high cpu/reduce cpu consumption

[cforce]

I've noticed significant CPU consumption during Beyla startup, exceeding 100% of a core, and continuous usage remains above 20%. This is far higher than I expected based on the performance documentation and is not acceptable for a continuous profiling setup. I'm unsure whether this is due to incorrect expectations, a bug, or improper usage?

reference 1 core Snapdragon 8 processor ARM

Avg - 13.22% | Max - 149.5%

Avg - 12.07% | Max - 33%

discovery:
  services:
    - exe_path: /
      namespace: MyApplication

# ebpf:
#   wakeup_len: 4096

cpu:
  enabled: true
  metrics:
    - usage_percent
    - load_average

process:
  enabled: true
  metrics:
    - cpu_usage_percent
    - memory_usage_percent
    - disk_io
  include_system_processes: true  # Include all system processes


# OTEL_EXPORTER_OTLP_ENDPOINT

otel_metrics_export:
  interval: 5s
  insecure_skip_verify: false
  protocol: "grpc"
  endpoint: "http://127.0.0.1:4317"
  features: ["application", "application_process", "network"]
  instrumentations: ["*"]

Could you provide a best-practice configuration setup for Beyla to OpenTelemetry Collector that ensures maximum observability while minimizing CPU usage? Additionally, how can we smooth out the data spikes instead of experiencing sharp peaks?