HWiNFO64 is the most comprehensive free hardware monitoring tool available for Windows. It reads hundreds of sensors across your CPU, GPU, motherboard, memory, and storage — more than any other utility. This guide covers how to set it up, which metrics actually matter, and how to build an on-screen display overlay for real-time monitoring while gaming.
Installing HWiNFO64
Download HWiNFO64 from hwinfo.com. The portable version works without installation — useful if you want to run it from a USB drive for diagnostics. The installer version integrates better with Windows startup.
On first launch, you will see two windows:
- System Summary: A quick overview of detected hardware
- Sensors window: The main view with all sensor readings
Click Start on the Sensors window and let it populate. HWiNFO64 polls hundreds of sensors, so the first load takes several seconds.
Understanding the Sensor Layout
The Sensors window organizes readings into collapsible sections, each corresponding to a hardware component:
- Motherboard — board temperatures, VRM temps, fan headers
- CPU — per-core temperatures, package power, clock speeds, voltage
- DRAM — memory voltage and temperature (if supported by your RAM)
- GPU — GPU temperature, hotspot temp, VRAM temp, fan speed, clock speeds, power draw
- Storage — NVMe and HDD SMART data including temperature and wear indicators
- Network — bandwidth per adapter
Each row shows: current value, minimum recorded, maximum recorded, and average. The Min and Max columns are invaluable for diagnosing issues — they capture peaks that you would miss watching in real time.
Key Metrics to Watch
Not all sensors are equally important. Focus on these:
CPU Metrics
| Metric | Healthy Range | Concern Zone |
|---|---|---|
| CPU Package Power | Varies by TDP | Above rated TDP (sustained) |
| CPU Core Temp (Max) | Under 85°C gaming | Above 95°C sustained |
| CPU Clock (All Cores) | Near boost spec | Significantly below spec |
| CPU Core Voltage | 1.0–1.4V depending on CPU | Above 1.5V |
CPU Package Power is more informative than temperature alone. If your CPU is hitting thermal limits (throttling), you will see package power drop while temperatures sit at the thermal junction limit (100°C for most Intel, 95°C for AMD Ryzen).
GPU Metrics
| Metric | What It Tells You |
|---|---|
| GPU Temperature (Die) | Core junction temp — main thermal indicator |
| GPU Hotspot Temperature | Highest temp on the die — 20–30°C above core is normal |
| GPU Memory Temperature | VRAM temp — critical on GDDR6X cards (normal up to ~100°C) |
| GPU Memory Used | VRAM consumption — look for pressure in heavy scenes |
| GPU Core Clock | Actual boost frequency — drops indicate throttling |
| GPU Power | Wattage draw — compare against TDP spec |
GPU Hotspot is the most important GPU thermal reading after die temperature. NVIDIA RTX 30/40 series GPUs frequently show hotspot temps 15–30°C above the core temp. Up to 105°C hotspot is within NVIDIA’s spec but warrants better airflow or thermal pad replacement.
VRAM temperature became critical with GDDR6X cards (RTX 3080, 3090, and 4080/4090). These chips run hot by design but sustained readings above 110°C can cause throttling.
Integrating with RTSS for an OSD
RivaTuner Statistics Server (RTSS) provides the overlay engine that renders HWiNFO64 sensor data on-screen while gaming. Both tools together create a powerful real-time HUD.
Setup Steps
- Download and install RTSS from
guru3d.com/rivatuner - Download and install the MSI Afterburner package — it bundles RTSS but you can use RTSS standalone
- Open HWiNFO64 Sensors window
- Click the small graph/OSD icon at the bottom left of the Sensors window (looks like a bar chart)
- This opens the Custom User Gadget or the SharedMemory export dialog — in recent versions, right-click any sensor row and select Add to OSD (RTSS)
- Alternatively, enable Shared Memory Support in HWiNFO64 settings (File > Settings > Plugins tab), then open RTSS and enable HWiNFO sensor reading
Once connected, RTSS shows your selected sensors as an overlay in the top-left (or configured position) of your games.
Recommended OSD Sensors
A clean, useful OSD setup includes:
CPU: Package Temp | All-Core Clock | CPU Power
GPU: Die Temp | Hotspot Temp | VRAM Used | Core Clock | GPU Power
System: RAM Used | Framerate (from RTSS itself)Keep the OSD compact — too many readings create visual noise. Focus on the sensors that tell you whether your hardware is healthy and what it is doing.
Custom Alerts
HWiNFO64 can trigger an alert (popup, sound, or log entry) when a sensor crosses a threshold. Access this by right-clicking any sensor and choosing Set Alert.
Useful alert thresholds:
| Sensor | Alert If Above |
|---|---|
| CPU Package Temp | 90°C |
| GPU Die Temp | 85°C |
| GPU Hotspot Temp | 105°C |
| GPU Memory Temp | 105°C |
| NVMe SSD Temp | 70°C |
Alerts are especially useful for long renders or benchmark sessions where you are not watching the screen continuously.
Logging for Benchmarks
HWiNFO64 can log all sensor data to a CSV file for later analysis. This is invaluable for comparing results before and after hardware changes (new thermal paste, different RAM speeds, overclocking).
To start logging:
- In the Sensors window, go to File > Start Logging
- Choose a file path and name (e.g.,
benchmark_run1.csv) - Run your test or benchmark
- Go to File > Stop Logging
The CSV contains timestamped readings for every sensor at your configured polling interval. Open it in Excel or import into a tool like HWiNFO Log Viewer (third-party, free) for graphical analysis.
Interpreting Normal vs Concerning Readings
CPU throttling: If CPU clock speeds drop below base clock during load and temperatures are at TJMax (100°C Intel, 95°C AMD), the CPU is thermally throttling. Solution: re-apply thermal paste, improve airflow, or adjust power limits.
GPU throttling: GPU clock drops mid-game while temperatures are high. Check GPU power — if it is at the power limit, the GPU is power-throttling. This is normal for power-limited laptops but indicates a thermal compound or airflow issue on desktops.
VRAM pressure: If GPU Memory Used approaches or reaches your card’s VRAM capacity during a game, you will see stutters and possibly driver crashes. Lower texture quality settings or upgrade to a card with more VRAM.
NVMe thermals: Modern NVMe SSDs throttle reads and writes when temperatures exceed 70°C. If your NVMe runs above this under sustained load, add a heatsink. Most mid-range and high-end motherboards include M.2 heatsinks.
HWiNFO64 combined with RTSS gives you visibility into what your hardware is actually doing — not what manufacturers claim it does. Run it during a gaming session, check the Max column afterward, and you will know immediately where your system’s bottlenecks and thermal concerns lie.