NVIDIA is teasing a 21 day ‘Ultimate Countdown’, just days after detailed pricing & launch date rumors for its upcoming Ampere GPUs emerged.
Ampere is the codename for NVIDIA’s next generation GPU architecture. The current RTX 20-series cards use the Turing architecture, and the new cards are expected to use RTX 30-series naming.
Ampere is already used in NVIDIA’s AI data center accelerators, so some details of the architecture are known. It moves from a 12nm fabrication process to the new (more efficient) 8nm and introduces 3rd generation tensor cores, which power DLSS upscaling and other AI features.
Across its GeForce social media channels, the #UltimateCountdown posts are accompanied by a new ’21 Days. 21 Years.’ banner image.
21 days from today works out as August 31, so keep an eye on NVIDIA later this month- or check UploadVR where we’ll break down any Ampere announcements.
The Latest Rumors
In the months before every new generation of GeForce GPUs launches, the rumor mill is active with claims of specs, product names, launch dates, and even pricing.
The most cited recent rumor is Wccftech’s report published over the weekend, with apparent basic specs and launch dates of five potential RTX 30-series products:
| Rumored Product | Rumored VRAM | Rumored Bus Width | Rumored Launch |
| RTX 3090? | 24 GB | 384-bit | Late September |
| RTX 3080 Ti? | 20 GB | 320-bit | October |
| RTX 3080? | 10 GB | 320-bit | Mid September |
| RTX 3070? | 8 GB | 256-bit | Late September |
| RTX 3060? | 8 GB | 256-bit | – |
CUDA Cores are NVIDIA’s stream processors, similar to the cores of a CPU. VRAM is the amount of fast random access memory available, and the memory bus dictates the amount of data transferable per clock cycle (GPUs typically cycle billions of types per second).
This rumor doesn’t include the increases to clock speed and other key specs that will likely improve via the move to the more efficient 8nm node process.
Ampere is also rumored to introduce GDDR6X memory. If these claims are real, there should be a considerable increase in performance over Turing.
While Wccftech has a track record of getting these specs right in the past, the outlet has also had its fair share of misfires too.
Regardless of the validity of current leaks, we’ll be keeping a close eye on NVIDIA in the lead up to the apparent August 31 event.
Are you looking to get a new GPU when Ampere drops, or are you holding off a few more years- or for AMD? Let us know in the comments below!
