Add arXiv citation and report the tolerances applied by the correctness checkers
Browse filesREADME and croissant.json: BibTeX for arXiv:2611.02752 (Croissant citeAs, version 1.0.1). index/benchmarks.parquet and index/golden_outputs.parquet: replace the catalog placeholders (threshold_raw, spec_atol) with the atol/rtol actually applied by the checkers plus tolerance_source. Input archives, baseline code, and inputs.parquet are unchanged.
- README.md +13 -4
- croissant.json +37 -36
- index/benchmarks.parquet +2 -2
- index/golden_outputs.parquet +2 -2
README.md
CHANGED
|
@@ -37,7 +37,7 @@ HPC-Bench evaluates large language models as optimizers of computational kernels
|
|
| 37 |
This dataset holds everything needed to run the benchmark that is not source code of the evaluation pipeline: the standardized inputs at five scales, the reference outputs used for correctness checking, the exact instrumented baseline programs, and a structured index that ties them together. The evaluation pipeline, the LLM-generated candidates studied in the paper, and the per-trial results live in the GitHub repository.
|
| 38 |
|
| 39 |
- Code and pipeline: https://github.com/Deep-Learning-Profiling-Tools/HPC-Bench
|
| 40 |
-
- Paper:
|
| 41 |
|
| 42 |
## Contents
|
| 43 |
|
|
@@ -130,7 +130,7 @@ One row per (setting, benchmark): 92 EX1, 92 EX2, 24 EX3.
|
|
| 130 |
| Baseline code | `baseline_code_path`, `baseline_code_bytes`, `baseline_code_sha256`, `baseline_code_files`, `baseline_kernel_files`, `source_repo`, `source_commit`, `source_path` |
|
| 131 |
| Inputs | `input_archive_path`, `input_archive_bytes`, `input_archive_sha256`, `input_sizes`, `n_input_sizes` |
|
| 132 |
| Reference outputs | `golden_output_sizes`, `n_golden_output_sizes`, `golden_output_bytes` |
|
| 133 |
-
| Correctness check | `checker`, `output_type`, `output_dtype`, `output_suffix`, `atol`, `rtol`, `
|
| 134 |
| Run configuration | `runtime_args_json` (command-line arguments per input size, from `configs/benchmark_args.json`) |
|
| 135 |
| Provenance | `upstream_suite`, `upstream_url`, `upstream_reference` |
|
| 136 |
|
|
@@ -149,7 +149,9 @@ A candidate is correct when the program's output, after the candidate kernel rep
|
|
| 149 |
| MachSuite | `machsuite` | `check.data` from the upstream suite | the program compares its own result against `check.data` and prints `Success`/`Fail`; the captured stdout is scanned |
|
| 150 |
| PARSEC (`canneal`) | `parsec` | text summary | the candidate's `Improvement:` value must be at least the reference value (absolute tolerance 1e-9, relative 1e-6) |
|
| 151 |
|
| 152 |
-
`atol`
|
|
|
|
|
|
|
| 153 |
|
| 154 |
Coverage: of the 1,016 (setting, benchmark, size) combinations present in the archives, 1,004 carry reference outputs. The exceptions are `lud` medium (EX1, EX2), `spmv` extra-large (EX1, EX2) and mini (EX2), `susan-c-mibench` medium (EX1, EX2), and all sizes of EX3 `srad_v2`; `haccmk` (EX1, EX2) and EX3 `nn` generate their data inside the program and have no per-size directories, and EX3 `backprop`, `bilateral`, and `gaussian` ship fewer than five scales. `index/inputs.parquet` records this per row.
|
| 155 |
|
|
@@ -189,7 +191,14 @@ The benchmark programs and input data are derived from PolyBench/C, Rodinia, Par
|
|
| 189 |
|
| 190 |
## Citation
|
| 191 |
|
| 192 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 193 |
|
| 194 |
## Croissant
|
| 195 |
|
|
|
|
| 37 |
This dataset holds everything needed to run the benchmark that is not source code of the evaluation pipeline: the standardized inputs at five scales, the reference outputs used for correctness checking, the exact instrumented baseline programs, and a structured index that ties them together. The evaluation pipeline, the LLM-generated candidates studied in the paper, and the per-trial results live in the GitHub repository.
|
| 38 |
|
| 39 |
- Code and pipeline: https://github.com/Deep-Learning-Profiling-Tools/HPC-Bench
|
| 40 |
+
- Paper: https://arxiv.org/abs/2611.02752
|
| 41 |
|
| 42 |
## Contents
|
| 43 |
|
|
|
|
| 130 |
| Baseline code | `baseline_code_path`, `baseline_code_bytes`, `baseline_code_sha256`, `baseline_code_files`, `baseline_kernel_files`, `source_repo`, `source_commit`, `source_path` |
|
| 131 |
| Inputs | `input_archive_path`, `input_archive_bytes`, `input_archive_sha256`, `input_sizes`, `n_input_sizes` |
|
| 132 |
| Reference outputs | `golden_output_sizes`, `n_golden_output_sizes`, `golden_output_bytes` |
|
| 133 |
+
| Correctness check | `checker`, `output_type`, `output_dtype`, `output_suffix`, `atol`, `rtol`, `tolerance_source`, `comparison_rule` |
|
| 134 |
| Run configuration | `runtime_args_json` (command-line arguments per input size, from `configs/benchmark_args.json`) |
|
| 135 |
| Provenance | `upstream_suite`, `upstream_url`, `upstream_reference` |
|
| 136 |
|
|
|
|
| 149 |
| MachSuite | `machsuite` | `check.data` from the upstream suite | the program compares its own result against `check.data` and prints `Success`/`Fail`; the captured stdout is scanned |
|
| 150 |
| PARSEC (`canneal`) | `parsec` | text summary | the candidate's `Improvement:` value must be at least the reference value (absolute tolerance 1e-9, relative 1e-6) |
|
| 151 |
|
| 152 |
+
`atol` and `rtol` are the tolerances the checker applies when correctness is checked with `scripts/arg.py benchmark check`, and `tolerance_source` says where they are defined: the PolyBench checker uses the per-benchmark threshold in `scripts/benchmark_catalog.py`; the Rodinia, Parboil, Codee, and MiBench checkers use the per-benchmark output specification in `scripts/checkers/output_specs.py`; the PARSEC checker uses constants in `scripts/checkers/parsec.py`. Integer outputs and images are compared exactly (`atol = rtol = 0`), and MachSuite rows carry no tolerance because the program checks itself. `comparison_rule` states the rule in words for every row.
|
| 153 |
+
|
| 154 |
+
Four CUDA-only benchmarks (`dwt2d`, `gaussian`, `huffman`, `hybridsort`) have reference outputs in their archives but no registered output specification, so the released checker framework records their checks as skipped and their candidates are not counted as correct in the released EX3 results; they are marked `checker = unregistered`, as is `nn`, which has no reference output.
|
| 155 |
|
| 156 |
Coverage: of the 1,016 (setting, benchmark, size) combinations present in the archives, 1,004 carry reference outputs. The exceptions are `lud` medium (EX1, EX2), `spmv` extra-large (EX1, EX2) and mini (EX2), `susan-c-mibench` medium (EX1, EX2), and all sizes of EX3 `srad_v2`; `haccmk` (EX1, EX2) and EX3 `nn` generate their data inside the program and have no per-size directories, and EX3 `backprop`, `bilateral`, and `gaussian` ship fewer than five scales. `index/inputs.parquet` records this per row.
|
| 157 |
|
|
|
|
| 191 |
|
| 192 |
## Citation
|
| 193 |
|
| 194 |
+
```bibtex
|
| 195 |
+
@article{cui2026hpc,
|
| 196 |
+
title={HPC-Bench: A Comprehensive Benchmark for High Performance Computing Codes},
|
| 197 |
+
author={Cui, Bowen and Yin, Junyu and Ramesh, Tejas and Lim, Leo and Hernandez, Oscar and Zhou, Keren},
|
| 198 |
+
journal={arXiv preprint arXiv:2611.02752},
|
| 199 |
+
year={2026}
|
| 200 |
+
}
|
| 201 |
+
```
|
| 202 |
|
| 203 |
## Croissant
|
| 204 |
|
croissant.json
CHANGED
|
@@ -82,9 +82,10 @@
|
|
| 82 |
"https://opensource.org/licenses/MIT",
|
| 83 |
"https://github.com/Deep-Learning-Profiling-Tools/HPC-Bench/blob/master/docs/BENCHMARKS.md"
|
| 84 |
],
|
| 85 |
-
"citeAs": "HPC-Bench: A Comprehensive Benchmark for High Performance Computing Codes
|
| 86 |
-
"version": "1.0.
|
| 87 |
"datePublished": "2026-10-31",
|
|
|
|
| 88 |
"keywords": [
|
| 89 |
"HPC",
|
| 90 |
"benchmark",
|
|
@@ -177,7 +178,7 @@
|
|
| 177 |
"description": "Structured index: benchmarks.",
|
| 178 |
"contentUrl": "https://huggingface.co/datasets/bcui2/HPC-Bench/resolve/main/index/benchmarks.parquet",
|
| 179 |
"encodingFormat": "application/x-parquet",
|
| 180 |
-
"sha256": "
|
| 181 |
},
|
| 182 |
{
|
| 183 |
"@type": "cr:FileObject",
|
|
@@ -195,7 +196,7 @@
|
|
| 195 |
"description": "Structured index: golden_outputs.",
|
| 196 |
"contentUrl": "https://huggingface.co/datasets/bcui2/HPC-Bench/resolve/main/index/golden_outputs.parquet",
|
| 197 |
"encodingFormat": "application/x-parquet",
|
| 198 |
-
"sha256": "
|
| 199 |
}
|
| 200 |
],
|
| 201 |
"recordSet": [
|
|
@@ -745,7 +746,7 @@
|
|
| 745 |
"@type": "cr:Field",
|
| 746 |
"@id": "benchmarks/checker",
|
| 747 |
"name": "checker",
|
| 748 |
-
"description": "Correctness checker used for the benchmark",
|
| 749 |
"dataType": "sc:Text",
|
| 750 |
"source": {
|
| 751 |
"fileObject": {
|
|
@@ -801,26 +802,11 @@
|
|
| 801 |
}
|
| 802 |
}
|
| 803 |
},
|
| 804 |
-
{
|
| 805 |
-
"@type": "cr:Field",
|
| 806 |
-
"@id": "benchmarks/rtol",
|
| 807 |
-
"name": "rtol",
|
| 808 |
-
"description": "Relative tolerance from the output specification",
|
| 809 |
-
"dataType": "sc:Float",
|
| 810 |
-
"source": {
|
| 811 |
-
"fileObject": {
|
| 812 |
-
"@id": "benchmarks-parquet"
|
| 813 |
-
},
|
| 814 |
-
"extract": {
|
| 815 |
-
"column": "rtol"
|
| 816 |
-
}
|
| 817 |
-
}
|
| 818 |
-
},
|
| 819 |
{
|
| 820 |
"@type": "cr:Field",
|
| 821 |
"@id": "benchmarks/atol",
|
| 822 |
"name": "atol",
|
| 823 |
-
"description": "Absolute tolerance
|
| 824 |
"dataType": "sc:Float",
|
| 825 |
"source": {
|
| 826 |
"fileObject": {
|
|
@@ -833,31 +819,31 @@
|
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"@type": "cr:Field",
|
| 836 |
-
"@id": "benchmarks/
|
| 837 |
-
"name": "
|
| 838 |
-
"description": "
|
| 839 |
"dataType": "sc:Float",
|
| 840 |
"source": {
|
| 841 |
"fileObject": {
|
| 842 |
"@id": "benchmarks-parquet"
|
| 843 |
},
|
| 844 |
"extract": {
|
| 845 |
-
"column": "
|
| 846 |
}
|
| 847 |
}
|
| 848 |
},
|
| 849 |
{
|
| 850 |
"@type": "cr:Field",
|
| 851 |
-
"@id": "benchmarks/
|
| 852 |
-
"name": "
|
| 853 |
-
"description": "
|
| 854 |
"dataType": "sc:Text",
|
| 855 |
"source": {
|
| 856 |
"fileObject": {
|
| 857 |
"@id": "benchmarks-parquet"
|
| 858 |
},
|
| 859 |
"extract": {
|
| 860 |
-
"column": "
|
| 861 |
}
|
| 862 |
}
|
| 863 |
},
|
|
@@ -1282,7 +1268,7 @@
|
|
| 1282 |
"@type": "cr:Field",
|
| 1283 |
"@id": "golden_outputs/checker",
|
| 1284 |
"name": "checker",
|
| 1285 |
-
"description": "Correctness checker used for the benchmark",
|
| 1286 |
"dataType": "sc:Text",
|
| 1287 |
"source": {
|
| 1288 |
"fileObject": {
|
|
@@ -1323,11 +1309,26 @@
|
|
| 1323 |
}
|
| 1324 |
}
|
| 1325 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1326 |
{
|
| 1327 |
"@type": "cr:Field",
|
| 1328 |
"@id": "golden_outputs/rtol",
|
| 1329 |
"name": "rtol",
|
| 1330 |
-
"description": "Relative tolerance
|
| 1331 |
"dataType": "sc:Float",
|
| 1332 |
"source": {
|
| 1333 |
"fileObject": {
|
|
@@ -1340,16 +1341,16 @@
|
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
"@type": "cr:Field",
|
| 1343 |
-
"@id": "golden_outputs/
|
| 1344 |
-
"name": "
|
| 1345 |
-
"description": "
|
| 1346 |
-
"dataType": "sc:
|
| 1347 |
"source": {
|
| 1348 |
"fileObject": {
|
| 1349 |
"@id": "golden_outputs-parquet"
|
| 1350 |
},
|
| 1351 |
"extract": {
|
| 1352 |
-
"column": "
|
| 1353 |
}
|
| 1354 |
}
|
| 1355 |
},
|
|
|
|
| 82 |
"https://opensource.org/licenses/MIT",
|
| 83 |
"https://github.com/Deep-Learning-Profiling-Tools/HPC-Bench/blob/master/docs/BENCHMARKS.md"
|
| 84 |
],
|
| 85 |
+
"citeAs": "@article{cui2026hpc,\n title={HPC-Bench: A Comprehensive Benchmark for High Performance Computing Codes},\n author={Cui, Bowen and Yin, Junyu and Ramesh, Tejas and Lim, Leo and Hernandez, Oscar and Zhou, Keren},\n journal={arXiv preprint arXiv:2611.02752},\n year={2026}\n}",
|
| 86 |
+
"version": "1.0.1",
|
| 87 |
"datePublished": "2026-10-31",
|
| 88 |
+
"dateModified": "2026-11-06",
|
| 89 |
"keywords": [
|
| 90 |
"HPC",
|
| 91 |
"benchmark",
|
|
|
|
| 178 |
"description": "Structured index: benchmarks.",
|
| 179 |
"contentUrl": "https://huggingface.co/datasets/bcui2/HPC-Bench/resolve/main/index/benchmarks.parquet",
|
| 180 |
"encodingFormat": "application/x-parquet",
|
| 181 |
+
"sha256": "a7373d4e1c11f661b8f0f74b655d294af49609c9e28c6a94e6c2ea5308f5e464"
|
| 182 |
},
|
| 183 |
{
|
| 184 |
"@type": "cr:FileObject",
|
|
|
|
| 196 |
"description": "Structured index: golden_outputs.",
|
| 197 |
"contentUrl": "https://huggingface.co/datasets/bcui2/HPC-Bench/resolve/main/index/golden_outputs.parquet",
|
| 198 |
"encodingFormat": "application/x-parquet",
|
| 199 |
+
"sha256": "17aefc7bc8a3026cda55d817204f5ffdfce6f48f68e25f67bea6f14e1a8c19d3"
|
| 200 |
}
|
| 201 |
],
|
| 202 |
"recordSet": [
|
|
|
|
| 746 |
"@type": "cr:Field",
|
| 747 |
"@id": "benchmarks/checker",
|
| 748 |
"name": "checker",
|
| 749 |
+
"description": "Correctness checker used for the benchmark (unregistered: no output specification in the released checker framework)",
|
| 750 |
"dataType": "sc:Text",
|
| 751 |
"source": {
|
| 752 |
"fileObject": {
|
|
|
|
| 802 |
}
|
| 803 |
}
|
| 804 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 805 |
{
|
| 806 |
"@type": "cr:Field",
|
| 807 |
"@id": "benchmarks/atol",
|
| 808 |
"name": "atol",
|
| 809 |
+
"description": "Absolute tolerance applied by the correctness checker",
|
| 810 |
"dataType": "sc:Float",
|
| 811 |
"source": {
|
| 812 |
"fileObject": {
|
|
|
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"@type": "cr:Field",
|
| 822 |
+
"@id": "benchmarks/rtol",
|
| 823 |
+
"name": "rtol",
|
| 824 |
+
"description": "Relative tolerance applied by the correctness checker",
|
| 825 |
"dataType": "sc:Float",
|
| 826 |
"source": {
|
| 827 |
"fileObject": {
|
| 828 |
"@id": "benchmarks-parquet"
|
| 829 |
},
|
| 830 |
"extract": {
|
| 831 |
+
"column": "rtol"
|
| 832 |
}
|
| 833 |
}
|
| 834 |
},
|
| 835 |
{
|
| 836 |
"@type": "cr:Field",
|
| 837 |
+
"@id": "benchmarks/tolerance_source",
|
| 838 |
+
"name": "tolerance_source",
|
| 839 |
+
"description": "Where the applied tolerance is defined (benchmark catalog, output specification, checker constants, or program self-check)",
|
| 840 |
"dataType": "sc:Text",
|
| 841 |
"source": {
|
| 842 |
"fileObject": {
|
| 843 |
"@id": "benchmarks-parquet"
|
| 844 |
},
|
| 845 |
"extract": {
|
| 846 |
+
"column": "tolerance_source"
|
| 847 |
}
|
| 848 |
}
|
| 849 |
},
|
|
|
|
| 1268 |
"@type": "cr:Field",
|
| 1269 |
"@id": "golden_outputs/checker",
|
| 1270 |
"name": "checker",
|
| 1271 |
+
"description": "Correctness checker used for the benchmark (unregistered: no output specification in the released checker framework)",
|
| 1272 |
"dataType": "sc:Text",
|
| 1273 |
"source": {
|
| 1274 |
"fileObject": {
|
|
|
|
| 1309 |
}
|
| 1310 |
}
|
| 1311 |
},
|
| 1312 |
+
{
|
| 1313 |
+
"@type": "cr:Field",
|
| 1314 |
+
"@id": "golden_outputs/atol",
|
| 1315 |
+
"name": "atol",
|
| 1316 |
+
"description": "Absolute tolerance applied by the correctness checker",
|
| 1317 |
+
"dataType": "sc:Float",
|
| 1318 |
+
"source": {
|
| 1319 |
+
"fileObject": {
|
| 1320 |
+
"@id": "golden_outputs-parquet"
|
| 1321 |
+
},
|
| 1322 |
+
"extract": {
|
| 1323 |
+
"column": "atol"
|
| 1324 |
+
}
|
| 1325 |
+
}
|
| 1326 |
+
},
|
| 1327 |
{
|
| 1328 |
"@type": "cr:Field",
|
| 1329 |
"@id": "golden_outputs/rtol",
|
| 1330 |
"name": "rtol",
|
| 1331 |
+
"description": "Relative tolerance applied by the correctness checker",
|
| 1332 |
"dataType": "sc:Float",
|
| 1333 |
"source": {
|
| 1334 |
"fileObject": {
|
|
|
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"@type": "cr:Field",
|
| 1344 |
+
"@id": "golden_outputs/tolerance_source",
|
| 1345 |
+
"name": "tolerance_source",
|
| 1346 |
+
"description": "Where the applied tolerance is defined (benchmark catalog, output specification, checker constants, or program self-check)",
|
| 1347 |
+
"dataType": "sc:Text",
|
| 1348 |
"source": {
|
| 1349 |
"fileObject": {
|
| 1350 |
"@id": "golden_outputs-parquet"
|
| 1351 |
},
|
| 1352 |
"extract": {
|
| 1353 |
+
"column": "tolerance_source"
|
| 1354 |
}
|
| 1355 |
}
|
| 1356 |
},
|
index/benchmarks.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7373d4e1c11f661b8f0f74b655d294af49609c9e28c6a94e6c2ea5308f5e464
|
| 3 |
+
size 93598
|
index/golden_outputs.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17aefc7bc8a3026cda55d817204f5ffdfce6f48f68e25f67bea6f14e1a8c19d3
|
| 3 |
+
size 21090
|