An Intel Nova Lake leak suggests the following technology of Core Extremely chips may have 60% extra multi-threaded efficiency, however the numbers simply don't add up
Regardless of not being anticipated to look till 2026, rumours and leaks about Intel’s subsequent technology of desktop CPUs proceed to return thick and quick. The most recent snippet claims that Nova Lake may have “management gaming efficiency” coming from greater than 10% higher single-threaded (ST) and 60% greater multi-threaded (MT) efficiency positive aspects.
Nonetheless, it is not a very complete leak, only a snapshot of what seems like an inside Intel doc, posted by X consumer Recreation.Retains.Loading (through 3DCenter). To start with, it would not explicitly state that it is about Nova Lake, the successor to Arrow Lake, however the truth that “New Low-Energy Island” is a characteristic just about factors to it being concerning the subsequent technology of desktop CPUs. That is only a cluster of low-power E-cores, used for background duties, that first appeared in Meteor Lake however not in any desktop chip up to now.
These obtained leaked 🤣🤣 pic.twitter.com/gE6TzL3qTWJune 28, 2025
Apart from that, there are simply two extra statements: “Management Gaming Efficiency” and “>1.1x greater ST and 1.6x MT efficiency”. The previous is commonplace advertising and marketing fluff, and there is nothing to essentially notice right here, apart from Intel is clearly attempting to get one over AMD within the gaming market. The Ryzen 7 9800X3D is so significantly better than the Core Extremely 9 285K in video games that it is apparent Intel cannot ignore it.
The only- and multi-threaded (ST/MT) efficiency positive aspects are one other matter, although. An uplift of 10% for ST is sort of definitely referring to the P-cores, and whereas that may not sound like a lot of a lift, it is par for the course with CPU architectural modifications as of late. In spite of everything, AMD claimed that Zen 5 had a 16% higher common IPC (directions per clock) than Zen 4, however video games did not present something like that at launch.
Nonetheless, 60% higher multi-threaded efficiency is an odd one, partly as a result of there is not any indication what the comparability is with and partly as a result of it has been rumoured that the Nova Lake vary will prime out with a 16 P-core, 32 E-core monster. If one assumes that Intel is evaluating that chip to the 285K (8 P-cores, 16 E-cores), then clock-for-clock, you are taking a look at a 100% uplift in MT efficiency.
So the place’s the opposite 40%? It is potential {that a} 16P+32E monster CPU simply cannot keep wherever close to its most clock speeds when absolutely loaded with threads, but when it is dropping that a lot, it is a bit worrying. One may argue that Intel is not planning on launching with a 48-core beast and as a substitute will begin off with an 8P+32E or a 12P+24E chip at greatest.
A Nova Lake processor with 36 or 40 threads would have an MT uplift over the 285K of fifty% to 67%, clock-for-clock, which is extra in keeping with the 60% declare within the leak, besides, the numbers simply do not make lots of sense proper now. Particularly that the one factor Intel chips do not want extra of is cores.
One concern I’ve with Intel throwing a mountain of cores into its subsequent desktop CPUs is how they entry and share the final degree of cache or LLC (at present L3, within the case of Arrow Lake). Every P-core and E-core cluster has its personal slice of LLC, however they will entry different slices through a hoop bus. The extra cores/clusters one provides to the ring, the extra ring stops should be current, and the upper the latency one will get with core-to-core cache accesses.
Cache latency is at present Intel’s greatest weak point in opposition to AMD’s Zen 5, particularly when 3D V-Cache is introduced into play. Whereas it is being claimed that Intel has one thing deliberate to counter the dominance of X3D chips in gaming, simply including extra cache is not the answer—the latency actually must be minimize proper down.
Intel’s older Raptor Lake chips (i.e. Core thirteenth/14th Gen) are higher than Arrow Lake (Core Extremely 200S) in video games, as a result of they will clock a lot greater, particularly the ring bus. The draw back is that they’ll suck up as a lot energy as you may throw at them within the course of and Intel needed to do one thing about that, therefore why Arrow Lake’s clock speeds and energy consumption are much more conservative.
I’ve spent lots of time over the yr messing about with voltages, clocks, and varied timings with a Core Extremely 9 285K and Extremely 7 265K to attempt to enhance how nicely the chips can fling information round. I can get some fairly respectable throughput figures, however the entry latency for L3 by no means actually improves sufficient to make any headway in opposition to a Ryzen 7 9800X3D.
Nova Lake might be an absolute behemoth in multi-threaded purposes, particularly these used within the skilled sector, however video games are nonetheless more than pleased with 12 to 16 threads. That is as a result of all the massive releases are developed for consoles, which solely provide that many threads to builders. What they need is a very good cache construction, with quick, low-latency information entry.
If Intel’s transfer to intention for “management gaming efficiency” consists completely of simply shoehorning a buckload of cores into its subsequent technology of desktop processors, I believe that AMD will not have a lot to fret about.
Greatest CPU for gaming 2025