I've been trying to figure this out for years. My RTX 2060 scores around 14,000 on PassMark and delivers normal performance in several AMD systems, but the same card scores roughly 4,500 when installed in my 8th-generation Intel computers. In practice, it gets about 30 FPS in benchmarks instead of the roughly 245 FPS I see on AMD hardware.
I've tested multiple Intel motherboards, CPUs, and power supplies, including an i5-8500T and an i7-8700, with the same result. The card reports similar maximum clocks in both systems—about 1995 MHz on the core and 7001 MHz on the memory—and I've already tried changing BIOS settings, resetting the BIOS, and enabling ReBAR. Reinstalling Windows did not help either.
The problem seemed to appear after updating to Windows 11 version 24H2; the system worked normally on 23H2. Is there a known compatibility or driver issue, or is something else limiting the GPU on these Intel systems? I'd like to reuse the PC as a spare machine for friends rather than buy more parts.
4 Answers
The i7-8700 is older than the newer AMD systems, but it should still run an RTX 2060 reasonably well, especially at 1080p. Memory speed, CPU limits, BIOS microcode updates, and security mitigations can affect benchmark results, but they normally would not explain a drop from roughly 14,000 to 4,500 by themselves. I would focus on finding what differs between the working 23H2 setup and the failing 24H2 setup before considering overclocking or replacing hardware.
The Intel platform itself should be compatible with an RTX 2060, so this probably isn’t a fundamental CPU/GPU incompatibility. Compare the systems with GPU-Z or HWiNFO while running the same benchmark. Check GPU power draw, core and memory clocks under load, PCIe link speed and width, GPU utilization, CPU utilization, temperatures, and whether the card is hitting a power or thermal limit. Matching reported maximum clocks does not necessarily mean the card is sustaining those clocks throughout the test.
Overclocking the 8th-generation CPU or increasing RAM speed might improve a CPU-limited result, but it should not be necessary for basic RTX 2060 compatibility. A large GPU benchmark gap points more toward a software, PCIe configuration, power-management, or driver issue. Since reinstalling Windows did not change anything, compare BIOS versions and chipset drivers as well, and test with unnecessary overlays and background applications disabled.
If the GPU is only reaching about 30 FPS while showing normal clocks, look at utilization and frame-time data rather than PassMark alone. A CPU, driver, PCIe-link, background-process, or Windows power-management problem could be holding it back. Also verify that the card is running at PCIe x16 rather than a reduced link width, and try a clean graphics-driver installation with the same driver version on both platforms.
That makes sense. I’ve built and supported a lot of PCs, which is why this has been so confusing. The Intel system worked correctly on 23H2, and the problem appeared after 24H2. I’ll compare it again in HWiNFO, especially GPU utilization, PCIe link state, power draw, and CPU behavior, before writing the machine off.