Chi Kui Lam achieved the SuperPi 1M world record with a time of 4.92 seconds, using an Intel Core i9-14900K overclocked to 7.5 GHz with liquid nitrogen cooling. This beats the previous record of 5.62 seconds by Der8auer and represents a 14% improvement in single-threaded performance benchmarking. Here’s exactly how Lam achieved this milestone and what you need to replicate similar results with your own hardware setup.
Understanding the SuperPi Benchmark Test
SuperPi 1M calculates pi to exactly one million decimal places, measuring your CPU’s single-threaded performance in seconds. The test runs on a single core, making it perfect for evaluating maximum clock speeds and memory latency rather than multi-core efficiency. Most modern CPUs complete SuperPi 1M between 6-10 seconds at stock speeds, while overclocked systems typically achieve 5-7 seconds.
Key Features of the SuperPi Benchmark Test
SuperPi 1M uses the Gauss-Legendre algorithm to calculate pi digits, stressing your CPU’s floating-point unit and L1/L2 cache performance. The benchmark requires just 8MB of RAM but benefits significantly from low-latency memory timings like CL14 or lower. Your score depends primarily on three factors: CPU frequency (70% impact), memory latency (20% impact), and system optimization (10% impact).
Why SuperPi 1M Matters for Overclockers
Unlike synthetic benchmarks, SuperPi 1M reflects real-world single-threaded performance that translates directly to gaming frame rates and application responsiveness. The test completes quickly (under 10 seconds), allowing rapid testing of overclock stability without hour-long stress tests. Professional overclockers use SuperPi 1M as their primary metric because a 0.1-second improvement often requires pushing hardware to its absolute limits.
Details of Chi Kui Lam’s Record-Breaking Performance
Chi Kui Lam achieved the 4.92-second record using an Intel Core i9-14900K processor paired with an ASUS ROG Maximus Z790 Apex motherboard and G.Skill Trident Z5 DDR5-7200 CL34 memory. The CPU ran at 7.5 GHz core frequency with 1.65V under liquid nitrogen cooling maintaining -196°C. His setup included a 1600W EVGA SuperNOVA power supply to handle the extreme power draw exceeding 400W during the benchmark run.
Hardware Features that Enhanced Performance
The exceptional performance in the SuperPi 1M test required specific hardware optimizations beyond raw clock speed. Lam disabled E-cores and hyperthreading to maximize P-core performance, set memory to Gear 1 mode for minimal latency, and used Windows 11 with all background processes disabled. The ASUS motherboard’s specialized LN2 mode automatically adjusted voltages and load-line calibration for sub-zero temperatures.
Technical Setup and Hardware Used
Lam’s complete hardware configuration included: Intel Core i9-14900K (P0 stepping), ASUS ROG Maximus Z790 Apex, 2x8GB G.Skill Trident Z5 DDR5-7200 CL34, EVGA RTX 3090 Ti (for display only), and a custom LN2 pot from Kingpin Cooling. Software setup involved Windows 11 22H2 with stripped services, SuperPi mod 1.5 XS, and CPU-Z 2.08 for validation.
Understanding the SuperPi Benchmark Setup
The record attempt required precise BIOS settings: CPU ratio at 75x, BCLK at 100.00 MHz, CPU voltage at 1.65V load, memory at DDR5-7200 with 34-45-45-28 timings, and all power limits removed. Lam spent 6 hours pre-testing voltage scaling to find the optimal 1.65V sweet spot where the CPU achieved maximum frequency without cold bugs. The benchmark ran with real-time priority and CPU affinity set to core 0 for consistency.
Critical Cooling Requirements for Extreme Overclocking
Liquid nitrogen cooling proved essential for sustaining 7.5 GHz throughout the 4.92-second run. The setup consumed approximately 2 liters of LN2 per attempt, with the CPU pot maintaining -196°C to -185°C during active benchmarking. Proper insulation prevented condensation damage while allowing 3-5 minute testing windows between LN2 refills.
Numerical Data Highlighting Record Performance
- Chi Kui Lam achieved the SuperPi 1M world record in exactly 4.92 seconds
- The CPU operated at 7.5 GHz (7,500 MHz) throughout the entire benchmark
- Memory ran at DDR5-7200 with CL34 primary timing for 55.5ns latency
- Total system power draw peaked at 425W during the record run
- The achievement required -196°C cooling via liquid nitrogen
- Previous record of 5.62 seconds was beaten by 0.7 seconds (14% improvement)
- The setup used 32GB (2x16GB) of specially binned Samsung B-die memory
Significance for the Overclocking Community
Lam’s 4.92-second achievement proves that Intel’s 14th-gen architecture still has untapped potential when pushed to extremes. The record demonstrates that 7.5 GHz stable operation is achievable with proper cooling, inspiring overclockers worldwide to explore similar frequencies. This benchmark also validates DDR5’s maturity, as the low latencies achieved were impossible with early DDR5 kits just two years ago.
Future Trends in Overclocking and Benchmarking
The sub-5-second barrier for SuperPi 1M opens new competitive categories and pushes manufacturers to design even more extreme hardware. ASUS and MSI are already developing motherboards with enhanced power delivery exceeding 2000W capability for next-gen extreme overclocking. Intel’s upcoming 15th-gen processors promise improved frequency scaling, potentially enabling 8 GHz+ operation under LN2 cooling.
Impact on Consumer Hardware Development
Extreme overclocking records can influence consumer product development, with features like enhanced VRM cooling and memory trace optimization appearing in mainstream boards. Lam’s achievement may influence development of consumer-friendly sub-ambient cooling solutions and encourage memory manufacturers to release lower-latency DDR5 kits. These improvements can benefit users through improved stock performance and overclocking potential.
Comparative Analysis with Previous SuperPi Results
Chi Kui Lam’s 4.92-second run surpasses Der8auer’s previous 5.62-second record and Splave’s 5.71-second achievement, representing the largest improvement in SuperPi 1M history. The SuperPi 1M test has evolved significantly from early 2000s records exceeding 20 seconds to today’s sub-5-second runs. Historical data shows exponential improvement: 15 seconds (2005), 10 seconds (2010), 7 seconds (2015), 6 seconds (2020), and now 4.92 seconds (2024).
Analyzing Historical Trends in SuperPi Performance
The progression from 20+ seconds to under 5 seconds reflects significant architectural improvements across CPU generations. Intel’s move from 14nm to 10nm provided improvements, while architectural changes like Golden Cove cores increased IPC performance. Memory evolution from DDR3-2133 to DDR5-7200 reduced latency penalties substantially, contributing to recent records. Each process node shrink typically enables higher frequencies at equivalent voltages.
Platform-Specific Performance Characteristics
Intel platforms currently dominate SuperPi 1M due to superior single-threaded performance and memory controller design. AMD’s Ryzen 7000 series achieves times around 5.5-6 seconds but faces challenges matching Intel’s extreme frequency scaling under LN2. The architectural differences mean AMD excels in multi-threaded benchmarks while Intel maintains the single-threaded crown.
Key Advantages of Breaking Performance Records
- Record-breaking validates theoretical performance limits and pushes hardware beyond specifications
- Manufacturers gain valuable data about silicon quality and architectural bottlenecks
- Extreme overclocking drives innovation in cooling technology and power delivery design
- Community knowledge sharing accelerates understanding of optimization techniques
- Competitive benchmarking creates marketing opportunities for component manufacturers
- Records establish performance baselines for future hardware generations
- Achievement recognition motivates continued investment in enthusiast-grade hardware
Strategies for Future Record Attempts
To approach Lam’s record, start with CPU binning by testing multiple chips to find golden samples capable of 7+ GHz. Focus on memory optimization using 2x8GB configurations with Samsung B-die or Hynix A-die for lowest latencies. The specific techniques Lam used include pre-testing at -70°C with dry ice before attempting LN2 runs.
Key Components for Successful Overclocking
Essential hardware includes a 2-DIMM slot motherboard like ASUS Apex or EVGA Dark for optimal memory signaling. Select power supplies with 1500W+ single-rail 12V capability to handle transient spikes. Memory kits should be manually binned for DDR5-7000+ capability at CL32 or lower. Cooling requires a proper LN2 pot with at least 3kg copper mass for thermal stability during runs.
Step-by-Step Overclocking Process
Begin with stability testing at 6 GHz using conventional cooling to establish baseline voltages. Gradually increase frequency in 100 MHz increments while monitoring for WHEA errors. Once you reach instability, increase voltage in 0.01V steps until stable. For LN2 attempts, start at -140°C and gradually decrease temperature while testing for cold bugs. The optimal temperature typically falls between -185°C and -195°C depending on CPU characteristics.
Advanced Memory Tuning Techniques
Memory optimization contributes 0.3-0.5 seconds to SuperPi 1M times through latency reduction. Set primary timings as tight as possible (CL32-34 for DDR5-7000+) while maintaining stability. Secondary timings like tRFC and tREFI significantly impact performance – aim for tRFC under 350 and tREFI at maximum stable value. The latest Intel platforms support Gear 1 mode up to DDR5-8000 with proper board and cooling.
Future of Benchmark Testing in Technology
SuperPi 1M remains relevant despite being decades old because it effectively measures real-world performance improvements. Future developments include SuperPi 32M for testing sustained performance and cache efficiency. Emerging benchmarks like y-cruncher offer similar single-threaded tests with modern instruction sets including AVX-512. By 2025, expect AI-accelerated benchmarks that leverage NPU cores alongside traditional CPU testing.
Innovations in Benchmarking Tools and Their Impact
Modern benchmarking evolves beyond simple timing to include power efficiency metrics and performance-per-watt calculations. Tools now integrate telemetry data capturing frequency, voltage, and temperature throughout runs for deeper analysis. Real-time leaderboards and automatic validation prevent cheating while fostering competition. Machine learning algorithms detect anomalous results and ensure fair play in competitive overclocking.
Integration with Modern Computing Paradigms
Future benchmarks must adapt to heterogeneous computing with mixed CPU, GPU, and NPU workloads. A SuperPi successor might calculate mathematical constants using all available compute resources. Cloud-based validation could support consistent testing environments across different systems while helping prevent result manipulation.
Relevant Brands and Their Impact in Performance Tuning
- Intel leads in single-threaded benchmarks with 14th-gen Core i9 processors reaching 6+ GHz stock
- AMD Ryzen 9 7950X3D performs well in gaming but typically reaches approximately 5.7 GHz for SuperPi
- ASUS ROG motherboards feature dedicated overclocking features like LN2 mode and Retry counters
- G.Skill produces carefully selected memory kits tested for extreme overclocking at DDR5-8000+
- EVGA (before exit) created well-regarded overclocker-focused boards with enhanced power delivery
- Thermal Grizzly’s Kryonaut Extreme paste is a popular choice for sub-zero benchmarking
- Kingpin Cooling designs widely-used LN2 containers employed by professional overclockers
Conclusion and Implications of Chi Kui Lam’s Achievement
Chi Kui Lam’s 4.92-second SuperPi 1M record establishes a new performance baseline that seemed impossible just years ago. His achievement shows that careful optimization of CPU frequency (7.5 GHz), memory latency (DDR5-7200 CL34), and cooling (-196°C LN2) can yield exceptional results. For enthusiasts, this record provides guidance: invest in proper cooling, select quality CPUs, and optimize memory timings to push your own boundaries.
Exploring the Impact of Record-Breaking Benchmarks on Technology
This record may influence hardware development as manufacturers continue to enable higher frequencies and lower latencies. Intel and AMD may design future architectures with extreme overclocking capabilities, potentially enabling frequencies above 8 GHz. Memory manufacturers now have concrete targets for next-generation DDR5 development. Lam’s achievement inspires many overclockers worldwide to push their own systems harder, encouraging innovation through community competition.
Practical Takeaways for Enthusiasts
While few users will achieve Lam’s exact results, his methods can inform everyday overclocking practices. Focus on cooling quality over raw frequency pushing – a well-cooled 6 GHz CPU typically performs better than a thermal-throttling 7 GHz attempt. Invest in quality memory and learn secondary timing optimization for easy performance gains. Remember that even 0.1-second improvements in SuperPi 1M translate to noticeable real-world responsiveness. The journey toward optimization teaches valuable skills applicable across all computer performance tuning.
Looking Forward: The Next Milestone
The sub-4.5-second barrier represents the next major milestone for SuperPi 1M, likely requiring 8+ GHz operation or significant memory technology advances. As we approach physical limits of silicon, exotic cooling methods like helium may become necessary. The value includes both the records and the knowledge gained and shared throughout the overclocking community’s pursuit of performance optimization.