Today marks the eagerly anticipated launch of NVIDIA’s latest showstopper in the graphics card industry, the GeForce RTX 5070. This new marvel was first unveiled at CES and is now officially hitting the shelves with an MSRP of $549 US. Graphics card enthusiasts, however, might be wondering if they will actually see this card available at that price due to past experiences with similar launches, where limited models were made available, leading to price hikes.
The GeForce RTX 5070 showcases the GB205-300-A1 silicon die, containing an impressive 6144 CUDA cores. Additionally, it houses 192 Tensor Cores, 48 RT Cores, and 80 ROPS. Unlike some of its predecessors that suffered from a production defect, the RTX 5070 is expected to maintain its full ROP count. Moreover, the card is crammed with 48 MB of L2 cache and boasts a hefty 31.1 billion transistors within a die size of 263mm².
Retaining the same 12 GB VRAM on a 192-bit bus, the RTX 5070 benefits from the new, quicker GDDR7 memory. The GPU boost clock ticks up to 2512 MHz, marginally higher than the 2475 MHz of the RTX 4070. With a TDP of 250W, the RTX 5070 commands 30W more than its predecessor, the RTX 4070, and a substantial 50W more than the RTX 3070.
Promising next-generation gaming experiences through neural rendering, NVIDIA touts the RTX 5070 as one of the most advanced platforms available for complete ray tracing and AI. The card claims a performance leap of up to 1.9 times and enhanced 3D rendering through DLSS 4 MFG over the RTX 4070, with a 20% bump in traditional raster performance. More than 75 games now support DLSS 4, offering the best image quality using the company’s DLSS SR/RR Transformer model along with superior performance backing Multi-Frame Generation in top-notch titles like Avowed, Red Dead Redemption II, and Microsoft Flight Simulator.
For gamers considering an upgrade from an RTX 3070, the RTX 5070 might just be the perfect choice, delivering over three times the performance with DLSS 4 and a 65% increase in traditional raster performance. In terms of GenAI performance, it is three times faster compared to the RTX 4070, thanks to Blackwell support for FP4. Video editors will also find export times slashed radically, achieving professional-grade 4:2:2 video output six times faster on the RTX 5070 compared to the RTX 4070, with its dedicated hardware accelerated 4:2:2 support.
However, in native RT gaming (without DLSS), the RTX 5070 offers a 20% performance boost compared to the RTX 4070. Yet, without the use of MFG (Multi-Frame-Gen), it cannot rival the RTX 4090 in rasterization or ray tracing performance. With the activation of 4x MGF mode, the RTX 5070 nearly matches the RTX 4090 in speed, which is a crucial point of consideration when looking at NVIDIA’s performance metrics.
The RTX 5070 will square off against the AMD Radeon RX 9070 within the same price bracket, setting the stage for a fierce rivalry between the two graphics giants. As the gaming community waits to see how these cards perform in real-world scenarios, it’s certain that the launch of the RTX 5070 adds an exciting new chapter to the ongoing saga of high-end gaming hardware.
