Nvidia GeForce RTX 5090 Faces Setbacks in Benchmark Performance
As gamers eagerly await the release of the Nvidia GeForce RTX 5070, the RTX 5090, which was initially seen as the top performer, is facing some setbacks. After briefly surpassing the RTX 4090 in PassMark’s G3D Mark benchmark earlier this month, the RTX 5090 has now dropped back into second place, trailing behind the RTX 4090. Despite early reports suggesting a small advantage of +2.85% for the newer Blackwell-based RTX 5090 over the Ada Lovelace-powered RTX 4090, further testing has eliminated this advantage.
RTX 5090 Drops Performance in Benchmark Testing
After a larger pool of samples was tested, the average score for the RTX 4090 rose slightly, from 38,422 to 38,436 (based on 14,422 samples), while the average score for the RTX 5090 dropped significantly, from 39,516 to 38,048 (based on 50 samples). This marked a -3.71% loss for the RTX 5090, and it now lags behind the RTX 4090. The fluctuation in performance suggests that the RTX 5090 may still have the potential to reclaim the top spot once more samples are tested and any issues are addressed.
Potential Causes for the Performance Gap
Several factors could explain the RTX 5090’s performance slump. For one, driver issues have been reported, with some RTX 5090 cards failing to function properly after software updates. The smaller sample size for the RTX 5090 (50 samples compared to 14,422 for the RTX 4090) means there is still room for improvement and more accurate results. Additionally, hardware problems have surfaced, including the discovery that some RTX 5090 boards are missing Raster Output Pipelines (ROPs)—crucial components for image rendering. These issues, along with low unit availability, may also contribute to the underperformance seen in benchmarks.
RTX 5090’s Strengths in Certain Tests
Despite these setbacks, the RTX 5090 still outperforms the RTX 4090 in some areas, particularly in DirectX 12. In this test, the RTX 5090 showed a significant advantage, with a +39.3% increase in average frames per second (209 vs. 150). However, the RTX 5090 falters in the GPU compute benchmark, where it lags behind the RTX 4090 by -17.5%, despite having 32.8% more CUDA cores. CUDA cores are essential for parallel processing tasks, and the underperformance in this area raises questions about the RTX 5090’s efficiency.
Uncertain Future for RTX 5090
It’s still early to draw conclusions about the RTX 5090’s overall performance, and it’s possible that Nvidia will resolve these issues through software updates or hardware revisions. However, these mixed results may cause potential buyers to reconsider purchasing such an expensive product, especially if performance inconsistencies continue. With the upcoming RTX 5070 still on the horizon, gamers will likely be watching these developments closely to see if the RTX 5090 can overcome its current challenges and reclaim its position at the top.
