Adjusting E-Cores in Alder Lake can enhance gaming speed.
Adjusting E-Cores in Alder Lake can enhance gaming speed.
Hey everyone! It seems there were no reviewers checking how disabling efficiency cores impacts performance. That’s why I went ahead and conducted my own mini trials. I’m excited to share what I found! I think these outcomes are worth exploring further with more games. My setup includes an i7-12700KF, RTX 3080, DDR5 5200 MHz, and Asus Z690 Prime-A. I used built-in benchmarks for each title: Arma 3, Rainbow Six Siege, F1 2018, and GTA V. Here’s a quick summary of my observations (mirror link provided). All tests were at 1080p, with each benchmark run three times.
In E-Off mode, turning off efficiency cores and enabling AVX512 through the BIOS led to the best results for Arma 3, with six siege games showing a slight but noticeable boost. Removing E-cores didn’t hurt performance in any case (still a small sample). W11 didn’t perform well, and using it caused issues—like crashes and BSODs. F1 had one crash and a random error message. GTA V needed manual settings tweaks to work properly.
Some games benefited more: Death Stranding showed improvement after testing with CapFrameX, while Arma 3, GTA V, and F1 performed similarly. E-cores off with AVX-512 also maintained stable frame rates.
My motherboard doesn’t have the legacy game compatibility mode in BIOS yet. Once it’s available, I’ll definitely give it another try! Would love to see more analysis from a journalist or deeper testing. Thanks for checking!
We observed some intriguing outcomes, but have you checked the CPU speed as well?
It kind of aligns, since the 11 (CPU scheduler) isn't fully developed yet?
Turning off E cores usually boosts performance when your workload doesn’t require more than 16 threads or 8 physical cores. When you disable E cores, P cores get access to the E core L3 cache exclusively. This change often makes a difference. Disabling hyperthreading on a 10900k lets physical cores use the extra L3 cache that logical cores previously accessed. For multitasking—such as streaming and gaming simultaneously—enabling E cores might be better. Also, with C-states active, loading certain programs can cause odd behavior; sometimes P cores enter sleep mode briefly. I noticed this with Stockfish: when hwinfo sensors were running, the engine paused instead of starting, but with C-states off, P cores resumed normal use right away.
These outcomes are fascinating. Thank you for providing them. My observations here include: All these titles are from earlier releases, which usually indicates lower threading capabilities. Even F1 2018 was likely developed before the 8th-gen Intel architecture and probably finished coding before Ryzen captured a significant market share to fine-tune for more than four cores. It’s perfectly fine to run benchmarks on such games since they remain popular, but it would be helpful to see data from titles that heavily utilize multiple cores—like Shadow of the Tomb Raider or the well-known Ashes of the Singularity—to determine if they gain performance advantages with E cores. Turning off E cores might actually boost P core performance or alter the boost curve. Have you checked the frequency settings? I’m not clear on the available documentation for 12th-gen behavior, but disabling E cores could enable P cores to reach a higher overall frequency. Did you test Windows 11 with E cores turned off? It’s designed for 12th-gen optimization, yet I’d like to know if its scheduler introduces any delays with hybrid cores. Perhaps the scheduler isn’t mature enough and still offers benefits.
Actually, I forgot to do this. I checked and made sure the CPU wasn’t throttling using HWINFO in any of the tests. I’ll run a few more benchmarks later to confirm the clock speeds. I managed to fit in additional runs before heading to work. I’ll try GTA V soon and update the results. Win 11 + E-Off: Arma 3: Avg 50.07 Six Siege: Min 355, Avg 439, Max 552 F1 2018: Min 208.33, Avg 269.67, Max 337.67 GTA V: I’ll run it soon! Arma 3 showed a small drop, Six Siege improved slightly (still not winning 10 Ecore off levels), F1 had a minor decline (likely within error range). I’ll check what modern games I can borrow from friends. I’d love to see a proper journalist do a 20+ game benchmark too.
CapframeX evaluated performance at 20MB and 30MB L3 with E cores turned off, including 2 P cores on the 12900k. This shows a noticeable boost in FPS from simply disabling E-cores—more L3 available. https://www.capframex.com/tests/Alder Lake-S 12900K ist der neue Gaming König
Win 11 seems to handle caching differently, possibly using fewer resources than Win 10. Your observations match some user reports about performance changes.
Additional testing has been conducted. P-cores reached 4.7 ghz during benchmarks for both E-cores On and Off, with fluctuations in some games. When E-cores were off and AVX-512 disabled, performance remained consistent. The motherboard lacks the legacy game compatibility mode in BIOS—once available, I’ll check it. A Reddit comment noted CapFrameX observed similar behavior in a death stranding scenario. If cache sharing is confirmed, Intel may face legal challenges similar to AMD’s case regarding shared cache affecting core classification.