GeForce RTX 5060 Ti PCIe Lanes Problem: Do You Have This Issue?

While testing the GeForce RTX 5060 Ti, I encountered a recurring issue with active PCIe lanes. On PCIe 5.0 motherboards, the card often fails to run in x8 mode even in a x16 slot, starting instead in slower x1, x2, or x4 modes. The same behavior occurred across three RTX 5060 Ti models from MSI, Gainward, and Asus. The issue likely stems from how these motherboards handle compatibility with PCIe 5.0 graphics cards using eight lanes.

The GeForce RTX 5060 Ti and RTX 5060 are the first models in the GeForce RTX 50 series with a PCI Express interface reduced to eight lanes. More expensive and powerful GeForce cards already use PCIe 5.0 with the full 16 lanes. It’s similar to the budget models from the previous generation, where the RTX 4060 Ti and RTX 4060 also had only eight active lanes. Even so, the RTX 5060 Ti’s bandwidth can match previous-gen cards with 16 lanes because RTX 40 only supported the older PCIe 4.0 standard, which has half the bandwidth.

That’s assuming everything works as it should. But during RTX 5060 Ti testing, I found that the card sometimes started with eight lanes, sometimes with four, two, or even just one.

Troubleshooting was tricky because after shutting down the system, removing the card, and reinstalling it, the number of PCIe lanes it initialized with kept changing. It seemed to depend on whether the card was supported and on the precise positioning of the bracket in the case’s mounting holes.

This led me to initially suspect the contacts in the PCI Express slot. The PCIe 5.0 connector on the cards is physically different from older versions—the contact pads on the card are slightly shorter in the new design.

Because both cases and boards allow some wiggle room during installation, one of my “investigative theories” was that a pin might not be making good contact due to imperfect seating.

Problem with the slot?

Although I’ve only been using this board for three months, about 20 cards have already gone through the slot, most of them installed multiple times. Some weighed nearly three kilograms. While I do use a support mount for heavier cards, I still wasn’t sure whether this kind of usage might have damaged one of the contacts. I’ve never experienced such problematic behavior with previous PCIe generations, even after hundreds of installations.

I didn’t immediately suspect an issue between the motherboard and the RTX 5060 Ti because, during roughly 200 PC startups with higher-end RTX 50 models, it was very rare for them to initialize with fewer lanes. The problem seemed to go away after reseating the card—or so I thought. In reality, simply powering the system off and on again may have fixed it during reinitialization.

Eventually, I brought out the camera. Even macro photos from different angles showed no signs of physical damage or excessive wear on the slot contacts.

Another reason physical damage seemed unlikely was that when I tried inserting a Radeon RX 9070 into the same slot, it booted without any issues and negotiated the full number of lanes.

I then contacted Nvidia and Gigabyte support, but neither had encountered a similar issue with the RTX 5060 Ti, which suggested the problem might be on my end. They gave me a few suggestions—updating the BIOS and toggling ReBAR on and off. I tried the ReBAR toggle, but decided to hold off on the BIOS update, since the version installed wasn’t that old. I also had a second processor on the way, which I needed to build a duplicate setup with the same components.

In the meantime, I tried a general workaround that had reportedly helped other users with earlier RTX 50 stability issues—forcing a lower PCIe version. The result was pretty much the same: the number of active lanes still varied randomly, just under the PCIe 4.0 standard. I didn’t rule out a BIOS issue, but left BIOS experiments for the second build, where it was unlikely that the slot would be damaged.

Second build, same issues

I definitively ruled out slot damage or poor contact after my second Ryzen 7 9800X3D arrived, allowing me to build a second system using the same motherboard model. All the components were identical in terms of model, but each was a brand-new unit.

I first flashed the same older BIOS version, F3j, to the new motherboard to match the setup of the original board. This was the last version available before RTX 5090 testing. Despite the RTX 5060 Ti being the first GPU I installed, the result was the same – the Gainward, MSI, and Asus models all initialized usually with fewer than eight lanes. I then updated to the latest BIOS (F4), reset BIOS to defaults, with clean-installed Windows, used the latest chipset drivers, and checked whether the integrated GPU being active or not affected the RTX 5060 Ti’s behavior. Nothing helped.

Unlike the first test build, which is housed in a traditional case, the second build was set up on an open bench. So I ruled out case-related variables or mounting issues, as well as the fact that most reviewers test horizontally with vertically-inserted GPUs.

Full number of lanes after restarts

Even before the RTX 5060 Ti’s launch, I noticed that rebooting the PC often resulted in more lanes being active than after a cold boot. Full slot speed could always be achieved with a few restarts.

But this wasn’t 100% reliable – during six days of testing, I once saw the lane count drop from eight to four, despite confirming slot speed before the test.

After the card’s launch, I searched for similar cases but found nothing. I checked through about 30 reviews. Only a few included GPU-Z screenshots showing the operating mode. From roughly ten screenshots, two showed a reduced number of active lanes. One used the same motherboard but a different processor (suggesting the board may be the issue), while the other was on a different AMD AM5 board from Asus. All other systems – mostly with Asus boards – ran cards at full lanes.

Still, that doesn’t guarantee those boards had no issues. If a screenshot is taken right after installation, a reboot may already have occurred, allowing the card to start at full lanes.

At this point, it was clear that my X870 Aorus Elite WiFi7 motherboard had an issue with detecting or initializing the number of active GPU lanes at startup, and that BIOS updates claiming improved PCIe compatibility weren’t solving it.

Different systems, different behavior

In addition to the two identical systems with the Aorus X870E and Ryzen 7 9800X3D, I tested two other setups with PCIe 5.0-compatible CPUs and motherboards – AMD Ryzen 9 7950X with an Asus ROG Strix B650E-E Gaming WiFi, and Intel Core i7-13700K with an Asus Prime Z690-P WiFi.

The Ryzen 9 7950X system had a BIOS updated in December. After about ten cold boots, the card always started with full lanes. The Intel setup hadn’t been used in a while. Its BIOS dated from late 2022, and the card booted with just one active lane. Multiple restarts didn’t help. But after a BIOS update, it began to start with all eight lanes. That doesn’t guarantee permanent success – I only ran ten cold boots across a few cards, not hundreds like in the test bench.

I also encountered another issue with the noise measurement setup. I use an older motherboard with a Core i5–8600K and Gigabyte Z370 HD3P. In this system, the card didn’t boot at all – the PC failed the Power-on Self-test and froze on a black screen. Previous RTX 50 models with full 16 lanes had no problems, including the RTX 5090, RTX 5080, RTX 5070 Ti, and Radeons RX 9070 XT and RX 9070. The fix was to remove the RTX 5060 Ti and flash the latest BIOS with ReBAR support – after that, the RTX 5060 Ti booted without issues.

How to verify whether you have the same issue will be covered in the next section.