The much-awaited NVIDIA RTX 5090 is finally here, and tech enthusiasts have been eagerly analyzing its performance benchmarks. While the GPU offers improvements over its predecessor, the RTX 4090, the gains appear to be more incremental than revolutionary. Let’s dive into some of the key benchmark results and how the RTX 5090 stacks up against other GPUs.
Performance in Synthetic Benchmarks
The RTX 5090 has shown notable improvements in synthetic benchmarks. Here are some key results:
- Geekbench 5 (OpenCL): The RTX 5090 scored 367,740 points, a 15% increase over the RTX 4090.
- Geekbench 5 (Vulkan): The card achieved 359,742 points, 37% higher than its predecessor.
- CUDA API Test: In this test, the RTX 5090 notched up 542,157 points, marking a 27% gain.
- Blender 3.6.0: A median score of 17,822.17, approximately 36% faster than the RTX 4090’s score of 13,064.17.
While these numbers are impressive, they have sparked discussions about whether they meet the expectations of a next-generation flagship GPU. With 32% more CUDA cores than the RTX 4090, many had anticipated a more significant leap in performance.
Real-World Gaming Benchmarks
While the synthetic benchmarks provide valuable insights, real-world gaming performance is where it matters most for enthusiasts. Here are some key takeaways from early gaming benchmarks:
- Cyberpunk 2077: With DLSS 4 enabled, the RTX 5090 delivers over 200 FPS at UHD resolutions. However, at native 4K, it struggles to hit 60 FPS in this demanding title.
- Star Wars: Outlaws: DLSS 4 boosts framerates to 250 FPS with reduced latency.
- Black Myth: Wukong and Alan Wake 2: Similar performance gains were observed, with DLSS 4 making a significant difference.
Comparing the RTX 5090 with Other GPUs
Let’s compare the RTX 5090 with some of its notable competitors:
| GPU | Architecture | CUDA Cores | Memory | Boost Clock (MHz) | Memory Bandwidth (GB/s) | Synthetic Benchmark Scores |
|---|---|---|---|---|---|---|
| RTX 5090 | Blackwell | 21,760 | 32 GB GDDR7 | 2,407 | 1,792 | 367,740 (Geekbench 5, OpenCL) |
| RTX 4090 | Ada Lovelace | 16,384 | 24 GB GDDR6X | 2,520 | 1,008 | 319,470 (Geekbench 5, OpenCL) |
| RTX 3080 | Ampere | 8,704 | 10 GB GDDR6X | 1,710 | 760 | 232,156 (Geekbench 5, OpenCL) |
| AMD Radeon RX 7900 XTX | RDNA 3 | 12,228 | 24 GB GDDR6 | 2,450 | 960 | Not Available |
Key Takeaways
- Performance Improvements: The RTX 5090 offers a considerable performance improvement over the RTX 4090 and RTX 3080. Its higher CUDA core count, faster memory, and improved architecture make it a powerhouse GPU.
- Gaming Experience: With DLSS 4, the RTX 5090 shows significant improvements in gaming performance, especially at UHD resolutions. However, in some demanding titles, achieving high framerates at native 4K remains challenging.
- Value Proposition: While the performance gains are substantial, the question of value remains. The RTX 5090’s price point might make it a more suitable choice for those seeking cutting-edge technology and performance for future 8K applications.
Are you excited about the RTX 5090, or are you considering other options? Let me know what you think!
