The groundbreaking GeForce RTX 5090 launched on January 6, 2025, powered by the cutting-edge GB202-300-A1 GPU built on TSMC’s ultra-efficient 5nm process. This powerhouse delivers staggering specs: 21,760 CUDA cores, 176 raster units, 680 texture units, 170 multiprocessing streaming units, 680 tensor cores, 172 ray tracing cores, and a massive 96MB L2 cache.
Performance reaches new heights with a base clock of 2017 MHz and boost clock soaring to 2407 MHz. It’s paired with 32GB of blazing-fast GDDR7 memory on a 512-bit bus, achieving 28Gbps memory speeds and an incredible 1.8TB/s bandwidth. The card delivers 104.8 TFLOPS of FP32 performance, with a 575W power requirement and an MSRP of $1999.
Our test rig features top-tier components:
– Intel Core i9-14900K processor
– ASUS ROG Strix Z790-E Gaming motherboard
– 32GB DDR5 RAM running at 6000MHz
– The star of the show: GeForce RTX 5090 32GB
– Dual Samsung 970 EVO 1TB SSDs
– Windows 11 operating system
Benchmark results reveal an interesting finding: while the RTX 5090 shows marginal performance gains in PCIe 5.0 mode for select games, the difference is barely noticeable. Across most titles, performance remains virtually identical between PCIe 4.0 and 5.0 configurations.
The verdict? Don’t lose sleep over whether your RTX 5090 runs on PCIe 5.0 or 4.0—the real-world impact is negligible, making this a non-issue for gamers and creators alike.
Isabella Campbell 
Mason Jensen 
Gemma King