Spaces:
Running
Running
File size: 9,665 Bytes
5e1a30c |
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 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 |
"""
Performance Timing Utilities for Epic 2 Demo
============================================
Provides timing context managers and performance instrumentation for accurate
measurement of component performance in the Epic 2 demo system.
"""
import time
import logging
from contextlib import contextmanager
from typing import Dict, Any, Optional, List
from dataclasses import dataclass, field
from threading import Lock
logger = logging.getLogger(__name__)
@dataclass
class TimingResult:
"""Represents a timing measurement result"""
stage_name: str
start_time: float
end_time: float
duration_ms: float
metadata: Dict[str, Any] = field(default_factory=dict)
@property
def duration_seconds(self) -> float:
return self.duration_ms / 1000.0
@dataclass
class PipelineTimings:
"""Aggregates timing results for a complete pipeline"""
total_start: float
total_end: Optional[float] = None
stages: List[TimingResult] = field(default_factory=list)
@property
def total_duration_ms(self) -> float:
if self.total_end is None:
return 0.0
return (self.total_end - self.total_start) * 1000.0
def get_stage_timings(self) -> Dict[str, Dict[str, Any]]:
"""Get stage timings in format expected by demo UI"""
timings = {}
for stage in self.stages:
timings[stage.stage_name] = {
"time_ms": stage.duration_ms,
"results": stage.metadata.get("results", 0),
"metadata": stage.metadata
}
return timings
def add_stage(self, stage_name: str, duration_ms: float, metadata: Dict[str, Any] = None):
"""Add a completed stage timing"""
current_time = time.time()
stage = TimingResult(
stage_name=stage_name,
start_time=current_time - (duration_ms / 1000.0),
end_time=current_time,
duration_ms=duration_ms,
metadata=metadata or {}
)
self.stages.append(stage)
class PerformanceInstrumentation:
"""Main performance timing instrumentation for Epic 2 demo"""
def __init__(self):
self._active_timings: Dict[str, PipelineTimings] = {}
self._lock = Lock()
def start_pipeline(self, pipeline_id: str) -> PipelineTimings:
"""Start timing a new pipeline"""
with self._lock:
timing = PipelineTimings(total_start=time.time())
self._active_timings[pipeline_id] = timing
return timing
def finish_pipeline(self, pipeline_id: str) -> Optional[PipelineTimings]:
"""Finish timing a pipeline and return results"""
with self._lock:
if pipeline_id in self._active_timings:
timing = self._active_timings[pipeline_id]
timing.total_end = time.time()
del self._active_timings[pipeline_id]
return timing
return None
@contextmanager
def time_stage(self, pipeline_id: str, stage_name: str, metadata: Dict[str, Any] = None):
"""Context manager for timing a pipeline stage"""
start_time = time.time()
try:
yield
finally:
end_time = time.time()
duration_ms = (end_time - start_time) * 1000.0
with self._lock:
if pipeline_id in self._active_timings:
timing = self._active_timings[pipeline_id]
timing.add_stage(stage_name, duration_ms, metadata or {})
logger.debug(f"Stage '{stage_name}' completed in {duration_ms:.2f}ms")
def get_timing(self, pipeline_id: str) -> Optional[PipelineTimings]:
"""Get current timing for a pipeline"""
with self._lock:
return self._active_timings.get(pipeline_id)
class ComponentPerformanceExtractor:
"""Extracts performance metrics from RAG system components"""
@staticmethod
def extract_retriever_metrics(retriever) -> Dict[str, Any]:
"""Extract detailed timing metrics from ModularUnifiedRetriever"""
metrics = {}
# Try to get performance metrics from the retriever
if hasattr(retriever, 'get_metrics'):
component_metrics = retriever.get_metrics()
if component_metrics:
# Extract stats from the actual format
retrieval_stats = component_metrics.get('retrieval_stats', {})
# Get sub-component statistics
sub_components = component_metrics.get('sub_components', {})
# Extract reranker statistics
reranker_stats = sub_components.get('reranker', {}).get('statistics', {})
fusion_stats = sub_components.get('fusion_strategy', {}).get('statistics', {})
# Create metrics in expected format
metrics['dense_retrieval'] = {
'time_ms': retrieval_stats.get('last_retrieval_time', 0) * 1000,
'results': component_metrics.get('indexed_documents', 0)
}
metrics['sparse_retrieval'] = {
'time_ms': retrieval_stats.get('avg_time', 0) * 1000,
'results': component_metrics.get('indexed_documents', 0)
}
metrics['fusion'] = {
'time_ms': fusion_stats.get('avg_graph_latency_ms', 0),
'results': fusion_stats.get('total_fusions', 0)
}
metrics['neural_reranking'] = {
'time_ms': reranker_stats.get('total_latency_ms', 0),
'results': reranker_stats.get('successful_queries', 0)
}
# Total retrieval time
metrics['total_retrieval_time_ms'] = retrieval_stats.get('total_time', 0) * 1000
return metrics
@staticmethod
def extract_generator_metrics(generator) -> Dict[str, Any]:
"""Extract detailed timing metrics from AnswerGenerator"""
metrics = {}
# Try to get performance metrics from the generator
if hasattr(generator, 'get_metrics'):
component_metrics = generator.get_metrics()
if component_metrics:
# Extract stats from the actual format
generation_count = component_metrics.get('generation_count', 0)
total_time = component_metrics.get('total_time', 0)
avg_time = component_metrics.get('avg_time', 0)
# Get sub-component information
sub_components = component_metrics.get('sub_components', {})
llm_client = sub_components.get('llm_client', {})
# Create metrics in expected format
metrics['prompt_building'] = {
'time_ms': avg_time * 1000 * 0.1, # Estimate 10% of total time
'results': generation_count
}
metrics['llm_generation'] = {
'time_ms': avg_time * 1000 * 0.8, # Estimate 80% of total time
'results': generation_count
}
metrics['response_parsing'] = {
'time_ms': avg_time * 1000 * 0.05, # Estimate 5% of total time
'results': generation_count
}
metrics['confidence_scoring'] = {
'time_ms': avg_time * 1000 * 0.05, # Estimate 5% of total time
'results': generation_count
}
# Total generation time
metrics['total_generation_time_ms'] = total_time * 1000
return metrics
@staticmethod
def create_demo_timing_format(retriever_metrics: Dict[str, Any],
generator_metrics: Dict[str, Any]) -> Dict[str, Any]:
"""Create timing format expected by the demo UI"""
return {
# Retrieval stages
"dense_retrieval": retriever_metrics.get('dense_retrieval', {"time_ms": 0, "results": 0}),
"sparse_retrieval": retriever_metrics.get('sparse_retrieval', {"time_ms": 0, "results": 0}),
"graph_enhancement": retriever_metrics.get('fusion', {"time_ms": 0, "results": 0}),
"neural_reranking": retriever_metrics.get('neural_reranking', {"time_ms": 0, "results": 0}),
# Generation stages
"prompt_building": generator_metrics.get('prompt_building', {"time_ms": 0, "results": 0}),
"llm_generation": generator_metrics.get('llm_generation', {"time_ms": 0, "results": 0}),
"response_parsing": generator_metrics.get('response_parsing', {"time_ms": 0, "results": 0}),
"confidence_scoring": generator_metrics.get('confidence_scoring', {"time_ms": 0, "results": 0}),
}
# Global performance instrumentation instance
performance_instrumentation = PerformanceInstrumentation()
@contextmanager
def time_query_pipeline(query: str):
"""Context manager for timing a complete query processing pipeline"""
pipeline_id = f"query_{int(time.time() * 1000)}"
timing = performance_instrumentation.start_pipeline(pipeline_id)
try:
yield timing, pipeline_id
finally:
final_timing = performance_instrumentation.finish_pipeline(pipeline_id)
if final_timing:
logger.info(f"Query pipeline completed in {final_timing.total_duration_ms:.2f}ms") |