ISOOC invites all interested parties to join the conversation about setting industry standards for overclocking world record validation requirements.
“Scientific discovery and scientific knowledge have been achieved only by those who have gone in pursuit of it without any practical purpose whatsoever in view.”
Max Planck
This scientific paradigm can also be applied to PC enthusiasts trying to break benchmark performance and frequency records. While such efforts may not appear to carry value beyond personal gratification and eliciting expression of awe from peers, collectively the overclocking community is pursuing the interest of building scientific knowledge and understanding of the performance and frequency limits of processors.
Validating such performance or frequency record claims has changed significantly over the years. In this blog post, we aim to shed light on the evolution of the operating frequency overclocking world records validation and explain how ISOOC can play a role in future standard-setting efforts.
History of Validating Clock Frequency Overclocking World Records
To frame the discussion of standardization of validation requirements, let’s have a closer look at the history of validating CPU clock frequency overclocking world records.
The history of validating CPU and memory operating frequency world records is documented by SkatterBencher on their personal website. The history dates back to the mid-nineties and coincides with the emergence of the internet as a means of communication.
In the early days of overclocking, people would simply share their results on message boards without much proof. Members of the message boards would accept or reject the report based on the credibility of the poster. One such example is the submission of a benchmark result by Richard Miles on August 28, 1996, to the Tomshardware database. The benchmark was completed on a Pentium processor overclocked to 225 MHz.
As more people joined the pursuit of achieving the highest operating frequency, there was an increasing demand for better, more secure validation methods as well as a more organized database for reporting achievements.
One of the earliest reputable attempts at improving the security of record validation is a software called WCPUID, developed by a Japanese PC enthusiast nicknamed H.Oda! in the early 2000s. The software offered a means to accurately report the operating frequency of various parts of the system, including CPU, base clock, and memory. Additionally, a verification screenshot included a CRC code which could be used to confirm the reliability of information presented in the screenshot.
The emergence of a more reliable validation surface sparked an explosion of databases and websites tracking the highest achieved frequency for CPUs, FSB, and memory. For tracking CPU frequency world records, the most prominent database at the time was the Japanese “WCPUID Max Ranking” leaderboard hosted by Son. However, it was certainly not the only database based on records validated with the WCPUID software. Several other websites and forums kept track of the frequency records, including for example the record database at VR-Zone, a Singaporean IT media.
Fierce competition among enthusiasts amidst an intense frequency race between semiconductor giants Intel and AMD drove the frequency record from 1370 MHz at the beginning of the new millennium to 6495 MHz only five years later. Unfortunately, the intense competition also attracted malevolent players to the game, creating a demand for software that would further improve the security of frequency reporting.
In 2005, CPUID launched its online frequency validation website for their popular CPU-Z software tool. While it has gone through many iterations and security improvements, for the past two decades the CPU-Z validation database has been the de facto industry standard for CPU and memory frequency world record validations.
Current Standards for Frequency Overclocking World Records
To make a legitimate claim to a CPU or memory overclocking world record, there are currently three standards in use: HWBOT, CPU-Z, and Guinness.
- The most widely accepted source to validate an overclocking claim is by submitting to the HWBOT CPU and Memory Frequency leaderboards. HWBOT employs its own set of rules and requirements for a record to be accepted, which includes a valid CPU-Z validation link as well as other requirements.
- The second standard is the one imposed by the CPU-Z validation database which requires the CPU to pass a heavy all-core workload before officially validating a result with a green “Validated” sticker.
- The third, least widely used, standard is set by Guinness World Records. This standard requires the frequency to be shown for at least three seconds and the final result to be approved by HWBOT administrators.
With the emergence of the system-on-chip (SoC) and the deep integration of technologies on the silicon chip, as well as the emergence of complex and dynamic clocking technologies, accurately assessing and measuring the operating frequency has become much more complicated. In recent years, there have been multiple occasions where overclocking records were claimed which turned out to be software reading errors.
ISOOC sees a need to review the current standards and adapt to the new situation. Before proposing how ISOOC can play a role in future standard-setting, let’s briefly elaborate on the challenge of operating frequency measurements.
Measuring Operating Clock Frequency
Current advanced clocking technologies typically consist of a reference clock source which is used by a PLL (phase-locked loop), FLL (frequency-locked loop), or AFLL (asymmetric frequency-locked loop) to generate the clocking signal driving the processor. In some cases, the operating frequency may dynamically adjust due to other parameters such as voltage noise. This dynamic clocking technology gives rise to the distinction between the target clock frequency and the effective clock frequency.
- The target clock frequency refers to programming the PLL or FLL. For example, to operate at 5.5 GHz with a reference clock of 100 MHz, you’d program the PLL to utilize a ratio of 55X.
- The effective clock frequency refers to the measurement of the total clock cycles across a period. For example, 5.5 billion clock cycles per second equals a 5.5 GHz operating frequency.
The effective clock frequency can diverge significantly from the user-defined target clock frequency. Power-saving or frequency-throttling technologies can reduce the operating frequency significantly depending on the processor configuration.
One example of such technology is Intel’s Fast Throttle which employs clock modulation to reduce the effective clock frequency in case the CPU overheats rather than adjusting the user-configured target frequency. Another example of such technology is the AMD’s Voltage Adaptive Operation which adjusts the effective clock frequency based on the voltage noise. If the voltage droops under load, the effective clock is adaptively adjusted to ensure stability without adjusting the target clock.
An additional layer of complexity is introduced by the integration of on-chip power management technologies. To program the processor, the user must rely on the firmware implementation by the semiconductor vendor. Firmware, just like any other form of software, may contain bugs or broken code.
One such example is Intel’s FLL OC Mode function which enables programming the FLL to high frequency range. Setting the correct FLL OC mode is a mandatory requirement to pass the CPU-Z validation process. Another such example is HWBOT’s requirement to validate memory frequency world records using an oscilloscope since Haswell due to incorrect memory frequency programming on certain platforms.
When considering the standard for setting a frequency world record, it is reasonable to require the part to actually operate at that frequency as opposed to the part being configured to target the clock frequency. However, measuring the operating frequency is a complicated matter.
There are two methods currently in use to report the actual operating frequency of a processor. The first method consists of taking timestamps from frequency counters in the CPU register as a means to record the effective clock frequency across an interval. The second method is to measure the effective clock cycles across the interval during a workload.
These methods apply specifically to the validation of CPU core frequency.
Building Consensus: A Call to Action
The ISOOC recognizes that establishing an industry standard requires collaboration. We invite all stakeholders with a vested interest in fair and accurate overclocking validation to join the discussion. This includes:
- Overclocking Enthusiasts: The backbone of the overclocking community. Their experience and insights are crucial for crafting a workable standard.
- Hardware Manufacturers (OEMs): CPU, motherboard, and memory manufacturers have a deep understanding of their products’ operation and capabilities. Their input is essential for ensuring the standard reflects real-world hardware behavior.
- Validation Tool Developers: Software and hardware developers who create benchmarking and validation tools play a critical role. Their expertise is needed to ensure the standard aligns with existing validation tools and encourages continued development.
- Competitive Benchmarking Platforms: Platforms like HWBOT provide a stage for overclockers to showcase their achievements. Their perspective on record-keeping and validation processes is valuable.
By working together, this collective knowledge can shape a future-proof standard that fosters healthy competition, protects against fraudulent claims, and celebrates the true pioneers of overclocking.
We at the ISOOC strongly believe that an industry-accepted standard will benefit everyone involved. It will elevate the credibility of overclocking records, inspire confidence in the community, and empower enthusiasts to push the boundaries of hardware performance even further. We invite all interested parties to reach out to ISOOC and join the conversation. Together, we can create a standard that celebrates the spirit of overclocking and ensures its longevity for generations to come.