Two performance targets, not a frame-rate range
After a post by Haze, VideoCardz says it saw internal Intel documents dated August 2026. The figures refer to two separate configurations on Intel’s own Gaming Index. The standard version targets up to 1.16 times the 285K’s score, and the version labelled “DS” up to 1.33 times. That does not imply a gain of 16–33% in every game.
Configuration | XMP memory | PL1 limit | Projected index |
|---|---|---|---|
Core Ultra 9 285K | DDR5-6400 | 250 W | 100 (baseline) |
Nova Lake-S, 28 cores | DDR5-8000 | 285 W | Up to 116 |
Nova Lake-S DS, 28 cores | DDR5-8000 | 296 W | Up to 133 |
Why a separate 20% estimate is circulating
In a post on 29 October, Haze estimated a gaming gain of around 20%, without detailing the configuration.
The 16% and 33% figures come from the August documents that VideoCardz says it reviewed separately.
Memory speed and power complicate the comparison
Both Nova Lake platforms use memory rated at 8,000 MT/s versus 6,400 MT/s for Arrow Lake, a 25% higher nominal transfer rate. Their TDP figures rise by 35 W and 46 W respectively. These are configured limits and do not tell us how much power the processors actually drew in games.
The 285K has eight performance cores and 16 efficient cores, 24 in total. Intel lists 125 W of base power and a maximum turbo power of 250 W. The projection uses precisely 250 W as the reference system’s PL1 limit. Comparing the prototype’s 296 W with the 125 W on the product sheet would mix two different settings.
The projected gap therefore applies to the whole platform. Without a detailed game list, results for each title and independent measurements at identical settings, there is no way to isolate the benefit from the cores, cache or memory.
Key takeaways
The figure is an August target of up to 33% for the DS configuration at 296 W. It is neither a promise about the retail model nor a result from our lab.