Checking if all e-cores problems are resolved on Intel 12th–14th Gen and being open to hybrid solutions
Checking if all e-cores problems are resolved on Intel 12th–14th Gen and being open to hybrid solutions
When examining the spoiler from my earlier message, it becomes clear that workload is spread evenly among all cores and threads. Each thread typically operates near 50%, meaning a configuration with six cores or twelve threads would handle nearly all processing while maintaining similar efficiency. After completing my tasks, I can turn off half of the cores in the BIOS and compare benchmark results in both setups. The key benefit of E-cores lies not in having more cores, but in allowing the operating system to shift less critical background processes onto E-cores. This frees up resources on P-cores, enabling the game to utilize a full P-core where it might have previously shared it. Consoles usually run minimal background operations—such as antivirus scans—and their OS is optimized for gaming, making them capable of similar or better performance with fewer resources (fewer or slower cores). Their hardware is also well-documented, allowing game engines and settings to be fine-tuned for specific configurations. Creating a multi-threaded engine isn’t simple; it requires thoughtful design from the start. Certain elements like audio, music, and networking can be moved to other threads with ease, but the main rendering thread remains a bottleneck. Until recently, splitting this work across threads was limited by older graphics APIs (like DX12/Vulkan), which only permitted one thread to communicate with the graphics context. For Unreal Engine, this changed significantly with Unreal Engine 5. There’s a clear distinction between engines that rely on extra threads for specific tasks versus those that can dynamically balance workloads across all available cores. Even then, most games built on top still need to be optimized to leverage these capabilities fully. Other engines may have reached this level sooner, but Unreal remains the most prominent example.
It offers a significant boost in productivity. I recently upgraded my home server from a 12400 to a 14700K to conserve storage space. By keeping thumbnail caches in RAM and clearing them regularly, I save valuable space. With 28 threads handling thumbnail tasks, loading a page with 100 images is noticeably slower than fetching them online. Currently, I only have a 12700K cooler, which causes thermal throttling to around 4Ghz. This setup is extremely impractical for most users, but since only two of us access it, it provides a major storage advantage for my needs. Despite the high power consumption of these Intel CPUs, they perform well at running at 4Ghz on both P and E cores.
That still works even with a somewhat minimal WIN1o or WIN11 setup. It’s not necessarily a completely simple installation, but it still allows basic AV, system info, and MSI Afterburner to function. However, phone home telemetry stops, updates turn off, and other services are disabled. The performance difference remains noticeable compared to heavier OSes or systems that rely on e-cores only for background tasks like downloading updates. I think the benefits of e-cores would likely be even greater with more processor cores, making them more effective for demanding background operations.
E-cores offer better power efficiency than P-cores. They work well for tasks that use many threads without requiring high performance per thread. The same efficiency gains could be achieved with more cores, but power consumption would rise unnecessarily. Games, on the other hand, don’t need many threads but must run as quickly as possible.
Early discussions often claimed P-cores handle gaming while E-cores manage background tasks, but this isn't accurate. In reality, E-cores don't follow that pattern and might even run on E-cores when they shouldn't, hurting performance. Even AMD chips like the 7950 with genuine 16 cores don't operate this way—they allocate cores differently, preventing some games from using the wrong set. High-core CPUs like the 7950X tend to stutter during background work, regardless of core count. Hardware Unboxed has covered this extensively. The actual impact comes from overall CPU performance, not just core numbers. What truly matters is how well the CPU handles tasks, not just how many cores it has. The idea that more cores always mean better gaming is flawed; cache size and architecture play bigger roles. AMD designs like the 7800X3D with ample cache often outperform higher-core models despite faster single-thread speeds. If you need the top gaming CPU, go with the 7800X3D. Its superior performance will likely keep up with Intel in current titles, and by the time 8-core CPUs become less competitive, these models will be outdated.
The challenge with hybrid cores lies in how effectively the operating system scheduler assigns tasks. Adding more cores doesn’t simplify the process much. @Wolverine2349 Ran some benchmarks for Cyberpunk 2077. It wasn’t a rigorous study, as I didn’t repeat tests for each setup. I restarted the game between runs because I wasn’t sure how quickly it would handle core changes. All tests used an RX 6600 at 1440p on the latest Manjaro (AMD Radeon RX 6600, version 16.0.6 with AMD tools). I toggled cores on and off via the command line. You’ll notice more cores generally help, but the gain is modest. The performance jump between 4C and 12C was under 8%, while 6C to 8C improved by about 1%. Cyberpunk seems to perform better in this case. Checking the System Monitor shows the workload spreads evenly across cores and threads, indicating solid multi-threading capabilities. In summary, 4C/8T averaged 86.4%, 6C/12T was 91.6%, and 8C/16T reached 92.6%. All stayed between 60% and 100%, with most hitting around 80-100%. The engine handles multi-threading well overall.