NVIDIA Lineup Update: New RTX 50XX GPUs Retain Memory Configurations

Recent rumors from insider Kopite7kimi suggest that the upcoming NVIDIA GeForce RTX 50XX series graphics cards, codenamed "Blackwell," will follow the traditional memory bus schemes established by the previous RTX 40XX "Ada" generation. This implies that the new generation will maintain time-tested configurations, providing familiar characteristics to users.

The current "Ada" lineup features models with memory bus widths ranging from 128 to 384 bits, where flagships like the GeForce RTX 4090 have a 384-bit bus and can include up to 24/48 GB of video memory. The "Blackwell" series is expected to offer the following modifications:

• RTX 5090 (384 bits)
• RTX 5080 (256 bits)
• RTX 5070 (192 bits)
• RTX 5060 (128 bits)
• RTX 5050 (128 bits)

Using standard memory chips with a capacity of 16 Gb (2 GB) and a data transfer speed of 32 Gbps, the expected technical specifications would be as follows:

• RTX 5090: 384 bits / 32 Gbps / up to 24 GB memory / max bandwidth 1536 GB/s
• RTX 5080: 256 bits / 32 Gbps / up to 16 GB memory / max bandwidth 1024 GB/s
• RTX 5070: 192 bits / 32 Gbps / up to 12 GB memory / max bandwidth 768 GB/s
• RTX 5060 and RTX 5050: 128 bits / 32 Gbps / up to 8 GB memory / max bandwidth 512 GB/s

It is important to note that these data are preliminary and subject to change. Particularly, the memory configuration for entry-level models will be an important aspect, considering the increasing video memory requirements for games at 1080p resolution and above.

The new "Blackwell" series is expected to be manufactured using TSMC's 3nm technology process and will support the DisplayPort 2.1 standard, as well as receive an updated Founders Edition cooling system. The premiere of the new generation is planned for the end of 2024, allowing users to enjoy advanced technologies and improved performance.