After the success of the Intel Arc B580, all eyes are on Nvidia and AMD regarding how they’ll respond. In particular, the RTX 5000 lineup has been a hot topic, and after a lot of discussion about when the first GPUs will drop, we’ve had our first look at the RTX 5080, thanks to a couple of leaked photos online.
If you’re wondering when the RTX 5080 will be released, we’ve got you covered; according to a recent leak, it is rumored to launch on 21st January 2025. The rest of the RTX 50 series will likely follow suit and the 5090 is also slated for launch this month.
What we know about the RTX 5080 from this leak
The RTX 5080 was leaked on Chiphell but has since been taken down. The images above are sourced via VideoCardz; the graphics card in question is the MSI Gaming Trio with factory overclocked settings. We’re not sure what this card’s overclocked boost clock is, but on the back of the packaging, we can see that it confirms 16GB of GDDR7 VRAM with a 256-bit bus, along with three Display Ports and one HDMI port. The memory specs for the RTX 50 series have already been leaked prior, but this gives more evidence for the 5080’s 16GB VRAM.
While gamers were hoping for a higher-bit connection, the upgrade to the memory type is massive. GDDR6X has a maximum speed of 24 Gbps, while GDDR7 tops out at 32 Gbps. According to Micron, combined with a 384-bit memory interface width, GDDR7 can deliver 1.5 TB/s of bandwidth, which is nothing short of impressive.
That being said, there isn’t much difference in the RTX 5080, as far as we can tell from this, as Nvidia stuck with 16GB VRAM, which is exactly the same as the RTX 4080. However, the RTX 3080 came in 10 and 12GB variants. So, we might have to wait until the next generation to see a VRAM increment in the RTX XX80 series lineup.
Leaked RTX 5080 specs versus the 4080
The RTX 5080 specs were leaked a few months back, showing relatively small improvements over the 4080.
| Specification | RTX 5080 (leaked) | RTX 4080 |
|---|---|---|
| GPU | Blackwell GB203-400-A1 | Ada Lovelace AD103-300-A1 |
| SM count | 84 | 76 |
| CUDA cores | 10,752 | 9,728 |
| VRAM | 16GB GDDR7 | 16GB GDDR6X |
| Memory bus | 256-bit | 256-bit |
| Memory clock | 30 Gbps | 22.4 Gbps |
| Memory bandwidth | 896 GB/s | 716.8 GB/s |
| Power connectors | 1 x 16-pin | 1 x 16-pin |
| TDP | 400W | 320W |
| DisplayPort version | DisplayPort 2.1a UHBR20 | DisplayPort 1.4a |
| Bus interface | PCIe 5.0 x16 | PCIe 4.0 x16 |
