Gaming

24 GB VRAM At 32 Gbps & 10752 Cores at 400W+

24 GB VRAM At 32 Gbps & 10752 Cores at 400W+

The first specifications of NVIDIA’s GeForce RTX 5080 SUPER have been leaked by Kopite7kimi, revealing 24 GB of VRAM and a fully enabled GB203 chip.

NVIDIA’s GeForce RTX 5080 SUPER To Pack Full GB203 GPU Die & 24 GB VRAM Clocked at 32 Gbps

It was reported a while ago that NVIDIA was working on a SUPER series refresh of the RTX 50 series. This lineup would feature a series of updated specs, especially to the memory configurations, and it looks like Kopite7kimi has more details to share.

According to the leaker who had been on point with his previous leaks, the NVIDIA GeForce RTX 5080 SUPER will retain the full GB203 Core configuration of the RTX 5080 with 10752 cores, but the GPU SKU is changed from GB203-400 to GB203-450, and the PCB is PG147-SKU35.

It is stated that the card will operate at a TBP of 400W+, which even at 400W, will be 40W higher than the Non-SUPER RTX 5080, though it does make sense if NVIDIA is going to clock the chip much higher and the memory specifications, which we will talk about in a bit, also seem to give us a good reason why the power numbers have been raised so much.

So, coming to the memory specifications, it is stated that the NVIDIA GeForce RTX 5080 SUPER will feature 24 GB of VRAM across a 256-bit bus while running at 32 Gbps, a 2 Gbps bump over the 30 Gbps used by the non-SUPER variant. These will be the latest 3 GB GDDR7 modules, and this gives us the full 1 TB/s of bandwidth, which we had hinted at a long time ago.

That’s a 6.6% bump in memory bandwidth and, given that the core config will remain the same, we can expect real-world gaming performance in the 5% range, similar to the RTX 4080 SUPER, though that extra VRAM onboard will look very enticing to those who have waited to buy the 16 GB RTX 5080. So the main updates on the SUPER include:

RTX 5080 SUPER vs RTX 5080 Non-SUPER:

  • Same Cores but Different GB203 SKU (450 vs 400)
  • Possibly Higher Clock Speeds
  • 24 GB VRAM vs 16 GB VRAM (+50%)
  • Same 256-bit Bus Interface
  • 6.6% Higher GDDR7 speeds (32 Gbps vs 30 Gbps)
  • 6.6% More Bandwidth (1024 GB/s vs 960 GB/s)
  • 11.1% Higher TBP (400W+ vs 360W)

Currently, there’s no word on when NVIDIA will launch the RTX 50 SUPER series and AIBs haven’t started talking about these cards yet, not even at the Computex showfloor, but it looks like we will get updates in the second half of this year with a launch possibly taking place in late 2025 or at CES 2026.

NVIDIA RTX 50 SUPER GPU Lineup (Preliminary):

Graphics Card Name NVIDIA GeForce RTX 5080 SUPER NVIDIA GeForce RTX 5080 NVIDIA GeForce RTX 5070 SUPER NVIDIA GeForce RTX 5070
GPU Name Blackwell GB203-450 Blackwell GB203-400 TBD Blackwell GB205-300-A1
GPU SMs 84 (84 Full) 84 (84 Full) TBD 50 (50 Full)
GPU Cores 10752 10752 TBD 6144
Clock Speeds TBD 2.62 GHz TBD 2.51 GHz
Memory Capacity 24 GB GDDR7 16 GB GDDR7 16 GB GDDR7 12 GB GDDR7
Memory Bus 256-bit 256-bit 192-bit 192-bit
Memory Speed 32 Gbps 30 Gbps TBD 28 Gbps
Bandwidth 1024 GB/s 960 GB/s TBD 672 GB/s
Power Interface 1 12V-2×6 (16-Pin) 1 12V-2×6 (16-Pin) TBD 1 12VHPWR (16-Pin)
Launch TBD 30th January, 2025 TBD 5th March, 2025
TBP 400W+ 360W TBD 250W
Price TBD $999 US TBD $549 US

Leave a Reply

Your email address will not be published. Required fields are marked *