NVIDIA is reportedly targeting a fourth-quarter release for its RTX 50 GPU series. While the data-center Blackwell GPU has been unveiled, its release date remains unconfirmed. The spotlight is now on the highly anticipated GeForce RTX 50 series, particularly among gamers.
In January, NVIDIA unveiled three new GeForce RTX 40 SUPER graphics cards, offering a mid-product refresh for high-end enthusiast-class GPUs. Presently, the RTX 40 series comprises 12 models, fewer than the RTX 30 lineup which had 20 models. However, excluding the LHR (Lite Hash Rate) series, the number of SKUs is roughly equivalent. This indicates that NVIDIA might have finalized the series by now.
As reported by Money UDN, board partners are expecting the release of high-end RTX 50 models in the fourth quarter, with particular attention on the RTX 5080 and RTX 5090 cards. It’s important to acknowledge that while these names are likely accurate, NVIDIA is still in the process of finalizing the specifications. The company is known for making last-minute adjustments or even canceling products before their market debut.
A prominent NVIDIA insider, highly regarded for accurate leaks, concurs that the RTX 50 launch is slated for this year. Kopite initially speculated about Blackwell featuring GDDR7 memory and hinted at a 512-bit memory bus for the largest GB202 GPU.
At the end of the year.
— kopite7kimi (@kopite7kimi) April 4, 2024
Speculations indicate that the GB202 GPU might include up to 192 Streaming Multiprocessors, whereas GB203 could have half that number (96 SMs). If NVIDIA maintains the current configuration for its FP32 (CUDA) cores, this would amount to 24,576 and 12,888 cores, respectively. However, to ensure favorable yields, it’s anticipated that neither the RTX 5090 nor the 5080 will feature these full configurations.
Source: UDN