Comprehensive Undervolting Guide for Intel CPUs

Posted by TechGuru77 on October 26, 2023 | Last updated: November 1, 2023 | 12,345 Views | 876 Replies

Hey everyone!

With the increasing power demands and thermal challenges of modern Intel CPUs, undervolting has become a popular method to improve performance, reduce temperatures, and increase power efficiency without sacrificing stability. This guide aims to provide a comprehensive overview of how to undervolt your Intel CPU safely and effectively.

What is Undervolting?

Undervolting is the process of reducing the voltage supplied to your CPU below its default factory settings. CPUs are typically given a voltage "headroom" by manufacturers to ensure stability across a wide range of silicon quality. By reducing this voltage, you can achieve similar performance levels at lower power consumption and therefore lower temperatures.

Benefits of Undervolting:

  • Lower Temperatures: Significantly reduces CPU temps, leading to less thermal throttling and quieter fan operation.
  • Improved Performance: By staying cooler, CPUs can sustain higher clock speeds for longer periods (especially beneficial for laptops and small form factor builds).
  • Reduced Power Consumption: Lower voltage means less power draw, leading to better battery life for laptops and lower electricity bills.
  • Increased Component Lifespan: Running cooler can potentially extend the life of your CPU and other components.

Prerequisites and Warnings:

Disclaimer:

Proceed at your own risk. While generally safe when done correctly, improper undervolting can lead to system instability, crashes, or in rare cases, hardware damage. Ensure you understand the risks involved.

What You'll Need:

  • CPU-Z: To monitor your CPU's current clock speeds and voltage.
  • HWiNFO64: For in-depth monitoring of CPU temperatures, voltages, power consumption, and clock speeds. Essential for stress testing.
  • Intel Extreme Tuning Utility (XTU): Intel's official utility for overclocking and undervolting. User-friendly for beginners.
  • Throttlestop: A powerful third-party utility for undervolting, especially useful for laptops. Offers more granular control than XTU.
  • Stress Testing Software: (e.g., Prime95, Cinebench R23, OCCT) to test stability after making changes.

Getting Started: Baseline Measurements

Before making any changes, it's crucial to establish a baseline:

  1. Open HWiNFO64 and run a sensor-only session.
  2. Run a demanding benchmark like Cinebench R23 (multi-core test for 10 minutes).
  3. Note down your CPU's average and peak temperatures, clock speeds, and voltage under load.
  4. Also, note down your idle temperatures and voltages.

Method 1: Using Intel XTU (Recommended for Beginners)

Intel XTU is a straightforward tool for undervolting. It might not be available for all systems or CPU generations, but if it is, it's a good starting point.

  1. Download and install Intel XTU.
  2. Open XTU. You'll see various performance tuning options.
  3. Navigate to the "Core Voltage" or "CPU Core Voltage Offset" section.
  4. Start by setting a small negative offset, e.g., -50mV (millivolts).
  5. Click "Apply".
  6. Run a stress test (e.g., Cinebench R23 for 5-10 minutes). Monitor temperatures and stability using HWiNFO64.
  7. If stable, repeat step 4 with a further -10mV or -25mV offset.
  8. If you experience crashes or Blue Screens of Death (BSOD), revert to the last stable setting and try a smaller adjustment or stop there.
  9. Once you find a stable undervolt, you can set XTU to start with Windows for automatic application of these settings.

Method 2: Using Throttlestop (Advanced Users & Laptops)

Throttlestop offers more control and is often necessary for laptops where XTU might be locked or limited.

  1. Download and extract Throttlestop. Run it as administrator.
  2. Disable BD PROCHOT: In the main Throttlestop window, uncheck "BD PROCHOT". This prevents other hardware sensors from throttling your CPU unnecessarily.
  3. Adjust CPU Core Voltage Offset:
    • Go to "FIVR" (Full Integrated Voltage Regulator) screen.
    • Under "CPU Core Voltage", set the "Offset" to a negative value (e.g., -75mV).
    • Ensure "Multiplier" is set to "Default" or "Auto" unless you know what you're doing.
    • Click "Apply".
  4. Test for Stability:
    • Close the FIVR window.
    • In the main Throttlestop window, click "Turn On".
    • Run a stress test (e.g., Cinebench R23 or OCCT CPU test for 15-30 minutes). Monitor temps and stability with HWiNFO64.
  5. Iterative Adjustments:
    • If stable, increase the negative offset (e.g., from -75mV to -100mV) via the FIVR screen and re-test.
    • If unstable (crashes, freezes), reduce the negative offset by 5mV or 10mV until stable.
  6. Save Settings: Once a stable undervolt is found, click "OK" in FIVR, then in the main Throttlestop window, click "Save". You can also configure Throttlestop to run at Windows startup.

Throttlestop Specific Features:

  • Voltage Reduction: You can often undervolt Cache/Ring voltage as well, which can also help with thermals. Start small and test rigorously.
  • SpeedShift (EPP): Adjusting the "EPP" (Energy Performance Preference) slider can influence how aggressively the CPU downclocks. Lower values mean more aggressive downclocking (better battery), higher values mean it stays at higher clocks (better performance).
  • Disable Turbo: For extreme power saving, you can disable turbo boost entirely, but this is usually not the goal of undervolting.

Stress Testing and Stability Checks

This is the most critical phase. You need to ensure your CPU is stable under various loads. A good stability test involves:

  • Long-duration benchmarks: Cinebench R23 (multi-core for at least 10 minutes), Prime95 (Small FFTs for temperature, Blend for general stability).
  • Real-world applications: Gaming, video editing, compiling code – whatever you normally do with your PC.
  • Monitor temperatures: Never let your CPU go above 90-95°C during stress tests. Ideally, aim to keep it below 85°C.
  • Watch for errors: Crashes, BSODs, application freezes, or errors reported by stress testing software indicate instability.

A common range for stable undervolting is between -75mV and -150mV, but this varies greatly between CPUs. Some can go as low as -200mV, while others might only tolerate -25mV.

Troubleshooting and Rollback

If your system becomes unstable:

  1. If you can boot into Windows, run XTU or Throttlestop and revert to a previous stable setting or disable the undervolt.
  2. If Windows won't boot, you'll need to clear the CMOS on your motherboard (refer to your motherboard manual) to reset BIOS settings to default. For laptops, you might need to power drain (remove battery if possible, hold power button for 30 seconds, then try booting).

Advanced Topics

Cache/Ring Voltage Offset:

Similar to core voltage, the ring/cache voltage can also be undervolted. This often requires more careful testing as it can impact cache-related operations.

Adaptive Voltage:

Some BIOS/software allow for adaptive voltage, where the voltage adjusts based on the CPU's load. This can be more complex to set up correctly.

LLC (Load-Line Calibration):

This setting in your BIOS helps counteract voltage droop under load. When undervolting, you might need to adjust LLC, but it's often best left at default or a mild setting unless you're an experienced overclocker.

Conclusion

Undervolting is a powerful tool for optimizing Intel CPUs. By following these steps and performing thorough testing, you can achieve a cooler, more efficient, and potentially faster PC. Remember to be patient, test thoroughly, and never compromise stability for maximum voltage reduction.

Happy undervolting!

- TechGuru77

Replies (876)

This is the original post.

Quote Reply

Great guide, TechGuru77! I managed to get a stable -120mV offset on my i9-13900K using Throttlestop. Temps dropped by about 8-10°C under load, which is amazing.

One question: How much can I safely undervolt the Cache voltage? I tried -50mV and it was fine, but I'm hesitant to go further.

Quote Reply

Thanks for the detailed write-up! I'm on an older i7-8700K and XTU was a breeze. Got -100mV stable. My laptop (XPS 15) is a different story, XTU is very limited. Will try Throttlestop when I have more time.

What are the signs of "too much" undervolting beyond BSODs?

Quote Reply

Replying to @Jonas_Lee: Good to hear about your 13900K! For cache voltage, it's highly CPU-dependent. Many find -50mV to -75mV to be safe. Some go lower, but it's very risky. Always test cache-sensitive workloads if you go beyond -50mV. A good starting point is to match your core voltage offset, but monitor closely.

Replying to @Sarah_Jones: Beyond BSODs, instability can manifest as random application crashes, data corruption (rare but possible), system freezes that require a hard reboot, or errors in stress tests that aren't outright crashes. If you notice any weird behavior, it's usually a sign to back off the undervolt slightly.

Quote Reply

Post a Reply