(
self,
e2e_latency: float,
timestamps: list[float],
shape_and_decoded_output: str,
gpu_metrics: GPURawMetrics | None,
)
| 101 | self.gpu_metrics = [] |
| 102 | |
| 103 | def accumulate( |
| 104 | self, |
| 105 | e2e_latency: float, |
| 106 | timestamps: list[float], |
| 107 | shape_and_decoded_output: str, |
| 108 | gpu_metrics: GPURawMetrics | None, |
| 109 | ) -> None: |
| 110 | self.e2e_latency.append(e2e_latency) |
| 111 | self._timestamps.append(timestamps) |
| 112 | self._accumulate_ttft_and_itl(timestamps) |
| 113 | self.shape_and_decoded_outputs.append(shape_and_decoded_output) |
| 114 | self.gpu_metrics.append(gpu_metrics) |
| 115 | |
| 116 | def _accumulate_ttft_and_itl(self, timestamps: list[float]) -> None: |
| 117 | timestamps = np.array(timestamps) |
no test coverage detected