[rocprof-compute] added f4f6 description and VALU FLOPS split for empirical peaks (#739)
* added f4f6 description and VALU FLOPS split * changed peak ammolite vars to local * reverted to dict peak initialization * ruff check format * updated VALU descriptions * updated VALU descriptions * Update parser.py * Update parser.py Added gracefull NameError handling Moved globals() update to init_metric_evaluation with ammolite__ vars and raw pmc_df * update formatting
This commit is contained in:
jamessiddeley-amd
committed by
GitHub
GitHub
parent
5201efe050
commit
f0955f5a83
@@ -771,7 +771,7 @@ def build_metric_value_string(dfs, dfs_type, normal_unit, profiling_config):
|
||||
|
||||
|
||||
def init_metric_evaluator(
|
||||
raw_pmc_df: Union[pd.DataFrame, dict], ammolite_vars: dict
|
||||
raw_pmc_df: Union[pd.DataFrame, dict], ammolite_vars: dict, empirical_peaks: dict
|
||||
) -> None:
|
||||
if isinstance(raw_pmc_df, dict):
|
||||
raw_pmc_df_keys = set(raw_pmc_df.keys())
|
||||
@@ -790,6 +790,7 @@ def init_metric_evaluator(
|
||||
# The process-local globals are used for performance optimization.
|
||||
globals().update(raw_pmc_df_items)
|
||||
globals().update(ammolite_vars)
|
||||
globals().update(empirical_peaks)
|
||||
|
||||
|
||||
def run_metric_evaluator(row_expr: str) -> str:
|
||||
@@ -821,6 +822,38 @@ def run_metric_evaluator(row_expr: str) -> str:
|
||||
console_error("analysis", str(ae))
|
||||
|
||||
|
||||
def create_empirical_peaks_dict(empirical_peaks_df):
|
||||
"""Create empirical peaks dictionary"""
|
||||
empirical_peaks = {}
|
||||
|
||||
if not empirical_peaks_df.empty:
|
||||
peak_data_row = empirical_peaks_df.iloc[0]
|
||||
for col in empirical_peaks_df.columns:
|
||||
empirical_peaks[f"ammolite__{col}_empirical_peak"] = peak_data_row[col]
|
||||
else:
|
||||
peak_names = [
|
||||
"FP16Flops",
|
||||
"FP32Flops",
|
||||
"FP64Flops",
|
||||
"MFMAF64Flops",
|
||||
"MFMAF32Flops",
|
||||
"MFMAF16Flops",
|
||||
"MFMABF16Flops",
|
||||
"MFMAF8Flops",
|
||||
"MFMAI8Ops",
|
||||
"HBMBw",
|
||||
"L2Bw",
|
||||
"L1Bw",
|
||||
"LDSBw",
|
||||
"MFMA_FLOPs_F6F4",
|
||||
]
|
||||
# initialize peaks to 0
|
||||
for peak_name in peak_names:
|
||||
empirical_peaks[f"ammolite__{peak_name}_empirical_peak"] = 0
|
||||
|
||||
return empirical_peaks
|
||||
|
||||
|
||||
@demarcate
|
||||
def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug, config):
|
||||
"""
|
||||
@@ -927,32 +960,10 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
"wave_size is not available in sysinfo.csv, please provide the correct "
|
||||
"value using --specs-correction"
|
||||
)
|
||||
if not empirical_peaks_df.empty:
|
||||
peak_data_row = empirical_peaks_df.iloc[0]
|
||||
for metric_name in empirical_peaks_df.columns:
|
||||
var_name = f"ammolite__{metric_name}_empirical_peak"
|
||||
locals()[var_name] = peak_data_row[metric_name]
|
||||
else:
|
||||
default_peaks = [
|
||||
"MFMAF64Flops",
|
||||
"MFMAF32Flops",
|
||||
"MFMAF16Flops",
|
||||
"MFMABF16Flops",
|
||||
"MFMAF8Flops",
|
||||
"MFMAI8Ops",
|
||||
"HBMBw",
|
||||
"L2Bw",
|
||||
"L1Bw",
|
||||
"LDSBw",
|
||||
"MFMA_FLOPs_F6F4",
|
||||
]
|
||||
# set values to 0 if no no empirical peaks from roofline.csv are provided
|
||||
for peak_name in default_peaks:
|
||||
var_name = f"ammolite__{peak_name}_empirical_peak"
|
||||
exec(f"{var_name} = 0", globals(), locals())
|
||||
|
||||
empirical_peaks = create_empirical_peaks_dict(empirical_peaks_df)
|
||||
|
||||
# TODO: fix all $normUnit in Unit column or title
|
||||
|
||||
# build and eval all derived build-in global variables
|
||||
ammolite__build_in = {}
|
||||
|
||||
@@ -966,6 +977,8 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
ammolite__build_in[key] = eval(compile(s, "<string>", "eval"))
|
||||
except TypeError:
|
||||
ammolite__build_in[key] = None
|
||||
except NameError:
|
||||
ammolite__build_in[key] = None
|
||||
except KeyError:
|
||||
ammolite__build_in[key] = None
|
||||
except AttributeError as ae:
|
||||
@@ -1022,12 +1035,32 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
)
|
||||
if matched_vars:
|
||||
for v in matched_vars:
|
||||
print(
|
||||
"Var ",
|
||||
v,
|
||||
":",
|
||||
eval(compile(v, "<string>", "eval")),
|
||||
)
|
||||
try:
|
||||
value = eval(
|
||||
compile(v, "<string>", "eval")
|
||||
)
|
||||
print("Var ", v, ":", value)
|
||||
except NameError:
|
||||
if "_empirical_peak" in v:
|
||||
if v in empirical_peaks:
|
||||
print(
|
||||
"Var ",
|
||||
v,
|
||||
":",
|
||||
empirical_peaks[v],
|
||||
)
|
||||
else:
|
||||
print(
|
||||
"Var ",
|
||||
v,
|
||||
": [empirical peak not found]", # noqa
|
||||
)
|
||||
else:
|
||||
print(
|
||||
"Var ",
|
||||
v,
|
||||
": [not available in main thread]", # noqa
|
||||
)
|
||||
matched_cols = re.findall(
|
||||
r"raw_pmc_df\['\w+'\]\['\w+'\]", row[expr]
|
||||
)
|
||||
@@ -1063,6 +1096,21 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
eval(compile(row[expr], "<string>", "eval"))
|
||||
)
|
||||
print("~" * 40)
|
||||
except NameError as ne:
|
||||
if "empirical_peak" in str(ne):
|
||||
console_warning(
|
||||
"Skipping debug evaluation. Empirical peak variables " # noqa
|
||||
"not available in main thread: {}".format( # noqa
|
||||
str(ne)
|
||||
)
|
||||
)
|
||||
else:
|
||||
console_warning(
|
||||
"Skipping debug evaluation. Variable not available: {}".format( # noqa
|
||||
str(ne)
|
||||
)
|
||||
)
|
||||
print("~" * 40)
|
||||
except TypeError:
|
||||
console_warning(
|
||||
"Skipping entry. Encountered a missing "
|
||||
@@ -1100,7 +1148,6 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
ammolite_vars = {
|
||||
key: val for key, val in locals().items() if key.startswith("ammolite__")
|
||||
}
|
||||
|
||||
# Empirically, 16 is about as much as we need.
|
||||
processes = min(16, multiprocessing.cpu_count() // 2)
|
||||
|
||||
@@ -1108,7 +1155,7 @@ def eval_metric(dfs, dfs_type, sys_info, empirical_peaks_df, raw_pmc_df, debug,
|
||||
with multiprocessing.Pool(
|
||||
processes=processes,
|
||||
initializer=init_metric_evaluator,
|
||||
initargs=(raw_pmc_df, ammolite_vars),
|
||||
initargs=(raw_pmc_df, ammolite_vars, empirical_peaks),
|
||||
) as pool:
|
||||
outs = pool.map(run_metric_evaluator, row_exprs)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user