
How Intel Fell Behind - Can Nova Lake Turn It Around?
AI Summary
This video provides a comprehensive overview of Intel's flagship processors over the last decade, examining their performance in modern games and tracing Intel's evolution in CPU architecture and manufacturing.
The journey begins with the Intel Core i7-7700K, released in 2017. While essentially a refresh of the 2015 6700K, it featured four cores, eight threads, 8MB of L3 cache, and a 91W TDP. Its primary upgrade was in the integrated GPU, with enhanced HEVC 10-bit and VP9 decoding, enabling official 4K Netflix streaming on Windows, making it popular for home theater PCs. However, for gaming, performance gains were minimal, with only a 7% clock speed increase leading to about a 5% performance uplift. The 7700K’s 14nm process was highly optimized, allowing for impressive clock speeds up to 5GHz, but thermal paste quality, rather than silicon, became the bottleneck for overclockers, leading to the practice of "delidding." Despite its limitations, the 7700K, even nearly a decade later, remains usable in many modern titles, averaging 78 FPS with 1% lows of 54 FPS across a 14-game suite. It struggles significantly with ray tracing, as seen in Cyberpunk 2077 and Spider-Man 2, where its CPU-intensive BVH processing is overwhelmed. Disabling ray tracing and lowering other CPU-intensive settings dramatically improves performance.
Intel, under pressure from AMD's eight-core Ryzen processors, released the Core i7-8700K in 2017, using the same LGA 1151 socket but requiring a new motherboard due to a BIOS-level lock. This "Coffee Lake" generation was still on a refined 14nm++ process, with identical IPC to Skylake. The key upgrade was the jump to six cores and 12 threads, with 12MB of L3 cache. This significantly boosted performance, especially in games that could utilize more cores. The 8700K was 36% faster than the 7700K in Spider-Man 2 with ray tracing and 49% faster in Cyberpunk, aligning with the core count increase.
The Core i9-9900K arrived a year later, bringing the i9 branding to the mainstream. It featured eight cores and 16 threads, matching AMD's Ryzen 7. A critical improvement was the reintroduction of soldered thermal interface material (TIM), addressing the heat issues of the 14nm++ process. Despite a 95W TDP rating, it routinely consumed over 200W, requiring robust motherboard VRMs. The 9900K maintained Intel's performance crown, offering solid gaming performance even today.
Just six months later, Intel launched the 10th generation Core i9-10900K on the new LGA 1200 socket. This marked the end of Intel's 14nm era, pushing the Skylake-derived architecture to its limits with 10 cores and 20 threads. The massive die size and heat generation led to physical packaging innovations like a thinned die and thicker IHS. The introduction of "Thermal Velocity Boost" (TVB) allowed single-core boosts up to 5.3GHz, but only if temperatures stayed below 70°C. The 10900K also frustratingly supported PCIe 4.0 traces on its motherboards, but the feature remained dormant until the 11th generation.
The 11th generation, codenamed "Rocket Lake," was a stop-gap measure due to delays in Intel's 10nm process. It backported the mobile "Sunny Cove" architecture (Cypress Cove) to the 14nm node. This "Frankenstein" approach resulted in massive core designs, forcing a reduction in core count. The flagship Core i9-11900K launched with only eight cores and 16 threads, a step back from the 10900K. This generation saw a slight performance regression in gaming compared to the 10th gen, while AMD’s Zen 3 architecture had already surpassed Intel.
The 12th generation, "Alder Lake," marked a radical redesign. Built on the "Intel 7" process (a refined 10nm node), it introduced a hybrid architecture with eight Performance-cores (P-cores) and eight Efficient-cores (E-cores), resulting in a 16-core, 24-thread configuration. Alder Lake was the first mainstream desktop architecture to support DDR5 memory and PCIe 5.0. While initially showing modest gains over the 11th gen with DDR4, pairing with DDR5 provided a significant uplift. The 12900K was 59% faster than the 7700K on DDR4, but with a much larger die size.
The Raptor Lake architecture, with the Core i9-13900K, refined Alder Lake. Key upgrades included significantly increased L2 cache for P-cores and E-core clusters, boosting multi-thread performance. The 13900K featured 24 cores and 32 threads, competing directly with AMD's high-core-count processors. It achieved extreme clock speeds, with the 13900KS reaching 6GHz out of the box. However, power consumption soared, with the 13900K routinely exceeding 300W, and Intel acknowledged 100°C operation as intended. Widespread stability and degradation issues were later reported, linked to Vmin shift instability, leading Intel to extend warranties. The 13900K was 15-16% faster than the 12900K in gaming, but at the cost of extreme power usage. The 14th gen was largely a minor refresh.
The Core Ultra 200 series, "Arrow Lake," represents a significant philosophical shift, prioritizing power efficiency. It abandoned hyperthreading, with P-cores now single-threaded, resulting in a 24-core, 24-thread configuration. Arrow Lake utilizes TSMC's 3nm node for its compute tile and employs a multi-chip module (MCM) design. This shift led to lower clock speeds and increased inter-tile latency, hurting gaming performance. The 200 series was slower than the 13th and 14th gen in gaming, initially showing only a 5-7% improvement over the 12900K. However, updated models like the 270K Plus offer better performance and pricing, with improved inter-tile communication boosting gaming performance to match the 14900K while consuming less power.
Looking ahead, the "Nova Lake" (Core Ultra 400 series) is expected to bring significant architectural changes, including new P-cores and E-cores, a dual compute tile design, and potentially up to 144MB of L3 cache to combat AMD's X3D processors. This will necessitate a new LGA 1954 socket, continuing Intel's practice of frequent socket changes, though there's hope for longer support.