Unreleased Core Ultra 9 290K Plus benchmark topples Geekbench rankings as the fastest consumer x86 chip

A new benchmark at Geekbench, courtesy of hardware sleuth HXL on X (formerly Twitter), shows Intel's unreleased Core Ultra 9 290K carving through x86 records with ease. However, since this SKU was a no-show at Intel's latest reveal, this benchmark is merely a glimpse of a chip that may never see the light of day.

On March 11, Intel officially pulled the curtain back on its Core Ultra 200 Plus series, the desktop "Arrow Lake Refresh". However, the reveal was limited to only two CPUs: the Core Ultra 5 250K Plus and the Core Ultra 7 270K Plus, conspicuously leaving behind a third SKU that had been haunting the rumor mill, the Core Ultra 9 290K Plus, which is also the topic of today's conversation. These "Plus" chips aren't a ground-up architectural overhaul; instead, they rely on faster clock speeds, increased die-to-die (D2D) interconnect frequencies, and the introduction of the Intel Binary Optimization Tool (iBOT) for select games and titles.

The test setup for this benchmark features ASUS' ROG STRIX Z890-E GAMING WIFI motherboard, coupled with 64GB of DDR5-6800 memory. Interestingly, the Core Ultra 9 290K Plus appears in Geekbench with its full retail branding rather than a generic engineering string, suggesting that Intel pulled the plug at the absolute last minute, possibly after samples had already reached motherboard partners. Specifications-wise, we are looking at 24 cores (8P+16E) and 24 threads, with a rated maximum frequency of 5.6 GHz, while frequencies during the benchmark hit almost 5.8 GHz. This almost mirrors the existing Core Ultra 9 285K configuration, except for the clock speeds, where the 285K is advertised to hit a maximum of 5.7 GHz.

2

Under the high-performance preset in Windows 11 and using iBOT, the unreleased Core Ultra 9 290K Plus scored 3,747 and 26,117 points in the single-core and multi-core categories, respectively. Using Geekbench's data, this is 17% faster than the 285K in single-core, and 16% faster in multi-core performance. Likewise, the 290K Plus demonstrates a 10% lead over AMD's flagship Ryzen 9 9950X3D in single-core performance and a 17% lead in multi-core performance. In fact, these single-core scores are reaching Apple Silicon territory (if we forget about efficiency for a second), nearing the M4 Max, and just 10-15% shy of the M5 Max.

This jump in performance can be explained by the improved interconnect frequencies and the use of iBOT, which intercepts inefficient instruction sequences and replaces them with optimized ones for Arrow Lake. Believe it or not, Geekbench is among the supported applications for iBOT. That being said, Intel likely scrapped the 290K Plus to keep the spotlight on the mainstream. By pushing the 250K Plus and 270K Plus instead, the company hopes to position LGA1851 as an attractive landing spot for users of 11th Gen (Rocket Lake) or older processors.

The value proposition is compelling, but it comes with notable caveats. Firstly, these new chips are likely the end of the road for the current socket, as Nova Lake is reported to move to the new LGA1954 platform. Secondly, the global DRAM crisis will continue to inflate total system prices; you'll end up paying $200 for the CPU and $500 for the RAM, which makes these chips a much harder sell for budget consumers, who might pivot to AM4 or even LGA1700, as they still have support for cheaper DDR4 RAM.