Monitoring GPU Performance

NVIDIA GPU

tmam leverages OpenTelemetry to provide visibility into your NVIDIA GPU performance. Track key metrics such as utilization, temperature, memory usage, and power consumption—enabling you to monitor hardware health and optimize AI workloads effectivel

Using Python SDK

Install tmam

import tmam

tmam.init(
otlp_endpoint="YOUR_OTEL_ENDPOINT", otlp_headers="YOUR_OTEL_HEADERS"
collect_gpu_stats=True 
)

GPU Metrics

This section highlights the GPU-related OpenTelemetry metrics collected by tmam. These metrics provide clear insights into GPU performance and resource utilization, making it easier to monitor system behavior in applications that rely on LLMs or other compute-intensive workloads. They complement trace data and are ideal for building dashboards to track GPU usage and performance in real time.