Pvelesko/rename executables to main by Geekdude · Pull Request #159 · ORNL/HeCBench

Geekdude · 2026-04-06T21:08:55Z

Summary

Standardize binary names across 189 Makefiles to produce an executable named main instead of various custom names
Custom names replaced: AMGMk, MD5Hash, SobolQRNG, rtm8, lulesh, kernel, b+tree.out, xsbench, XSBench, streamcluster, srad, myocyte, leukocyte, kmeans, hybridsort, heartwall, MaxwellsGPU3d, miniFE.x, vj-cpu, snicit, qtclustering, miniDGS, etc.
Updated all Makefile variants (.hipcl, .hipsycl, .aomp, .nvc)
Updated run:, clean:, and target rules to reference main

Motivation

Most benchmarks already use main as the executable name, but ~60 benchmarks had custom names. This makes automation harder since scripts need to know each benchmark's binary name. With this change, all benchmarks consistently produce ./main, simplifying tooling like autohecbench.py (which already defaults to binary = "main").

Test plan

Verify a sample of renamed benchmarks compile and run correctly
Confirm make run works for affected benchmarks

Ensures benchmarks return non-zero exit code when verification fails, allowing automated test harnesses to detect failures via process exit codes. Changes are purely additive (insertions only). Adds `if (condition) exit(1);` after PASS/FAIL output in 838 benchmark files covering all variants (CUDA, HIP, OpenMP, SYCL). Adds `#include <cstdlib>` where needed. Preserves original file encoding (ISO-8859) and line endings (CRLF).

Relocates the failure exit from immediately after the PASS/FAIL print to just before `return 0;`, ensuring host and device memory are properly released before the process exits. Changes `exit(1)` to `return 1` in 596 files where free/cudaFree/hipFree/sycl::free calls follow the verification output.

…ult check The prefetch() function was missing exit(1) on failure, and main() referenced testResult which is local to prefetch/naive functions. Both functions now exit(1) on failure directly.

Standardize binary names across all benchmarks to produce an executable named 'main' instead of various custom names (AMGMk, MD5Hash, SobolQRNG, rtm8, lulesh, kernel, b+tree.out, xsbench, XSBench, streamcluster, srad, myocyte, leukocyte, kmeans, hybridsort, heartwall, MaxwellsGPU3d, etc.). Changes include: - Updated program/EXE/BIN_NAME variables in Makefiles - Changed binary target names and -o flags - Updated run targets to reference 'main' - Updated clean targets to remove 'main' - Fixed all Makefile variants (.hipcl, .hipsycl, .aomp, .nvc)

replace vj-gpu with main

- face-omp (aomp, nvc), face-sycl: restore vj-cpu in program variable, rename vj-gpu target to main - logan-cuda/hip: revert Makefile changes (restore demo target, c++14) - lsqt-cuda/hip: add missing builddir target for obj_gpu directory - multimaterial-hip, -hip/hipcl, -sycl, -omp/aomp, -omp/nvc: restore multimat and multimat_F targets, rename multimat_FL to main - projectile-hip: remove accidentally committed Projectile binary - face-sycl: fix $(program) target override that broke vj-cpu recipe - multimaterial-sycl: add missing all: target so make builds everything

Copilot

Pull request overview

Standardizes benchmark build outputs by renaming executables to main across many Makefile variants and aligns runtime automation by ensuring benchmarks return non-zero exit codes on verification failure.

Changes:

Renamed numerous Makefile targets/program variables to produce a main executable and updated run/clean rules accordingly.
Updated many benchmarks to exit/return non-zero when validation fails (PASS/FAIL now maps to process exit status).
Added missing standard headers (e.g., <cstdlib>) in some benchmarks to support exit() usage.

Reviewed changes

Copilot reviewed 300 out of 1021 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
src/floydwarshall-sycl/main.cpp	Exits with non-zero on FAIL (but introduces unreachable code).
src/floydwarshall-omp/main.cpp	Exits with non-zero on FAIL (but introduces unreachable code).
src/floydwarshall-hip/main.cu	Exits with non-zero on FAIL (but introduces unreachable code).
src/floydwarshall-cuda/main.cu	Exits with non-zero on FAIL (but introduces unreachable code).
src/filter-sycl/main.cpp	Adds non-zero exit on mismatch (but duplicates the check).
src/dp-sycl/main.cpp	Adds non-zero exit on mismatch (but duplicates the check).
src/atomicPerf-sycl/main.cpp	Adds non-zero exit on failure (but duplicates the check many times).
src/accuracy-sycl/main.cpp	Introduces `ok` flag for exit status (but likely incorrect scoping/indentation).
src/face-sycl/Makefile	Renames GPU target to `main` but `run` target still invokes `vj-cpu`.
src/b+tree-hip/Makefile	Renames output to `main` but `clean` target doesn’t remove `main`.
(many additional Makefiles / benchmarks)	Similar executable renames and non-zero exit code updates across the suite.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-06T21:11:01Z

src/floydwarshall-sycl/main.cpp

  else
  {
    printf("FAIL\n");
+    exit(1);
    if (numNodes <= 8) 
    {


The newly added exit(1); makes the diagnostic printing block below unreachable. If the intent is to keep printing details on small graphs, move the non-zero exit/return to the end of the else block (after the optional dumps), or replace exit(1) with a flag and return 1 at the end of main.

Copilot · 2026-04-06T21:11:01Z

src/filter-sycl/main.cpp

  sycl::free(d_nres, q);

+  if (!match) return 1;
+  if (!match) return 1;


The failure check is duplicated. Remove one of the two identical if (!match) return 1; lines to avoid redundancy.

Suggested change

if (!match) return 1;

Copilot · 2026-04-06T21:11:01Z

src/dp-sycl/main.cpp

+  if (dst != dst_ref) return 1;
+  if (dst != dst_ref) return 1;


The same condition is repeated three times. Keep a single check (or use return dst != dst_ref;) to reduce noise and prevent copy/paste artifacts from spreading.

Suggested change

if (dst != dst_ref) return 1;

if (dst != dst_ref) return 1;

Copilot · 2026-04-06T21:11:02Z

src/atomicPerf-sycl/main.cpp

+  if (fail) return 1;
+  if (fail) return 1;
+  if (fail) return 1;
+  if (fail) return 1;
+  if (fail) return 1;


The if (fail) return 1; statement is duplicated multiple times. This should be a single check after fail is finalized; remove the extra lines.

Suggested change

if (fail) return 1;

if (fail) return 1;

if (fail) return 1;

if (fail) return 1;

if (fail) return 1;

Copilot · 2026-04-06T21:11:02Z

src/accuracy-sycl/main.cpp


    int count;
    q.memcpy(&count, d_count, sizeof(int)).wait();
+	bool ok = (count == count_ref);


The added ok declaration appears to be introduced with inconsistent indentation (tab) and may be scoped too narrowly (e.g., inside a loop/block) while being used later for the final exit status. Declare ok in the outer scope of main and assign to it (do not redeclare) so the final if (!ok) return 1; reliably reflects the verification result.

Copilot · 2026-04-06T21:11:02Z

src/accuracy-sycl/main.cpp

  free(label);
  free(data);

+  if (!ok) return 1;


The added ok declaration appears to be introduced with inconsistent indentation (tab) and may be scoped too narrowly (e.g., inside a loop/block) while being used later for the final exit status. Declare ok in the outer scope of main and assign to it (do not redeclare) so the final if (!ok) return 1; reliably reflects the verification result.

Copilot · 2026-04-06T21:11:02Z

src/face-sycl/Makefile

 #===============================================================================

-program = vj-cpu vj-gpu
+program = main vj-cpu


This Makefile introduces a main target but the run target still invokes only ./vj-cpu. Given the PR goal of standardizing on main and other face Makefile variants run the GPU version via ./main, consider updating run here to execute ./main for the GPU path (or add a separate run step for it) to keep behavior consistent.

Copilot · 2026-04-06T21:11:02Z

src/face-sycl/Makefile

 run: build
 	$(LAUNCHER) ./vj-cpu ../face-cuda/Face.pgm ../face-cuda/info.txt ../face-cuda/class.txt Output-cpu.pgm


This Makefile introduces a main target but the run target still invokes only ./vj-cpu. Given the PR goal of standardizing on main and other face Makefile variants run the GPU version via ./main, consider updating run here to execute ./main for the GPU path (or add a separate run step for it) to keep behavior consistent.

Copilot · 2026-04-06T21:11:03Z

src/b+tree-hip/Makefile

+	$(LAUNCHER) ./main file ../data/b+tree/mil.txt command ../data/b+tree/command.txt

 clean:
 	rm	-rf *.o *.out \


The build output is renamed to main, but clean still removes *.out and does not remove the main executable. Update the clean rule to delete main so make clean restores a clean workspace.

Suggested change

rm -rf *.o *.out \

rm -rf *.o *.out main \

pvelesko and others added 6 commits April 6, 2026 16:46

Fix prefetch-hip: add missing exit(1) and remove out-of-scope testRes…

d5bcf24

…ult check The prefetch() function was missing exit(1) on failure, and main() referenced testResult which is local to prefetch/naive functions. Both functions now exit(1) on failure directly.

Update Makefile.aomp

a13db7f

replace vj-gpu with main

Copilot AI review requested due to automatic review settings April 6, 2026 21:08

Copilot AI reviewed Apr 6, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Pvelesko/rename executables to main#159

Pvelesko/rename executables to main#159
Geekdude wants to merge 6 commits intomasterfrom
pvelesko/rename-executables-to-main

Geekdude commented Apr 6, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Copilot AI Apr 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		run: build
		$(LAUNCHER) ./vj-cpu ../face-cuda/Face.pgm ../face-cuda/info.txt ../face-cuda/class.txt Output-cpu.pgm

Conversation

Geekdude commented Apr 6, 2026

Summary

Motivation

Test plan

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants