG_SEED 0x0C728394u

/* VERIFY_BUILD (defined by `ckpt verify`) shrinks the workload: verification
* crosses the same static region boundaries regardless of iteration count,
* and BOR halt mode makes the full workload impractically slow. */
#ifdef VERIFY_BUILD
#define ITERATIONS 64U
/* The same static region boundaries are crossed regardless of iteration
* count, and 20000 iterations outlast a harvesting trace and are
* impractically slow under BOR halt mode. */
#if defined(VERIFY_BUILD) || defined(INTERMITTENT_BUILD)
#define ITERATIONS 128U
#else
#define ITERATIONS 20000U
#endif
Expand Down
10 changes: 5 additions & 5 deletions benchmarks/intermittent/dijkstra.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,11 +12,11 @@

#define NUM_NODES 32U

/* VERIFY_BUILD (defined by `ckpt verify`) shrinks the workload: the relaxation
* loop runs once per queue entry per node and BOR halt mode pays a reset per
* region boundary, so all 32 queries are impractically slow. Shrinking the
* query count keeps the graph, and with it every loop trip count, unchanged. */
#ifdef VERIFY_BUILD
/* The relaxation loop runs once per queue entry per node, so all 32 queries
* outlast a harvesting trace and are impractically slow under BOR halt mode.
* The query count keeps the graph, and with it every loop trip count,
* unchanged. */
#if defined(VERIFY_BUILD) || defined(INTERMITTENT_BUILD)
#define NUM_QUERIES 4U
#else
#define NUM_QUERIES NUM_NODES
Expand Down
9 changes: 4 additions & 5 deletions benchmarks/intermittent/qsort_input.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,11 +3,10 @@

#include <stdint.h>

/* VERIFY_BUILD (defined by `ckpt verify`) shrinks the workload: the insertion
* sort shifts O(ARRAY_SIZE^2) times and BOR halt mode pays a reset per region
* boundary, so the full array is impractically slow. The vertex table below is
* unsized, so only the loop bounds shrink. */
#ifdef VERIFY_BUILD
/* The insertion sort shifts O(ARRAY_SIZE^2) times, so the full 640-vertex
* array outlasts a harvesting trace and is impractically slow under BOR halt
* mode. The vertex table below is unsized, so only the loop bounds shrink. */
#if defined(VERIFY_BUILD) || defined(INTERMITTENT_BUILD)
#define ARRAY_SIZE 96
#else
#define ARRAY_SIZE 640
Expand Down
Loading
Loading