The GeForce RTX 5090, a new flagship GPU powered by NVIDIA’s “Blackwell” architecture that features fifth-generation Tensor Cores and fourth-generation RT Cores, will deliver 20% improved performance on average versus its Ada-based predecessor, the GeForce RTX 4090, according to a set of alleged benchmark scores stemming from “a few reviewers” that surfaced online for the GPU today ahead of official reviews, which are expected to roll out later this week.
“On average, gamers can expect about a 20% performance improvement over the RTX
4090, according to reviewers we spoke with,” reads a report. “In synthetic tests, these figures are higher: 36% FireStrike (Performance/Extreme/UItra), 33% TimeSpy (Performance/Extreme) and 46% for other tests (Speed Way, Port Royal, Steel Nomad).”
“To ensure this data is as comparable as possible, we are using official 3DMark scores for the 4090, 4080S, and 7900XTX,” it continues. “The score below for the RTX 5090 represents the average of three reviews involving the latest Intel and AMD setups. Additionally, we always use Graphics scores when available, though newer benchmarks no longer separate Overall and Graphics scores.”
“The GeForce RTX 5090 GPU — the fastest GeForce RTX GPU to date — features 92 billion transistors, providing over 3,352 trillion AI operations per second (TOPS) of computing power,” NVIDIA has said of its latest flagship GPU for gamers and creators. “Blackwell architecture innovations and DLSS 4 mean the GeForce RTX 5090 GPU outperforms the GeForce RTX 4090 GPU by up to 2x.”