Friday, October 30th 2026
Intel Nova Lake-S Promises Up To 17% Single-Thread and 20% Gaming Performance Improvements
A lot of information relating to Intel Nova Lake-S has emerged leading up to the official launch in early 2027, including official information from Intel detailing the Nova Lake architecture and a recent benchmark leak that promised significant single-thread performance improvements on a 28-core Nova Lake-S part. However, a recent leak out of @Haze2K1 on X, shared by VideoCardz, purports that the real-world performance improvements for Intel's Nova Lake desktop CPUs are in the double digit range.
While the leaker has been around for a while, there is no guarantee that this information is accurate, since the only cited source is "circulating chatter," however, the claim is that Intel's Nova Lake will outperform comparable last-generation Arrow Lake CPUs by 8-17% in raw single-thread performance, 15% in general-purpose workloads, and up to 20% in gaming. There is some nuance to these claims, with the 17% improvement figure exclusive to the higher-end CPUs that feature bLLC, while the 17% boost allegedly comes from the implementation of Intel's new Advanced Performance Extensions.
Some of that performance improvement will also come from memory controller improvements, since Nova Lake is meant to use DDR5-8000 memory out of the box, where Arrow Lake was relegated to 6400 MT/s. How this all shakes out, however, remains to be seen, since Intel has only promised more information at CES in January 2027.
Sources:
VideoCardz, Haze on X
While the leaker has been around for a while, there is no guarantee that this information is accurate, since the only cited source is "circulating chatter," however, the claim is that Intel's Nova Lake will outperform comparable last-generation Arrow Lake CPUs by 8-17% in raw single-thread performance, 15% in general-purpose workloads, and up to 20% in gaming. There is some nuance to these claims, with the 17% improvement figure exclusive to the higher-end CPUs that feature bLLC, while the 17% boost allegedly comes from the implementation of Intel's new Advanced Performance Extensions.
54 Comments on Intel Nova Lake-S Promises Up To 17% Single-Thread and 20% Gaming Performance Improvements
I guess we'll see, but Intel's big comeback couldn't be at a worse time. Prices are discouraging, and PC sales are way down as well. Throw in DDR5-8000, and this will be no cheap upgrade for buyers looking for a significant boost.
Also I don't ever recall anyone doing a test where using Performance upscaling presets, if they affect CPU scaling performance in same way as just lowering native resolution. Because with FSR4 I now regularly use Performance preset as it's so good quality I can't even notice the difference in games like Marvel Rivals and Overwatch. Even DeadZone Rogue apart from some more noise in reflections due to lower internal resolution, but the rest looks so good. Internal resolution is then around 1080p which makes comparison a bit more realistic then.
@MufasaV
Thing with Zen 6 is it can be socketed in existing AM5 boards and has a big head start considering how performant Zen5 already is. Biggest advantage will be AM5 because good luck buying a pair of DDR5 8000 MT sticks in this broken economy...
720p ---> 4K Ultra perf
720p ---> 1440 Perf
720p ---> 1080p Quality
Look for some recently published game setting:
Recommended .. targeting 1440p@60fps with Upscaling
Edit: examples where 720p is used
Why are you not ordering Intel CPUs now :roll:
I gotta say I love articles like these because all they do is underline you don't need to be touching anything at all for the foreseeable future :D
- 15% IPC
- 10-15% clock boost 6.2-7.0ghz
- not even counting the X3D version
The purpose of a CPU benchmark is to benchmark the CPU, so the lowest graphical settings are used to remove the GPUs influence. Do we need to get out the crayons to explain this to you? This has been explained to death yet there are still those who complain about low rez testing because "WeLl NoBoDy PlAyS aT tHoSe SeTtInGs".
The percentage of people using a resolution has no bearing on its utility to test performance. If we were to follow your Looney tunes logic, the it's 4k that should be dropped, as more people play at aub 1080p then they do 4k. Funny how you don't advocate for that......hmmmm......Going from 20 percent below the 9800x3d to 10 percent ahead isn't enough?
Oh how spoiled we have become.....
techpowerup:
why is the original leak interpreted like this, It looks full of errors and random interpretations, which is very unprofessional
Are you saying Zen 6 wont be better than Zen 5, then that's blatantly false. Even Zen 5 was still better than Zen 4, even at mere 5% in games.
If you're saying that Zen 6 wont have a higher uplift than NVL then that's irrelevant because it doesn't need to, to main parity or advantage. Zen 5 is already ahead of Intel. If Intel manages to beat Zen 5 X3D by 10% then Zen 6 doesn't need to improve by 30%. Just 10%. And that's the regular version. Zen 6 X3D will come later with even more gains.You're assuming the leaked data is also not based on 720p tests?
Because i didn't see mentioned in the leaks what game or benchmark these percentages were derived from.
I very much dounbt it's 1440p or 4K testing where NVL is suddenly 30% faster than ARL.Uphill battle for Intel. Besting two year old Zen 5 X3D by 10% with a supposedly most advanced (and expensive) node is not exactly the win some people think it is. Add to that a new platform people need to buy. They'll have a hell of a time convincing existing DDR4 Intel system users to upgrade. LGA1851 users could upgrade by swapping the board, but it's still a hassle and there arent many of them to begin with.
And with Sea of wires and hopefully the FCLK bottleneck removed on current systems.
DDR5-6000 provides a theoretical maximum bandwidth of 96.0 GB/s, which translates to an actual real-world bandwidth of roughly 84 to 88 GB/s in synthetic tests like AIDA64.
DDR5-8000 provides a theoretical maximum bandwidth of 128.0 GB/s, which translates to an actual real-world bandwidth of roughly 115 to 125 GB/s in synthetic tests like AIDA64.
So even without DDR5 8000 will be a nice boost here from the current 64 GB/s limit on a single CCD
Then add more clock speed i'm thinking 6.2-6.5 Fmax Clock and other arch changes I think we will be eating good.
Intel probably have figures of how much they expect in gains, but it will vary a lot. At launch we'll probably see very few productive applications/tools with APX support.