Files
2026-07-13 12:40:42 +08:00

2015 lines
80 KiB
Python
Executable File

# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import collections
import re
from enum import Enum
from paddle.base.core import TracerEventType, TracerMemEventType
from paddle.utils.flops import flops
from .statistic_helper import (
intersection_ranges,
merge_ranges,
merge_self_ranges,
sum_ranges,
)
_AllTracerEventType = [
TracerEventType.Operator,
TracerEventType.Dataloader,
TracerEventType.ProfileStep,
TracerEventType.CudaRuntime,
TracerEventType.Kernel,
TracerEventType.Memcpy,
TracerEventType.Memset,
TracerEventType.UserDefined,
TracerEventType.OperatorInner,
TracerEventType.Forward,
TracerEventType.Backward,
TracerEventType.Optimization,
TracerEventType.Communication,
TracerEventType.PythonOp,
TracerEventType.PythonUserDefined,
]
_CommunicationOpName = ['allreduce', 'broadcast', 'rpc']
class SortedKeys(Enum):
r"""
SortedKeys is used to specify how to sort items when printing ``paddle.profiler.Profiler.summary`` table.
The meaning of each SortedKeys is as following
- **SortedKeys.CPUTotal** : Sorted by CPU total time.
- **SortedKeys.CPUAvg** : Sorted by CPU average time.
- **SortedKeys.CPUMax** : Sorted by CPU max time.
- **SortedKeys.CPUMin** : Sorted by CPU min time.
- **SortedKeys.GPUTotal** : Sorted by GPU total time.
- **SortedKeys.GPUAvg** : Sorted by GPU average time.
- **SortedKeys.GPUMax** : Sorted by GPU max time.
- **SortedKeys.GPUMin** : Sorted by GPU min time.
"""
CPUTotal = 0
CPUAvg = 1
CPUMax = 2
CPUMin = 3
GPUTotal = 4
GPUAvg = 5
GPUMax = 6
GPUMin = 7
def _nodename2opname(name):
r'''
convert static host node name to operator name
'''
op_name = name.replace(' compute', '')
op_name = op_name.replace(' dygraph', '')
op_name = op_name.replace(' pybind_imperative_func', '')
return op_name
class HostStatisticNode:
r'''
Wrap original node for calculating statistic metrics.
'''
def __init__(self, hostnode):
self.hostnode = hostnode
self.children_node = []
self.runtime_node = []
self.cpu_time = 0
self.self_cpu_time = 0
self.gpu_time = 0 # kernel time
self.self_gpu_time = 0
self.general_gpu_time = 0 # besides kernel, include time of gpu events like memcpy and memset
self.self_general_gpu_time = 0
self.flops = 0
def cal_flops(self):
if self.hostnode.type == TracerEventType.Operator:
if hasattr(self.hostnode, 'input_shapes'):
op_name = _nodename2opname(self.hostnode.name)
self.flops = flops(
op_name,
self.hostnode.input_shapes,
self.hostnode.attributes,
)
def cal_statistic(self):
self.cpu_time = self.hostnode.end_ns - self.hostnode.start_ns
self.self_cpu_time = self.cpu_time
self.cal_flops()
for child in self.children_node:
child.cal_flops()
child.cal_statistic()
self.gpu_time += child.gpu_time
self.general_gpu_time += child.general_gpu_time
self.self_cpu_time -= child.end_ns - child.start_ns
self.flops += child.flops
for rt in self.runtime_node:
rt.cal_statistic()
self.self_cpu_time -= rt.end_ns - rt.start_ns
self.gpu_time += rt.gpu_time
self.self_gpu_time += rt.gpu_time
self.general_gpu_time += rt.general_gpu_time
self.self_general_gpu_time += rt.general_gpu_time
for device in self.hostnode.device_node:
if device.type == TracerEventType.Kernel:
self.gpu_time += device.end_ns - device.start_ns
self.self_gpu_time += device.end_ns - device.start_ns
self.general_gpu_time += device.end_ns - device.start_ns
self.self_general_gpu_time += device.end_ns - device.start_ns
@property
def end_ns(self):
return self.hostnode.end_ns
@property
def start_ns(self):
return self.hostnode.start_ns
def __getattr__(self, name):
return getattr(self.hostnode, name)
def traverse_tree(nodetrees):
results = collections.defaultdict(list)
for thread_id, rootnode in nodetrees.items():
stack = []
stack.append(rootnode)
threadlist = results[thread_id]
while stack:
current_node = stack.pop()
threadlist.append(current_node)
for childnode in current_node.children_node:
stack.append(childnode)
return results
def get_device_nodes(hostnode):
'''
Get all device nodes called in the time range of hostnode.
'''
stack = []
device_nodes = []
stack.append(hostnode)
while stack:
current_node = stack.pop()
for childnode in current_node.children_node:
stack.append(childnode)
for runtimenode in current_node.runtime_node:
for devicenode in runtimenode.device_node:
device_nodes.append(devicenode)
return device_nodes
def _build_layer_from_tree(nodetrees):
def build_layer(node, depth=0):
if "GradNode" in node.name:
return [], 0
if node.type in [
TracerEventType.Backward,
TracerEventType.Optimization,
]:
return [], 0
if node.type == TracerEventType.Operator:
stat_node = HostStatisticNode(node)
stat_node.cal_statistic()
return stat_node, stat_node.flops
layer = []
nflops = 0
for c in node.children_node:
l, f = build_layer(c, depth + 1)
if l:
nflops += f
layer.append(l)
if node.type == TracerEventType.Forward:
stat_node = HostStatisticNode(node)
stat_node.cal_statistic()
stat_node.flops = nflops
return [stat_node, layer], nflops
return layer, nflops
ret = []
for _, rootnode in nodetrees.items():
layer, _ = build_layer(rootnode)
ret.append(layer)
return ret
def _format_large_number(n, precision=2):
if n // 1e12 > 0:
return f"{round(n / 1e12, precision)} T"
if n // 1e9 > 0:
return f"{round(n / 1e9, precision)} G"
if n // 1e6 > 0:
return f"{round(n / 1e6, precision)} M"
if n // 1e3 > 0:
return f"{round(n / 1e3, precision)} K"
return f"{round(n, precision)}"
def _format_time(n, precision=2):
if n // 1e9 > 0:
return f"{round(n / 1e9, precision)} s"
if n // 1e6 > 0:
return f"{round(n / 1e6, precision)} ms"
if n // 1e3 > 0:
return f"{round(n / 1e3, precision)} us"
return f"{round(n, precision)} ns"
def _gen_layer_flops(node, repeat=1):
ret = []
offset = []
loop = []
def print_layer_tree(node, depth=0):
if isinstance(node, list):
for n in node:
print_layer_tree(n, depth + 1)
elif node.type in [TracerEventType.Forward, TracerEventType.Operator]:
if len(offset) == 0:
offset.append(depth)
name = _nodename2opname(node.name)
if (
depth == offset[-1] and len(ret) > 0 and ret[0].startswith(name)
): # repeat begin
loop.append(1)
if len(loop) >= repeat:
return "".join(ret)
align = " " * (depth - offset[-1])
tm = _format_time(node.cpu_time)
flops_n = _format_large_number(node.flops)
flops_s = _format_large_number(node.flops * 1e9 / node.cpu_time)
ret.append(
f"{align}{name} latency: {tm}, FLOPs: {flops_n}, FLOPS: {flops_s}\n"
)
for n in node[1:]:
print_layer_tree(n)
return "".join(ret)
def gen_layer_flops(nodetrees, repeat=1):
r'''
gen_layer_flops generate flops/runtime information depend on layer/operator.
'''
layer_tree = _build_layer_from_tree(nodetrees)
return _gen_layer_flops(layer_tree, repeat)
def wrap_tree(nodetrees):
'''
Using HostStatisticNode to wrap original profiler result tree, and calculate node statistic metrics.
'''
node_statistic_tree = {}
results = collections.defaultdict(list)
newresults = collections.defaultdict(list)
for thread_id, rootnode in nodetrees.items():
stack = []
stack.append(rootnode)
root_statistic_node = HostStatisticNode(rootnode)
newstack = []
newstack.append(root_statistic_node)
node_statistic_tree[thread_id] = root_statistic_node
threadlist = results[thread_id]
newthreadlist = newresults[thread_id]
while stack:
current_node = stack.pop()
threadlist.append(current_node)
current_statistic_node = newstack.pop()
newthreadlist.append(current_statistic_node)
for childnode in current_node.children_node:
stack.append(childnode)
child_statistic_node = HostStatisticNode(childnode)
current_statistic_node.children_node.append(
child_statistic_node
)
newstack.append(child_statistic_node)
for runtimenode in current_node.runtime_node:
runtime_statistic_node = HostStatisticNode(runtimenode)
current_statistic_node.runtime_node.append(
runtime_statistic_node
)
# recursive calculate node statistic values
for thread_id, root_statistic_node in node_statistic_tree.items():
root_statistic_node.cal_statistic()
return node_statistic_tree, newresults
class TimeRangeSummary:
r"""
Analyse time ranges for each TracerEventType, and summarize the time.
"""
def __init__(self):
self.CPUTimeRange = collections.defaultdict(list)
self.GPUTimeRange = collections.defaultdict(
lambda: collections.defaultdict(list)
) # GPU events should be divided into different devices
self.CPUTimeRangeSum = collections.defaultdict(int)
self.GPUTimeRangeSum = collections.defaultdict(
lambda: collections.defaultdict(int)
)
self.call_times = collections.defaultdict(int)
def parse(self, nodetrees):
r"""
Analysis node trees in profiler result, and get time range for different tracer event type.
"""
thread2hostnodes = traverse_tree(nodetrees)
for threadid, hostnodes in thread2hostnodes.items():
CPUTimeRange = collections.defaultdict(list)
GPUTimeRange = collections.defaultdict(
lambda: collections.defaultdict(
lambda: collections.defaultdict(list)
)
) # device_id/type/stream_id
for hostnode in hostnodes[1:]: # skip root node
CPUTimeRange[hostnode.type].append(
(hostnode.start_ns, hostnode.end_ns)
)
self.call_times[hostnode.type] += 1
for runtimenode in hostnode.runtime_node:
CPUTimeRange[runtimenode.type].append(
(runtimenode.start_ns, runtimenode.end_ns)
)
self.call_times[runtimenode.type] += 1
for devicenode in runtimenode.device_node:
GPUTimeRange[devicenode.device_id][devicenode.type][
devicenode.stream_id
].append((devicenode.start_ns, devicenode.end_ns))
self.call_times[devicenode.type] += 1
for event_type, time_ranges in CPUTimeRange.items():
time_ranges = merge_self_ranges(time_ranges, is_sorted=False)
self.CPUTimeRange[event_type] = merge_ranges(
self.CPUTimeRange[event_type], time_ranges, is_sorted=True
)
for device_id, device_time_ranges in GPUTimeRange.items():
for event_type, event_time_ranges in device_time_ranges.items():
for stream_id, time_ranges in event_time_ranges.items():
time_ranges = merge_self_ranges(
time_ranges, is_sorted=False
)
self.GPUTimeRange[device_id][event_type] = merge_ranges(
self.GPUTimeRange[device_id][event_type],
time_ranges,
is_sorted=True,
)
for event_type, time_ranges in self.CPUTimeRange.items():
self.CPUTimeRangeSum[event_type] = sum_ranges(time_ranges)
for device_id, device_time_ranges in self.GPUTimeRange.items():
for event_type, time_ranges in device_time_ranges.items():
self.GPUTimeRangeSum[device_id][event_type] = sum_ranges(
time_ranges
)
def get_gpu_devices(self):
return self.GPUTimeRange.keys()
def get_gpu_range_sum(self, device_id, event_type):
return self.GPUTimeRangeSum[device_id][event_type]
def get_cpu_range_sum(self, event_type):
return self.CPUTimeRangeSum[event_type]
class DistributedSummary:
r"""
Analysis communication and computation time range, and their overlap.
The computation time is all kernel except kernels for communication like nccl.
"""
def __init__(self):
self.cpu_communication_range = []
self.gpu_communication_range = []
self.communication_range = []
self.computation_range = []
self.overlap_range = []
self.cpu_calls = 0
self.gpu_calls = 0
def parse(self, nodetrees):
'''
Collect all communication and computation time ranges.
'''
thread2hostnodes = traverse_tree(nodetrees)
for threadid, hostnodes in thread2hostnodes.items():
for hostnode in hostnodes[1:]: # skip root node
# case 1: TracerEventType is Communication
if hostnode.type == TracerEventType.Communication:
self.cpu_communication_range.append(
(hostnode.start_ns, hostnode.end_ns)
)
device_nodes = get_device_nodes(hostnode)
for device_node in device_nodes:
if device_node.type == TracerEventType.Kernel:
self.gpu_communication_range.append(
(device_node.start_ns, device_node.end_ns)
)
# case 2: TracerEventType is Operator but is communication op
elif hostnode.type == TracerEventType.Operator and any(
name in hostnode.name.lower()
for name in _CommunicationOpName
):
self.cpu_communication_range.append(
(hostnode.start_ns, hostnode.end_ns)
)
device_nodes = get_device_nodes(hostnode)
for device_node in device_nodes:
if device_node.type == TracerEventType.Kernel:
self.gpu_communication_range.append(
(device_node.start_ns, device_node.end_ns)
)
# case 3: Others, filter kernels named with nccl
else:
for runtimenode in hostnode.runtime_node:
for devicenode in runtimenode.device_node:
if devicenode.type == TracerEventType.Kernel:
kernel_name = devicenode.name.lower()
if (
'nccl' in kernel_name
or 'xccl' in kernel_name
):
self.gpu_communication_range.append(
(devicenode.start_ns, devicenode.end_ns)
)
else:
self.computation_range.append(
(devicenode.start_ns, devicenode.end_ns)
)
self.cpu_calls = len(set(self.cpu_communication_range))
self.gpu_calls = len(set(self.gpu_communication_range))
self.cpu_communication_range = merge_self_ranges(
self.cpu_communication_range, is_sorted=False
)
self.gpu_communication_range = merge_self_ranges(
self.gpu_communication_range, is_sorted=False
)
self.communication_range = merge_ranges(
self.cpu_communication_range,
self.gpu_communication_range,
is_sorted=True,
)
self.computation_range = merge_self_ranges(
self.computation_range, is_sorted=False
)
self.overlap_range = intersection_ranges(
self.communication_range, self.computation_range, is_sorted=True
)
class EventSummary:
r"""
Analyse operator event in profiling data, correlate with its device event.
"""
class ItemBase:
def __init__(self, name):
self.name = name
self.call = 0
self.cpu_time = 0
self.gpu_time = 0
self.max_cpu_time = 0
self.min_cpu_time = float('inf')
self.max_gpu_time = 0
self.min_gpu_time = float('inf')
self.devices = {}
self.operator_inners = {}
self.general_gpu_time = 0
self.min_general_gpu_time = float('inf')
self.max_general_gpu_time = 0
self._flops = 0
@property
def flops(self):
return self._flops
@property
def avg_cpu_time(self):
return self.cpu_time / self.call
@property
def avg_gpu_time(self):
return self.gpu_time / self.call
@property
def avg_general_gpu_time(self):
return self.general_gpu_time / self.call
def add_cpu_time(self, time):
if time > self.max_cpu_time:
self.max_cpu_time = time
if time < self.min_cpu_time:
self.min_cpu_time = time
self.cpu_time += time
def add_gpu_time(self, time):
if time > self.max_gpu_time:
self.max_gpu_time = time
if time < self.min_gpu_time:
self.min_gpu_time = time
self.gpu_time += time
def add_general_gpu_time(self, time):
if time > self.max_general_gpu_time:
self.max_general_gpu_time = time
if time < self.min_general_gpu_time:
self.min_general_gpu_time = time
self.general_gpu_time += time
def add_call(self):
self.call += 1
def add_flops(self, flops):
self._flops += flops
def add_item(self, node):
raise NotImplementedError
class DeviceItem(ItemBase):
def add_item(self, node):
self.call += 1
self.add_gpu_time(node.end_ns - node.start_ns)
class OperatorItem(ItemBase):
def add_item(self, node):
self.add_call()
self.add_cpu_time(node.cpu_time)
self.add_gpu_time(node.gpu_time)
self.add_general_gpu_time(node.general_gpu_time)
self.add_flops(node.flops)
for child in node.children_node:
if child.type != TracerEventType.Operator:
if child.name not in self.operator_inners:
self.operator_inners[child.name] = (
EventSummary.OperatorItem(child.name)
)
self.operator_inners[child.name].add_item(child)
for runtimenode in node.runtime_node:
for devicenode in runtimenode.device_node:
name = devicenode.name
if name not in self.devices:
self.devices[name] = EventSummary.DeviceItem(name)
self.devices[name].add_item(devicenode)
class ForwardItem(ItemBase):
def add_item(self, node):
self.add_call()
self.add_cpu_time(node.cpu_time)
self.add_gpu_time(node.gpu_time)
self.add_general_gpu_time(node.general_gpu_time)
self.add_flops(node.flops)
for child in node.children_node:
if child.type != TracerEventType.Operator:
if child.name not in self.operator_inners:
self.operator_inners[child.name] = (
EventSummary.OperatorItem(child.name)
)
self.operator_inners[child.name].add_item(child)
class GeneralItem(ItemBase):
def add_item(self, node):
self.add_call()
self.add_cpu_time(node.cpu_time)
self.add_gpu_time(node.gpu_time)
self.add_general_gpu_time(node.general_gpu_time)
def __init__(self):
self.items = {} # for operator summary
self.thread_items = collections.defaultdict(
dict
) # for operator summary
self.userdefined_items = {} # for userdefined summary
self.userdefined_thread_items = collections.defaultdict(
dict
) # for userdefined summary
self.model_perspective_items = {} # for model summary
self.memory_manipulation_items = {} # for memory manipulation summary
self.kernel_items = {} # for kernel summary
def parse(self, nodetrees):
r"""
Analysis operator event in the nodetress.
"""
node_statistic_trees, thread2host_statistic_nodes = wrap_tree(nodetrees)
for (
threadid,
host_statistic_nodes,
) in thread2host_statistic_nodes.items():
for host_statistic_node in host_statistic_nodes[
1:
]: # skip root node
if host_statistic_node.type == TracerEventType.Operator:
self.add_operator_item(host_statistic_node)
if (
host_statistic_node.type == TracerEventType.UserDefined
or host_statistic_node.type
== TracerEventType.PythonUserDefined
):
if (
'memcpy' in host_statistic_node.name.lower()
or 'memorycopy' in host_statistic_node.name.lower()
or 'memset' in host_statistic_node.name.lower()
):
self.add_memory_manipulation_item(host_statistic_node)
else:
if (
host_statistic_node.type
== TracerEventType.PythonUserDefined
):
self.add_userdefined_item(host_statistic_node)
self.add_kernel_item(host_statistic_nodes[0])
for threadid, root_statistic_node in node_statistic_trees.items():
deque = collections.deque()
deque.append(root_statistic_node)
while deque:
current_node = deque.popleft()
for child in current_node.children_node:
if (
child.type == TracerEventType.Forward
or child.type == TracerEventType.Dataloader
or child.type == TracerEventType.Backward
or child.type == TracerEventType.Optimization
):
self.add_model_perspective_item(
child
) # find first model perspective node
else:
if child.type == TracerEventType.ProfileStep:
self.add_model_perspective_item(child)
deque.append(child)
def add_forward_item(self, operator_node):
pass
def add_operator_item(self, operator_node):
if operator_node.name not in self.items:
self.items[operator_node.name] = EventSummary.OperatorItem(
operator_node.name
)
self.items[operator_node.name].add_item(operator_node)
if operator_node.name not in self.thread_items[operator_node.thread_id]:
self.thread_items[operator_node.thread_id][operator_node.name] = (
EventSummary.OperatorItem(operator_node.name)
)
self.thread_items[operator_node.thread_id][operator_node.name].add_item(
operator_node
)
def add_userdefined_item(self, userdefined_node):
if userdefined_node.name not in self.userdefined_items:
self.userdefined_items[userdefined_node.name] = (
EventSummary.GeneralItem(userdefined_node.name)
)
self.userdefined_items[userdefined_node.name].add_item(userdefined_node)
if (
userdefined_node.name
not in self.userdefined_thread_items[userdefined_node.thread_id]
):
self.userdefined_thread_items[userdefined_node.thread_id][
userdefined_node.name
] = EventSummary.GeneralItem(userdefined_node.name)
self.userdefined_thread_items[userdefined_node.thread_id][
userdefined_node.name
].add_item(userdefined_node)
def add_memory_manipulation_item(self, memory_manipulation_node):
if memory_manipulation_node.name not in self.memory_manipulation_items:
self.memory_manipulation_items[memory_manipulation_node.name] = (
EventSummary.GeneralItem(memory_manipulation_node.name)
)
self.memory_manipulation_items[memory_manipulation_node.name].add_item(
memory_manipulation_node
)
def add_model_perspective_item(self, model_perspective_node):
if model_perspective_node.type == TracerEventType.Forward:
name = 'Forward'
elif model_perspective_node.type == TracerEventType.Backward:
name = 'Backward'
elif model_perspective_node.type == TracerEventType.Optimization:
name = 'Optimization'
elif model_perspective_node.type == TracerEventType.Dataloader:
name = 'Dataloader'
elif model_perspective_node.type == TracerEventType.ProfileStep:
name = 'ProfileStep'
else:
return
if name not in self.model_perspective_items:
self.model_perspective_items[name] = EventSummary.GeneralItem(name)
self.model_perspective_items[name].add_item(model_perspective_node)
def add_kernel_item(self, root_node):
device_nodes = get_device_nodes(root_node)
for device_node in device_nodes:
if device_node.type == TracerEventType.Kernel:
name = device_node.name
if name not in self.kernel_items:
self.kernel_items[name] = EventSummary.DeviceItem(name)
self.kernel_items[name].add_item(device_node)
class MemorySummary:
r"""
Analyse memory events in profiling data.
"""
class MemoryItem:
def __init__(self, event_name, place, memory_type='Allocated'):
self.event_name = event_name
self.place = place
self.allocation_count = 0
self.free_count = 0
self.allocation_size = 0
self.free_size = 0
self.increase_size = 0
self.memory_type = memory_type
def add_memory_record(self, size, allocation_type):
if (
allocation_type == TracerMemEventType.Allocate
or allocation_type == TracerMemEventType.ReservedAllocate
):
self.allocation_count += 1
self.allocation_size += size
elif (
allocation_type == TracerMemEventType.Free
or allocation_type == TracerMemEventType.ReservedFree
):
self.free_count += 1
self.free_size -= size # size is sign(-) when free.
else:
print("No corresponding type.")
self.increase_size = self.allocation_size - self.free_size
def __init__(self):
self.allocated_items = collections.defaultdict(
dict
) # for memory summary, device type: event
self.reserved_items = collections.defaultdict(
dict
) # for memory summary, device type: event
self.peak_allocation_values = collections.defaultdict(int)
self.peak_reserved_values = collections.defaultdict(int)
def _analyse_node_memory(self, event_name, node):
for memnode in node.mem_node: # self mem node
if (
memnode.type == TracerMemEventType.Allocate
or memnode.type == TracerMemEventType.Free
):
if event_name not in self.allocated_items[memnode.place]:
self.allocated_items[memnode.place][event_name] = (
MemorySummary.MemoryItem(
event_name, memnode.place, 'Allocated'
)
)
self.allocated_items[memnode.place][
event_name
].add_memory_record(memnode.increase_bytes, memnode.type)
elif (
memnode.type == TracerMemEventType.ReservedAllocate
or memnode.type == TracerMemEventType.ReservedFree
):
if event_name not in self.reserved_items[memnode.place]:
self.reserved_items[memnode.place][event_name] = (
MemorySummary.MemoryItem(
event_name, memnode.place, 'Reserved'
)
)
self.reserved_items[memnode.place][
event_name
].add_memory_record(memnode.increase_bytes, memnode.type)
self.peak_allocation_values[memnode.place] = max(
self.peak_allocation_values[memnode.place],
memnode.peak_allocated,
)
self.peak_reserved_values[memnode.place] = max(
self.peak_reserved_values[memnode.place], memnode.peak_reserved
)
def parse(self, nodetrees):
r"""
Analyse memory event in the nodetress.
"""
thread2hostnodes = traverse_tree(nodetrees)
for threadid, host_nodes in thread2hostnodes.items():
for host_node in host_nodes[1:]: # skip root node
if host_node.type == TracerEventType.OperatorInner:
continue
if host_node.type == TracerEventType.Operator:
for child in host_node.children_node:
self._analyse_node_memory(host_node.name, child)
self._analyse_node_memory(host_node.name, host_node)
class StatisticData:
r"""
Hold all analysed results.
"""
def __init__(self, node_trees, extra_info):
self.node_trees = node_trees
self.extra_info = extra_info
self.time_range_summary = TimeRangeSummary()
self.event_summary = EventSummary()
self.distributed_summary = DistributedSummary()
self.memory_summary = MemorySummary()
self.time_range_summary.parse(node_trees)
self.event_summary.parse(node_trees)
self.distributed_summary.parse(node_trees)
self.memory_summary.parse(node_trees)
def _build_table(
statistic_data,
sorted_by=SortedKeys.CPUTotal,
op_detail=True,
thread_sep=False,
time_unit='ms',
row_limit=100,
max_src_column_width=75,
views=None,
):
from .profiler import SummaryView
"""Prints a summary of events."""
# format table row
SPACING_SIZE = 2
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
def add_column(padding, text_dir='<'):
row_format_list[0] += (
'{: ' + text_dir + str(padding) + '}' + (' ' * SPACING_SIZE)
)
header_sep_list[0] += '-' * padding + (' ' * SPACING_SIZE)
line_length_list[0] += padding + SPACING_SIZE
def add_title(padding, text):
left_length = padding - len(text)
half = left_length // 2
return '-' * half + text + '-' * (left_length - half)
result = []
def append(s):
result.append(s)
result.append('\n')
def format_time(time, unit='ms', indent=0):
r"""
Transform time in ns to time in unit.
"""
if time == float('inf'):
return '-'
else:
result = float(time)
if unit == 's':
result /= 1e9
elif unit == 'ms':
result /= 1e6
elif unit == 'us':
result /= 1e3
return '{}{:.2f}'.format(' ' * indent, result)
def format_ratio(ratio, indent=0):
r"""
Transform ratio within [0, 1] to percentage presentation.
"""
return '{}{:.2f}'.format(' ' * indent, ratio * 100)
total_time = statistic_data.time_range_summary.get_cpu_range_sum(
TracerEventType.ProfileStep
)
if views is None or SummaryView.DeviceView in views:
# ----- Print Device Summary ----- #
headers = ['Device', 'Utilization (%)']
name_column_width = 30
DEFAULT_COLUMN_WIDTH = 20
add_column(name_column_width)
for _ in headers[1:]:
add_column(DEFAULT_COLUMN_WIDTH)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Device Summary"))
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
row_values = [
'CPU(Process)',
format_ratio(
float(statistic_data.extra_info['Process Cpu Utilization'])
),
]
append(row_format.format(*row_values))
row_values = [
'CPU(System)',
format_ratio(
float(statistic_data.extra_info['System Cpu Utilization'])
),
]
append(row_format.format(*row_values))
for gpu_name in statistic_data.time_range_summary.get_gpu_devices():
gpu_time = float(
statistic_data.time_range_summary.get_gpu_range_sum(
gpu_name, TracerEventType.Kernel
)
)
utilization = gpu_time / total_time
row_values = [f'GPU{gpu_name}', format_ratio(utilization)]
append(row_format.format(*row_values))
append(header_sep)
append(
"Note:\nCPU(Process) Utilization = Current process CPU time over all cpu cores / elapsed time, so max utilization can be reached 100% * number of cpu cores.\n"
"CPU(System) Utilization = All processes CPU time over all cpu cores(busy time) / (busy time + idle time).\n"
"GPU Utilization = Current process GPU time / elapsed time."
)
append('-' * line_length)
append('')
append('')
if total_time == 0:
return ''.join(result)
if views is None or SummaryView.OverView in views:
# ----- Print Overview Summary ----- #
headers = ['Event Type', 'Calls', 'CPU Time', 'Ratio (%)']
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
DEFAULT_COLUMN_WIDTH = 25
for _ in headers:
add_column(DEFAULT_COLUMN_WIDTH)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Overview Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
cpu_type_time = collections.defaultdict(int)
gpu_type_time = collections.defaultdict(int)
cpu_call_times = collections.defaultdict(int)
gpu_call_times = collections.defaultdict(int)
cpu_call_times.update(statistic_data.time_range_summary.call_times)
gpu_call_times.update(statistic_data.time_range_summary.call_times)
for (
event_type,
value,
) in statistic_data.time_range_summary.CPUTimeRangeSum.items():
if event_type != TracerEventType.Communication:
cpu_type_time[event_type] = value
if statistic_data.distributed_summary.cpu_communication_range:
cpu_type_time[TracerEventType.Communication] = sum_ranges(
statistic_data.distributed_summary.cpu_communication_range
)
cpu_call_times[TracerEventType.Communication] = (
statistic_data.distributed_summary.cpu_calls
)
for event_type in [
TracerEventType.Dataloader,
TracerEventType.Forward,
TracerEventType.Backward,
TracerEventType.Optimization,
]:
event_type_name = str(event_type).split('.')[1]
if (
event_type in cpu_call_times
and event_type_name
in statistic_data.event_summary.model_perspective_items
):
cpu_call_times[event_type] = (
statistic_data.event_summary.model_perspective_items[
event_type_name
].call
)
cpu_type_time[event_type] = (
statistic_data.event_summary.model_perspective_items[
event_type_name
].cpu_time
)
gpu_time_range = collections.defaultdict(list)
for (
device_id,
device_time_ranges,
) in statistic_data.time_range_summary.GPUTimeRange.items():
for event_type, time_range in device_time_ranges.items():
gpu_time_range[event_type] = merge_ranges(
gpu_time_range[event_type], time_range, is_sorted=True
)
for event_type, time_range in gpu_time_range.items():
gpu_type_time[event_type] = sum_ranges(time_range)
if statistic_data.distributed_summary.gpu_communication_range:
gpu_type_time[TracerEventType.Communication] = sum_ranges(
statistic_data.distributed_summary.gpu_communication_range
)
gpu_call_times[TracerEventType.Communication] = (
statistic_data.distributed_summary.gpu_calls
)
sorted_items = sorted(
cpu_type_time.items(), key=lambda x: x[1], reverse=True
)
event_type, time = sorted_items[0]
row_values = [
'{}'.format(str(event_type).split('.')[1]),
cpu_call_times[event_type],
format_time(time, unit=time_unit),
format_ratio(float(time) / total_time),
]
append(row_format.format(*row_values))
for event_type, time in sorted_items[1:]:
row_values = [
' {}'.format(str(event_type).split('.')[1]),
cpu_call_times[event_type],
format_time(time, unit=time_unit),
format_ratio(float(time) / total_time),
]
append(row_format.format(*row_values))
append(header_sep)
headers = ['', 'Calls', 'GPU Time', 'Ratio (%)']
append(row_format.format(*headers))
append(header_sep)
for event_type, time in gpu_type_time.items():
row_values = [
' {}'.format(str(event_type).split('.')[1]),
gpu_call_times[event_type],
format_time(time, unit=time_unit),
format_ratio(float(time) / total_time),
]
append(row_format.format(*row_values))
append(header_sep)
append(
"Note:\nIn this table, We sum up all collected events in terms of event type.\n"
"The time of events collected on host are presented as CPU Time, and as GPU Time if on device.\n"
"Events with different types may overlap or inclusion, e.g. Operator includes OperatorInner, so the sum of ratios is not 100%.\n"
"The time of events in the same type with overlap will not calculate twice, and all time is summed after merged.\n"
"Example:\n"
"Thread 1:\n"
" Operator: |___________| |__________|\n"
"Thread 2:\n"
" Operator: |____________| |___|\n"
"After merged:\n"
" Result: |______________| |__________|\n"
)
append('-' * line_length)
append('')
append('')
if views is None or SummaryView.ModelView in views:
# ----- Print Model Summary Report ----- #
model_perspective_items = (
statistic_data.event_summary.model_perspective_items
)
if len(model_perspective_items) > 1:
all_row_values = []
accumulation_time = 0
gpu_accumulation_time = 0
gpu_total_time = (
statistic_data.event_summary.model_perspective_items[
'ProfileStep'
].gpu_time
)
for name in [
'ProfileStep',
'Dataloader',
'Forward',
'Backward',
'Optimization',
]:
if name in model_perspective_items:
item = model_perspective_items[name]
if gpu_total_time == 0:
gpu_ratio = 0
else:
gpu_ratio = float(item.gpu_time) / gpu_total_time
name = f'{name}' if 'ProfileStep' in name else f' {name}'
row_values = [
f'{name}',
item.call,
f'{format_time(item.cpu_time, unit=time_unit)} / {format_time(item.avg_cpu_time, unit=time_unit)} / {format_time(item.max_cpu_time, unit=time_unit)} / {format_time(item.min_cpu_time, unit=time_unit)} / {format_ratio(float(item.cpu_time) / total_time)}',
f'{format_time(item.gpu_time, unit=time_unit)} / {format_time(item.avg_gpu_time, unit=time_unit)} / {format_time(item.max_gpu_time, unit=time_unit)} / {format_time(item.min_gpu_time, unit=time_unit)} / {format_ratio(gpu_ratio)}',
]
all_row_values.append(row_values)
if 'ProfileStep' not in name:
accumulation_time += item.cpu_time
gpu_accumulation_time += item.gpu_time
other_time = total_time - accumulation_time
other_gpu_time = gpu_total_time - gpu_accumulation_time
if gpu_total_time == 0:
gpu_ratio = 0
else:
gpu_ratio = float(other_gpu_time) / gpu_total_time
row_values = [
' Others',
'-',
f'{format_time(other_time, unit=time_unit)} / - / - / - / {format_ratio(float(other_time) / total_time)}',
f'{format_time(other_gpu_time, unit=time_unit)} / - / - / - / {format_ratio(gpu_ratio)}',
]
all_row_values.append(row_values)
# Calculate the column width
calltime_width = 6
cpu_data_description_width = 40
gpu_data_description_width = 40
for row_values in all_row_values:
if (
isinstance(row_values[1], int)
and len(str(row_values[1])) > calltime_width
):
calltime_width = len(str(row_values[1]))
if len(row_values[2]) > cpu_data_description_width:
cpu_data_description_width = len(row_values[2])
if len(row_values[3]) > gpu_data_description_width:
gpu_data_description_width = len(row_values[3])
headers = [
'Name',
'Calls',
'CPU Total / Avg / Max / Min / Ratio(%)',
'GPU Total / Avg / Max / Min / Ratio(%)',
]
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
name_column_width = 15
add_column(name_column_width)
add_column(calltime_width)
add_column(cpu_data_description_width)
add_column(gpu_data_description_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Model Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
for row_values in all_row_values:
append(row_format.format(*row_values))
append(header_sep)
append(
"Note:\nIn this table, GPU time is the sum of all device(GPU) events called in the phase.\n"
"Unlike overview summary, if two device(GPU) events execute on different streams with overlap time, we sum them directly here.\n"
)
append('-' * line_length)
append('')
append('')
if views is None or SummaryView.DistributedView in views:
# ----- Print Distribution Summary Report ----- #
if statistic_data.distributed_summary.communication_range:
headers = [
'Name',
'Total Time',
'Ratio (%)',
]
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
DEFAULT_COLUMN_WIDTH = 25
for _ in headers:
add_column(DEFAULT_COLUMN_WIDTH)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Distribution Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
communication_time = sum_ranges(
statistic_data.distributed_summary.communication_range
)
computation_time = sum_ranges(
statistic_data.distributed_summary.computation_range
)
overlap_time = sum_ranges(
statistic_data.distributed_summary.overlap_range
)
row_values = [
'ProfileStep',
format_time(total_time, unit=time_unit),
format_ratio(float(total_time) / total_time),
]
append(row_format.format(*row_values))
row_values = [
' Communication',
format_time(communication_time, unit=time_unit),
format_ratio(float(communication_time) / total_time),
]
append(row_format.format(*row_values))
row_values = [
' Computation',
format_time(computation_time, unit=time_unit),
format_ratio(float(computation_time) / total_time),
]
append(row_format.format(*row_values))
row_values = [
' Overlap',
format_time(overlap_time, unit=time_unit),
format_ratio(float(overlap_time) / total_time),
]
append(row_format.format(*row_values))
append(header_sep)
append(
"Note:\nCommunication time: Communication Event time, Communication Op time and its kernel time on gpu.\n"
"Computation time: Kernel time, except kernels belong to communication(nccl kernels).\n"
"Overlap time: Communication time intersects with computation time.\n"
"Example:\n"
"Communication:\n"
" CPU: |_________________|\n"
" GPU: |______________|\n"
" Total: |_________________| |______________|\n"
"Computation time(Kernel):\n"
" GPU: |________________|\n"
"Overlap time: |___________|\n"
)
append('-' * line_length)
append('')
append('')
if views is None or SummaryView.OperatorView in views:
# ----- Print Operator Summary Report ----- #
if statistic_data.event_summary.items:
all_row_values = []
name_column_width = 52
if thread_sep:
thread_items = statistic_data.event_summary.thread_items
else:
thread_items = {
'All threads merged': statistic_data.event_summary.items
}
for thread_id, items in thread_items.items():
all_row_values.append(f"Thread: {thread_id}")
if sorted_by == SortedKeys.CPUTotal:
sorted_items = sorted(
items.items(), key=lambda x: x[1].cpu_time, reverse=True
)
elif sorted_by == SortedKeys.CPUAvg:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].avg_cpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.CPUMax:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].max_cpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.CPUMin:
sorted_items = sorted(
items.items(), key=lambda x: x[1].min_cpu_time
)
elif sorted_by == SortedKeys.GPUTotal:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUAvg:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].avg_general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMax:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].max_general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMin:
sorted_items = sorted(
items.items(), key=lambda x: x[1].min_general_gpu_time
)
total_op_cpu_time = 0
total_op_gpu_time = 0
for name, item in sorted_items:
total_op_cpu_time += item.cpu_time
total_op_gpu_time += item.general_gpu_time
for name, item in sorted_items:
if total_op_cpu_time == 0:
cpu_ratio = 0
else:
cpu_ratio = float(item.cpu_time) / total_op_cpu_time
if total_op_gpu_time == 0:
gpu_ratio = 0
else:
gpu_ratio = (
float(item.general_gpu_time) / total_op_gpu_time
)
row_values = [
name,
item.call,
f'{format_time(item.cpu_time, unit=time_unit)} / {format_time(item.avg_cpu_time, unit=time_unit)} / {format_time(item.max_cpu_time, unit=time_unit)} / {format_time(item.min_cpu_time, unit=time_unit)} / {format_ratio(cpu_ratio)}',
'{} / {} / {} / {} / {}'.format(
format_time(item.general_gpu_time, unit=time_unit),
format_time(
item.avg_general_gpu_time, unit=time_unit
),
format_time(
item.max_general_gpu_time, unit=time_unit
),
format_time(
item.min_general_gpu_time, unit=time_unit
),
format_ratio(gpu_ratio),
),
item.flops,
]
all_row_values.append(row_values)
if op_detail:
for (
innerop_name,
innerop_node,
) in item.operator_inners.items():
if item.cpu_time == 0:
cpu_ratio = 0
else:
cpu_ratio = (
float(innerop_node.cpu_time) / item.cpu_time
)
if item.general_gpu_time == 0:
gpu_ratio = 0
else:
gpu_ratio = (
float(innerop_node.general_gpu_time)
/ item.general_gpu_time
)
if len(innerop_name) + 2 > name_column_width:
innerop_name = innerop_name[
: name_column_width - 5
]
innerop_name += "..."
row_values = [
f' {innerop_name}',
innerop_node.call,
'{} / {} / {} / {} / {}'.format(
format_time(
innerop_node.cpu_time, unit=time_unit
),
format_time(
innerop_node.avg_cpu_time,
unit=time_unit,
),
format_time(
innerop_node.max_cpu_time,
unit=time_unit,
),
format_time(
innerop_node.min_cpu_time,
unit=time_unit,
),
format_ratio(cpu_ratio),
),
'{} / {} / {} / {} / {}'.format(
format_time(
innerop_node.general_gpu_time,
unit=time_unit,
),
format_time(
innerop_node.avg_general_gpu_time,
unit=time_unit,
),
format_time(
innerop_node.max_general_gpu_time,
unit=time_unit,
),
format_time(
innerop_node.min_general_gpu_time,
unit=time_unit,
),
format_ratio(gpu_ratio),
),
'-',
]
all_row_values.append(row_values)
for (
device_node_name,
device_node,
) in innerop_node.devices.items():
if innerop_node.general_gpu_time == 0:
gpu_ratio = 0
else:
gpu_ratio = (
float(device_node.gpu_time)
/ innerop_node.general_gpu_time
)
if (
len(device_node_name) + 4
> name_column_width
):
device_node_name = device_node_name[
: name_column_width - 7
]
device_node_name += "..."
row_values = [
f' {device_node_name}',
device_node.call,
'- / - / - / - / -',
'{} / {} / {} / {} / {}'.format(
format_time(
device_node.gpu_time, unit=time_unit
),
format_time(
device_node.avg_gpu_time,
unit=time_unit,
),
format_time(
device_node.max_gpu_time,
unit=time_unit,
),
format_time(
device_node.min_gpu_time,
unit=time_unit,
),
format_ratio(gpu_ratio),
),
'-',
]
all_row_values.append(row_values)
for (
device_node_name,
device_node,
) in item.devices.items():
if item.general_gpu_time == 0:
gpu_ratio = 0
else:
gpu_ratio = (
float(device_node.gpu_time)
/ item.general_gpu_time
)
if len(device_node_name) + 2 > name_column_width:
device_node_name = device_node_name[
: name_column_width - 5
]
device_node_name += "..."
row_values = [
f' {device_node_name}',
device_node.call,
'- / - / - / - / -',
'{} / {} / {} / {} / {}'.format(
format_time(
device_node.gpu_time, unit=time_unit
),
format_time(
device_node.avg_gpu_time, unit=time_unit
),
format_time(
device_node.max_gpu_time, unit=time_unit
),
format_time(
device_node.min_gpu_time, unit=time_unit
),
format_ratio(gpu_ratio),
),
'-',
]
all_row_values.append(row_values)
# Calculate the column width
calltime_width = 6
cpu_data_description_width = 40
gpu_data_description_width = 40
flops_width = 10
for row_values in all_row_values:
if isinstance(row_values, str):
continue
if (
isinstance(row_values[1], int)
and len(str(row_values[1])) > calltime_width
):
calltime_width = len(str(row_values[1]))
if len(row_values[2]) > cpu_data_description_width:
cpu_data_description_width = len(row_values[2])
if len(row_values[3]) > gpu_data_description_width:
gpu_data_description_width = len(row_values[3])
headers = [
'Name',
'Calls',
'CPU Total / Avg / Max / Min / Ratio(%)',
'GPU Total / Avg / Max / Min / Ratio(%)',
'FLOPs',
]
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
add_column(name_column_width)
add_column(calltime_width)
add_column(cpu_data_description_width)
add_column(gpu_data_description_width)
add_column(flops_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Operator Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
for row_values in all_row_values:
if isinstance(row_values, str):
append(add_title(line_length, row_values))
else:
append(row_format.format(*row_values))
append(header_sep)
append('')
append('')
if views is None or SummaryView.KernelView in views:
# ----- Print Kernel Summary Report ----- #
if statistic_data.event_summary.kernel_items:
all_row_values = []
kernel_items = statistic_data.event_summary.kernel_items
if sorted_by == SortedKeys.GPUAvg:
sorted_items = sorted(
kernel_items.items(),
key=lambda x: x[1].avg_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMax:
sorted_items = sorted(
kernel_items.items(),
key=lambda x: x[1].max_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMin:
sorted_items = sorted(
kernel_items.items(), key=lambda x: x[1].min_gpu_time
)
else:
sorted_items = sorted(
kernel_items.items(),
key=lambda x: x[1].gpu_time,
reverse=True,
)
total_kernel_gpu_time = 0
for name, item in sorted_items:
total_kernel_gpu_time += item.gpu_time
for name, item in sorted_items:
if total_kernel_gpu_time == 0:
gpu_ratio = 0
else:
gpu_ratio = float(item.gpu_time) / total_kernel_gpu_time
row_values = [
name,
item.call,
f'{format_time(item.gpu_time, unit=time_unit)} / {format_time(item.avg_gpu_time, unit=time_unit)} / {format_time(item.max_gpu_time, unit=time_unit)} / {format_time(item.min_gpu_time, unit=time_unit)} / {format_ratio(gpu_ratio)}',
]
all_row_values.append(row_values)
headers = [
'Name',
'Calls',
'GPU Total / Avg / Max / Min / Ratio(%)',
]
# Calculate the column width
name_column_width = 90
calltime_width = 6
gpu_data_description_width = 40
for row_values in all_row_values:
if (
isinstance(row_values[1], int)
and len(str(row_values[1])) > calltime_width
):
calltime_width = len(str(row_values[1]))
if len(row_values[2]) > gpu_data_description_width:
gpu_data_description_width = len(row_values[2])
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
add_column(name_column_width)
add_column(calltime_width)
add_column(gpu_data_description_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Kernel Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
kernel_name_pattern = re.compile(r'(.+?)(<.*>)(\(.*\))')
for row_values in all_row_values:
match = kernel_name_pattern.match(row_values[0])
if match:
name = match.group(1) + match.group(2)
else:
name = row_values[0]
if len(name) > name_column_width:
row_values[0] = name[: name_column_width - 3] + '...'
else:
row_values[0] = name
append(row_format.format(*row_values))
append(header_sep)
append('')
append('')
if views is None or SummaryView.MemoryManipulationView in views:
# ----- Print Memory Manipulation Summary Report ----- #
if statistic_data.event_summary.memory_manipulation_items:
all_row_values = []
memory_manipulation_items = (
statistic_data.event_summary.memory_manipulation_items
)
gpu_total_time = (
statistic_data.event_summary.model_perspective_items[
'ProfileStep'
].general_gpu_time
)
for name, item in memory_manipulation_items.items():
if gpu_total_time == 0:
gpu_ratio = 0
else:
gpu_ratio = float(item.general_gpu_time) / gpu_total_time
row_values = [
name,
item.call,
f'{format_time(item.cpu_time, unit=time_unit)} / {format_time(item.avg_cpu_time, unit=time_unit)} / {format_time(item.max_cpu_time, unit=time_unit)} / {format_time(item.min_cpu_time, unit=time_unit)} / {format_ratio(float(item.cpu_time) / total_time)}',
f'{format_time(item.general_gpu_time, unit=time_unit)} / {format_time(item.avg_general_gpu_time, unit=time_unit)} / {format_time(item.max_general_gpu_time, unit=time_unit)} / {format_time(item.min_general_gpu_time, unit=time_unit)} / {format_ratio(gpu_ratio)}',
]
all_row_values.append(row_values)
headers = [
'Name',
'Calls',
'CPU Total / Avg / Max / Min / Ratio(%)',
'GPU Total / Avg / Max / Min / Ratio(%)',
]
# Calculate the column width
name_column_width = 0
calltime_width = 6
cpu_data_description_width = 40
gpu_data_description_width = 40
for row_values in all_row_values:
if len(row_values[0]) > name_column_width:
name_column_width = len(row_values[0])
if (
isinstance(row_values[1], int)
and len(str(row_values[1])) > calltime_width
):
calltime_width = len(str(row_values[1]))
if len(row_values[2]) > cpu_data_description_width:
cpu_data_description_width = len(row_values[2])
if len(row_values[3]) > gpu_data_description_width:
gpu_data_description_width = len(row_values[3])
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
add_column(name_column_width)
add_column(calltime_width)
add_column(cpu_data_description_width)
add_column(gpu_data_description_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "Memory Manipulation Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
for row_values in all_row_values:
append(row_format.format(*row_values))
append(header_sep)
append('')
append('')
if views is None or SummaryView.UDFView in views:
# ----- Print UserDefined Summary Report ----- #
if statistic_data.event_summary.userdefined_items:
all_row_values = []
gpu_total_time = (
statistic_data.event_summary.model_perspective_items[
'ProfileStep'
].general_gpu_time
)
if thread_sep:
userdefined_thread_items = (
statistic_data.event_summary.userdefined_thread_items
)
else:
userdefined_thread_items = {
'All threads merged': statistic_data.event_summary.userdefined_items
}
for thread_id, items in userdefined_thread_items.items():
all_row_values.append(f"Thread: {thread_id}")
if sorted_by == SortedKeys.CPUTotal:
sorted_items = sorted(
items.items(), key=lambda x: x[1].cpu_time, reverse=True
)
elif sorted_by == SortedKeys.CPUAvg:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].avg_cpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.CPUMax:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].max_cpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.CPUMin:
sorted_items = sorted(
items.items(), key=lambda x: x[1].min_cpu_time
)
elif sorted_by == SortedKeys.GPUTotal:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUAvg:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].avg_general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMax:
sorted_items = sorted(
items.items(),
key=lambda x: x[1].max_general_gpu_time,
reverse=True,
)
elif sorted_by == SortedKeys.GPUMin:
sorted_items = sorted(
items.items(), key=lambda x: x[1].min_general_gpu_time
)
for name, item in sorted_items:
if gpu_total_time == 0:
gpu_ratio = 0
else:
gpu_ratio = (
float(item.general_gpu_time) / gpu_total_time
)
row_values = [
name,
item.call,
f'{format_time(item.cpu_time, unit=time_unit)} / {format_time(item.avg_cpu_time, unit=time_unit)} / {format_time(item.max_cpu_time, unit=time_unit)} / {format_time(item.min_cpu_time, unit=time_unit)} / {format_ratio(float(item.cpu_time) / total_time)}',
'{} / {} / {} / {} / {}'.format(
format_time(item.general_gpu_time, unit=time_unit),
format_time(
item.avg_general_gpu_time, unit=time_unit
),
format_time(
item.max_general_gpu_time, unit=time_unit
),
format_time(
item.min_general_gpu_time, unit=time_unit
),
format_ratio(gpu_ratio),
),
]
all_row_values.append(row_values)
# Calculate the column width
name_column_width = 0
calltime_width = 6
cpu_data_description_width = 40
gpu_data_description_width = 40
for row_values in all_row_values:
if isinstance(row_values, str):
continue
if len(row_values[0]) > name_column_width:
name_column_width = len(row_values[0])
if (
isinstance(row_values[1], int)
and len(str(row_values[1])) > calltime_width
):
calltime_width = len(str(row_values[1]))
if len(row_values[2]) > cpu_data_description_width:
cpu_data_description_width = len(row_values[2])
if len(row_values[3]) > gpu_data_description_width:
gpu_data_description_width = len(row_values[3])
headers = [
'Name',
'Calls',
'CPU Total / Avg / Max / Min / Ratio(%)',
'GPU Total / Avg / Max / Min / Ratio(%)',
]
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
add_column(name_column_width)
add_column(calltime_width)
add_column(cpu_data_description_width)
add_column(gpu_data_description_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(add_title(line_length, "UserDefined Summary"))
append(f'Time unit: {time_unit}')
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
for row_values in all_row_values:
if isinstance(row_values, str):
append(add_title(line_length, row_values))
else:
append(row_format.format(*row_values))
append('')
append('')
if views is None or SummaryView.MemoryView in views:
# ----- Print Memory Summary Report ----- #
if (
statistic_data.memory_summary.allocated_items
or statistic_data.memory_summary.reserved_items
):
for (
device_type,
memory_events,
) in statistic_data.memory_summary.allocated_items.items():
all_row_values = []
sorted_items = sorted(
memory_events.items(),
key=lambda x: x[1].increase_size,
reverse=True,
)
for event_name, item in sorted_items:
row_values = [
event_name,
item.memory_type,
item.allocation_count,
item.free_count,
item.allocation_size,
item.free_size,
item.increase_size,
]
all_row_values.append(row_values)
sorted_reserved_items = sorted(
statistic_data.memory_summary.reserved_items[
device_type
].items(),
key=lambda x: x[1].increase_size,
reverse=True,
)
for event_name, item in sorted_reserved_items:
row_values = [
event_name,
item.memory_type,
item.allocation_count,
item.free_count,
item.allocation_size,
item.free_size,
item.increase_size,
]
all_row_values.append(row_values)
# Calculate the column width
headers = [
'Name',
'Type',
'Allocation Count',
'Free Count',
'Allocation Size',
'Free Size',
'Increased Size',
]
row_format_list = [""]
header_sep_list = [""]
line_length_list = [-SPACING_SIZE]
name_column_width = 50
number_column_width = 15
add_column(name_column_width)
add_column(12)
add_column(number_column_width)
add_column(number_column_width)
add_column(number_column_width)
add_column(number_column_width)
add_column(number_column_width)
row_format = row_format_list[0]
header_sep = header_sep_list[0]
line_length = line_length_list[0]
# construct table string
append(
add_title(line_length, f"Memory Summary - {device_type}")
)
append(
'Peak Allocated Memory: {}'.format(
statistic_data.memory_summary.peak_allocation_values[
device_type
]
)
)
append(
'Peak Reserved Memory: {}'.format(
statistic_data.memory_summary.peak_reserved_values[
device_type
]
)
)
append(header_sep)
append(row_format.format(*headers))
append(header_sep)
for row_values in all_row_values:
if isinstance(row_values, str):
append(add_title(line_length, row_values))
else:
append(row_format.format(*row_values))
append('')
append('')
return ''.join(result)