},
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

import okio.Buffer
import okio.ForwardingSink
import okio.ForwardingSource
import okio.Sink
import okio.Source
import okio.buffer

/**
 * Transmits a single HTTP request and a response pair. This layers connection management and events
 * on [ExchangeCodec], which handles the actual I/O.
 */
class Exchange(
  internal val call: RealCall,
  internal val eventListener: EventListener,
  internal val finder: ExchangeFinder,

      Option<"enable_dynamic_range_per_channel_quantization_for_dense_layers_",
              "enable-dynamic-range-per-channel-quantization-for-dense-layers", "bool",
              "true", "Whether enable per-channel quantized weights for Fully Connected layers (default is per tensor).">,
      Option<"min_elements_for_weights_",
              "min-elements-for-weights", "int64_t", "1024",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

        "Number of quantized layers with quantized outputs: %d/%d",
        total_quantized_func_count - float_output_func_count,
        total_quantized_func_count));
    lines.push_back(absl::StrFormat("Number of quantize layers added: %d",
                                    quantize_func_count));
    lines.push_back(absl::StrFormat("Number of dequantize layers added: %d",

### Default CA Flow through istio-agent

![CA Flow](docs/ca.svg)

A single SDS request from Envoy goes through a few different layers in istio-agent.

1. First, the request is handled by the SDS server. This is mostly just an intermediate translation layer exposing
   the `SecretManager` to Envoy, without much business logic. For each resource requested by Envoy, the SDS server

  // quantization.
  bool verify_numeric = false;
  // Whether to add verification for layer by layer, or on whole model. When
  // disabled (per-layer) float and quantized ops will be run from same input
  // (output of previous quantized layer). When enabled, float and quantized ops
  // will run with respective float and quantized output of previous ops.
  bool whole_model_verify = false;

      {{- if or .region .zone }}
      ,
      {{- end }}
      "sub_zone": "{{ .sub_zone }}"
      {{- end }}
    },
    "metadata": {{ .meta_json_str }}
  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",
            "static_layer": {{ .runtime_flags }}
          },
          {
              "name": "admin",
              "admin_layer": {}
          }
      ]