Skip to content

Commit

Permalink
bug fix: hpctoolkit reader (#116)
Browse files Browse the repository at this point in the history 
---------

Co-authored-by: Abhinav Bhatele <bhatele@cs.umd.edu>
  • Loading branch information
@movsesyanae @bhatele
movsesyanae and bhatele authored Nov 14, 2023
1 parent b6b5f47 commit 0d847b0
Showing 1 changed file with 39 additions and 23 deletions.
62 changes: 39 additions & 23 deletions pipit/readers/hpctoolkit_reader.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -592,13 +592,13 @@ def __read_single_entry_point(self, entry_point_pointer: int) -> None:
# map context for this context
string_index = self.common_string_index_map[pretty_name_pointer]
context = {
"relation": -1,
"lexical_type": -1,
"function_index": -1,
"source_file_index": -1,
"source_file_line": -1,
"load_module_index": -1,
"load_module_offset": -1,
"relation": None,
"lexical_type": None,
"function_index": None,
"source_file_index": None,
"source_file_line": None,
"load_module_index": None,
"load_module_offset": None,
"string_index": string_index,
}
# context = {"string_index": string_index}
Expand Down Expand Up @@ -710,7 +710,7 @@ def __read_children_contexts(
sub_flex_2 = int.from_bytes(
flex[8:10], byteorder=self.byte_order, signed=self.signed
)
flex = flex[10:]
flex = flex[16:]
source_file_index = self.__get_source_file_index(sub_flex_1)
source_file_line = sub_flex_2

Expand Down Expand Up @@ -929,7 +929,6 @@ def __read_hit_section(self, section_pointer: int, section_size: int) -> None:
# hit pointer
hit_pointer = self.file.tell()

# print((self.file.tell() - section_pointer))
# Number of identifications in this tuple (u16)
num_tuples = int.from_bytes(
self.file.read(2), byteorder=self.byte_order, signed=self.signed
Expand All @@ -940,7 +939,7 @@ def __read_hit_section(self, section_pointer: int, section_size: int) -> None:

# Identifications for an application thread
# Read H.I.T.s
tuples_list = []
tuples_map = {}
for i in range(num_tuples):
# One of the values listed in the profile.db
# Identifier Names section. (u8)
Expand All @@ -964,8 +963,21 @@ def __read_hit_section(self, section_pointer: int, section_size: int) -> None:
self.file.read(8), byteorder=self.byte_order, signed=self.signed
)
identifier_name = self.meta_reader.get_identifier_name(kind)
tuples_list.append((identifier_name, physical_id))
self.hit_map[hit_pointer] = tuples_list
tuples_map[identifier_name] = physical_id

self.hit_map[hit_pointer] = self.__clean_hit(tuples_map)

def __clean_hit(self, tuples_map: dict) -> dict:
# add None for information not present
if "Node" not in tuples_map:
tuples_map["Node"] = None
if "RANK" not in tuples_map:
tuples_map["RANK"] = None
if "THREAD" not in tuples_map:
tuples_map["THREAD"] = None
if "CORE" not in tuples_map:
tuples_map["CORE"] = None
return tuples_map

def __read_common_header(self) -> None:
"""
Expand Down Expand Up @@ -1144,6 +1156,7 @@ def __read_trace_headers_section(
"Name": [],
"Thread": [],
"Process": [],
"Core": [],
"Host": [],
"Node": [],
"Source File Name": [],
Expand Down Expand Up @@ -1242,9 +1255,10 @@ def __read_single_trace_header(self, header_pointer: int) -> None:
else:
self.data["Event Type"].append("Leave")
self.data["Timestamp (ns)"].append(timestamp)
self.data["Process"].append(hit[1][1])
self.data["Thread"].append(hit[2][1])
self.data["Host"].append(hit[0][1])
self.data["Process"].append(hit["RANK"])
self.data["Thread"].append(hit["THREAD"])
self.data["Host"].append(hit["NODE"])
self.data["Core"].append(hit["CORE"])
self.data["Node"].append(last_node)
self.data["Source File Name"].append(context_information["file"])
self.data["Source File Line Number"].append(
Expand Down Expand Up @@ -1276,9 +1290,10 @@ def __read_single_trace_header(self, header_pointer: int) -> None:
else:
self.data["Event Type"].append("Enter")
self.data["Timestamp (ns)"].append(timestamp)
self.data["Process"].append(hit[1][1])
self.data["Thread"].append(hit[2][1])
self.data["Host"].append(hit[0][1])
self.data["Process"].append(hit["RANK"])
self.data["Thread"].append(hit["THREAD"])
self.data["Host"].append(hit["NODE"])
self.data["Core"].append(hit["CORE"])
self.data["Node"].append(entry_node)
self.data["Source File Name"].append(context_information["file"])
self.data["Source File Line Number"].append(
Expand Down Expand Up @@ -1311,9 +1326,10 @@ def __read_single_trace_header(self, header_pointer: int) -> None:
else:
self.data["Event Type"].append("Leave")
self.data["Timestamp (ns)"].append(timestamp)
self.data["Process"].append(hit[1][1])
self.data["Thread"].append(hit[2][1])
self.data["Host"].append(hit[0][1])
self.data["Process"].append(hit["RANK"])
self.data["Thread"].append(hit["THREAD"])
self.data["Host"].append(hit["NODE"])
self.data["Core"].append(hit["CORE"])
self.data["Node"].append(last_node)
self.data["Source File Name"].append(context_information["file"])
self.data["Source File Line Number"].append(context_information["line"])
Expand Down Expand Up @@ -1358,5 +1374,5 @@ def read(self) -> pipit.trace.Trace:

# cct is needed to create trace in hpctoolkit,
# so always return it as part of the trace
self.trace_df = trace_df
return pipit.trace.Trace(None, trace_df, self.meta_reader.cct)
self.trace_df = trace_df.dropna(axis=1, how="all")
return pipit.trace.Trace(None, self.trace_df, self.meta_reader.cct)

0 comments on commit 0d847b0

Please sign in to comment.