File size: 1,472 Bytes
02c783d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
# Copyright(C) [2025] Advanced Micro Devices, Inc. All rights reserved.
import os
import torch
class Names:
GEN_FOLDER = "gen"
GEN_SUFFIX = "_gen_triton_code"
REF_SUFFIX = "_ref_triton_code"
RET_SEPERATOR = "*#*#"
PYTEST_SEPARATOR = "&"*100
GPU = torch.cuda.get_device_name(0).replace(" ", "_") if torch.cuda.is_available() else None
PASS_NUM = 'pass_num'
FILE_NAME = 'file_name'
CALL_STATUS = 'call_status'
EXEC_STATUS = 'exec_status'
STDOUT = 'stdout'
STDERR = 'stderr'
DIFFICULTY = 'difficulty'
PREDICT = 'predict'
FILE = 'file'
DIFFICULTY = 'difficulty'
LABEL = 'label'
SPEEDUP = 'speedup'
REPO_ROOT = os.path.abspath(os.path.dirname(__file__))
TMP_ROOT = "tmp2"
TBG_ROOT = os.path.join(REPO_ROOT, "data", "TritonBench")
TBG_DATA_ROOT= os.path.join(TBG_ROOT, "data", "TritonBench_G_v1")
TBG_PERF_GOLD_ROOT = os.path.join(TBG_ROOT, "performance_metrics", "perf_G", f"{Names.GPU}_golden_metrics")
NATIVE_PERF_GOLD_ROOT = os.path.join(TBG_ROOT, "performance_metrics", "perf_G", "golden_metrics")
TBG_PERF_GOLD_DATA_ROOT = os.path.join(TBG_ROOT, "performance_metrics", "perf_G", f"{Names.GPU}_golden_results")
ROCm_ROOT = os.path.join(REPO_ROOT, "data", "ROCm")
ROCm_DATA_ROOT= os.path.join(ROCm_ROOT, "data", "ROCm_v1")
ROCm_DATA_AUTOTUNE_ROOT= os.path.join(ROCm_ROOT, "data", "ROCm_v1_autotune")
ROCM_PERF_GOLD_DATA_ROOT = os.path.join(ROCm_ROOT, "data", "performance", "golden_results") |