Tom is a UK-based word spewer with a taste for everything weird and wonderful about games kick-started by a transformative play-through of Metal Gear Solid many moons ago. Adores Hollow Knight, probably a bit too much.
WePC is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
Last Updated:
Ahead of NVIDIA’s expected Ampere GeForce RTX 30-Series reveal later today, an accidental early listing from GPU manufacturer Gainward has revealed specifications for the upcoming cards.
Unsurprisingly, the full product pages have since been removed, but not before being snapped up for all to see. According to the listings, Gainward plans to push out four RTX 30-series GPUs, two apiece for the RTX 3090, and RTX 3080 SKUs. The RTX 3070 is notably absent. Though this doesn’t necessarily imply it won’t feature during NVIDIA’s announcement later today, it may, however, signal a launch further down the line.
The four GPUs are the GeForce RTX 3090 Phoenix, the GeForce RTX 3090 Phoenix GS, the GeForce RTX 3080 Phoenix, and the GeForce RTX 3080 Phoenix GS. GS refers to ‘Golden Sample’ and translates to factory-overclocked clock speeds and slight boost clock speed improvement.
GeForce RTX 3090 Phoenix GS
- CUDA cores: 5248
- Boost Clock speed: 1725 MHz
- Memory: 24GB GDDR6X
- Memory Interface: 384 bits
- Memory clock: 9750 MHz (19.5 Gbps)
- Memory Bandwidth: 936 GB/s
- TGP: 350 W
- PCIe: Gen 4
- Max power consumption: 350W
- Output: HDMI 2.1, DisplayPort 1.4a
GeForce RTX 3090 Phoenix
- CUDA cores: 5248
- Boost Clock speed: 1695 MHz
- Memory: 24GB GDDR6X
- Memory Interface: 384 bits
- Memory clock: 9750 MHz (19.5 Gbps)
- Memory Bandwidth: 936 GB/s
- TGP: 350 W
- PCIe: Gen 4
- Max power consumption: 350W
- Output: HDMI 2.1, DisplayPort 1.4a
GeForce RTX 3080 Phoenix GS
- CUDA cores: 4352
- Boost Clock speed: 1740 MHz
- Memory: 10GB GDDR6X
- Memory Interface: 320 bits
- Memory clock: 9500 MHz (19 Gbps)
- Memory Bandwidth: 760 GB/s
- TGP: 320 W
- PCIe: Gen 4
- Max power consumption: 320W
- Output: HDMI 2.1, DisplayPort 1.4a
GeForce RTX 3080 Phoenix
- CUDA cores: 4352
- Boost Clock speed: 1710 MHz
- Memory: 10GB GDDR6X
- Memory Interface: 320 bits
- Memory clock: 9500 MHz (19 Gbps)
- Memory Bandwidth: 760 GB/s
- TGP: 320 W
- PCIe: Gen 4
- Max power consumption: 320W
- Output: HDMI 2.1, DisplayPort 1.4a
All four GPUs feature the 7nm fabrication process (although no word on where NVIDIA is sourcing the silicon – TSMC or Samsung?), 2nd Gen Ray Tracing cores, 3rd Gen Tensor Cores, NVIDIA DLSS, NVIDIA GeForce Experience, NVIDIA G-Sync, NVIDIA GPU Boost, Vulkan RT API, and OpenGL 4.6. All will feature the same 2.7-slot triple-fan cooling solution. Aesthetically the cards are indistinguishable from one another with a rather sleek design, chunky backplate, and what appears to be an RGB lighting arrangement of some description.
Interestingly, Gainward notes that the Phoenix line-up uses two 8-pin PCI Express power connectors, seemingly confirming that the recently announced 12-pin connector arrangement will be exclusive to NVIDIA’s Founders Edition family. The manufacturer also recommends a minimum 750W, or greater, power supply to run the GPUs.
Overall, the specifications align comfortably with many of the most recent GeForce RTX 30-series leaks, though Gainward’s slip-up offers the most unambiguous confirmation to date. Of course, the timing is a little off, and someone over at Gainward will likely be receiving quite the telling off for their blunder.
Over to NVIDIA for the big GeForce Special Event blow out, which starts at 5 pm BST/9 am PT.