RTX 4070 performs at the same speed as GTX 1080
RTX 4070 performs at the same speed as GTX 1080
still trying isn't harmful, yet I recall many comparable situations—often people assumed it wasn't the CPU, but ultimately it was the processor that was the issue.
I don’t believe there are any titles completely limited by an 8700k paired with a GTX 1080. It doesn’t matter whether the 4070 causes some bottlenecking or not. I just mention a few to keep the possibility open for another option. Otherwise, there would be games where the 1080 Ti matched the 1080 perfectly. The OP should redo the test and specify what they’re checking so we can draw solid conclusions. (Sorry about the image quality)
Perhaps the memory issue or Vsync affected performance. On my Phenom II 1100T with DDR3, the RTX 3060 struggled more than the Quadro K4200 (GTX 670 equivalent). Even at low settings, stuttering occurred without strain on CPU or GPU, forcing me to rely on RAM. The 3060 works in my upgraded build with DDR4 2133. I played Metro Exodus fine on my older 1100T/K4200 setup—despite being below specs, it handled it well under medium settings (WUXGA). It’s better to maximize performance from what you have. Only games compatible with Windows 7 can run Cyberpunk due to missing SSE 4.1/4.2 support.
At 1080p resolution, frame rates hovered between 240 and 280 FPS on both cards. Running GPU used full capacity at 1080 but only 40% utilization was seen with the 4070.
provided additional details to assist you, ensuring no frame limit or v-sync restrictions are applied.
I included the references, please find them useful. The images you shared provide great clarity.
Ram is a 2666MHz DDR4 chip and I have 64GB of it installed.
Your preferences are affecting the results. But this is clearly a CPU limitation. Your 4070 handles the 1080 workload well. This is somewhat normal since ADA/Ampere drivers are heavier for the CPU. You did notice better GPU performance with the 4070, but that’s expected. I appreciate your benchmarks—they accurately illustrate how bottlenecks function. However, some might misinterpret them. The 1080 run was actually 50% CPU constrained. About half the frames were CPU-bound, half GPU-bound. Standard deviations are similar. The main reason you didn’t see more gains was the heavier drivers, which caused a noticeable drop in performance when matched to your game settings. Those shadows are costly because z-maps can duplicate scenes, but at medium quality they barely impact results. LOD, shading, reflections, and AO still use GPU cores, so lowering them helps boost FPS and reduce GPU load.