NVIDIA GeForce RTX 5070 Ti Unveiled With Reduced ROPs; Issues Now Affecting Other Premium RTX 50 Series GPUs

The NVIDIA GeForce RTX 5070 Ti is reportedly facing the same issue as its higher-end counterpart, the RTX 5090, by exhibiting a deficiency in Render Output Units (ROPs).Instead of the advertised number, many units are being found with a reduced ROP count.

Uncovering the RTX 5070 Ti: 88 ROPs Instead of 96

NVIDIA’s credibility is taking a hit as it has been confirmed that a fraction of their RTX 5070 Ti and RTX 5090/5090D graphics cards come with fewer ROPs than what was specified. Specifically, reports indicate that while some units show 88 ROPs, NVIDIA’s official specifications listed 96 ROPs. This discrepancy raises significant concerns about the transparency of NVIDIA’s product offerings, particularly when it impacts performance metrics that gamers rely on.

Despite NVIDIA’s claims that only 0.5% of the produced GPUs suffered from this issue, there’s a growing sense of skepticism in the community. A post on Facebook, shared by a user, showcased the GPU-Z screenshot which confirms the presence of these diminished ROPs in the market. This raises the question of how many consumers may unknowingly purchase these GPUs.

RTX 5070 Ti reduced ROPs — Image Credit: @GawroskiT

Given this reduction, performance is bound to suffer. NVIDIA suggests that the average performance drop is around 4%, but this can vary significantly based on individual game requirements and performance benchmarks. Users are spending substantial amounts for these GPUs, often exceeding the expected retail price due to decreased inventory levels. Thus, any decline in performance can feel more pronounced.

RTX 5090 Unboxing — Image Credits: NVIDIA

Diving deeper into the specifications, the drop in ROPs has resulted in a marked decrease in Pixel Fillrate, directly affecting graphical performance. The performance of the RTX 5070 Ti with 96 ROPs is expected to deliver around 287.7 GPixel/s, whereas the affected versions only manage to output 223.7 GPixel/s. Although the boost clock speed on these compromised models is set at 2.99 GHz, which can push their Pixel Fillrate to around 263.12 GPixel/s, it still results in a considerable loss of about 9%.

Even though NVIDIA’s figures indicate that the faulty units are a mere fraction, the total number of affected cards can quickly increase into the thousands, raising further concerns about the reliability and transparency of NVIDIA’s quality control processes in recent times.

Source & Images