TECH NEWS – Nvidia is reportedly planning to “supercharge” the Blackwell architecture, as one of these enhanced models has now been leaked.

Reports have surfaced that Nvidia is working on Super variants of its RTX 5000 GPUs, offering stronger specifications—especially in terms of memory. Now, known and trusted leaker Kopite7kimi has shared more details. According to him, the RTX 5080 Super will retain the full GB203 core configuration with 10,752 FP32 cores, but its GPU SKU will shift from GB203-400 to GB203-450, and the PCB will be PG147-SKU35. The card is expected to run with a TBP of over 400W, which—even at 400W—is 40W more than the standard RTX 5080.

As for memory, the Nvidia GeForce RTX 5080 Super is said to come with 24 GB of VRAM on a 256-bit bus, running at 32 Gbps—a 2 Gbps improvement over the 30 Gbps non-Super version. These will be the latest 3 GB GDDR7 modules, allowing the card to deliver 1 TB/s of bandwidth. That’s a 6.6% increase in memory bandwidth. Since the core setup remains unchanged, real-world gaming performance is expected to be within 5%, similar to the RTX 4080 Super. However, the expanded VRAM will likely be attractive to those who held off on the 16 GB RTX 5080.

How It Compares to the RTX 5080

Same core count, different GB203 SKU (450 vs. 400), potentially higher clock speeds (still unconfirmed), 50% more VRAM (24 GB vs. 16 GB), same 256-bit bus, 6.6% faster GDDR7 (32 Gbps vs. 30 Gbps), 6.6% higher bandwidth (1024 GB/s vs. 960 GB/s), and 11.1% higher TBP (400+ W vs. 360 W).

There’s still no official word on when Nvidia plans to launch the RTX 5000 Super lineup. Even at Computex, manufacturers remained quiet on the topic. We’ll likely hear more later in the year, with a potential release at the end of 2025 or during CES 2026 in January.

Source: WCCFTech

GeForce RTX 5080 Super
PG147-SKU35
GB203-450-A1
10752FP32
256-bit GDDR7 24G 32Gbps
400+W

— kopite7kimi (@kopite7kimi) May 20, 2025