NVIDIA’s AMP’d Up GeForce RTX 3060, GeForce RTX 3070 & GeForce RTX 3080 To Tackle AMD’s Big Navi RDNA 2 Graphics Cards

NVIDIA’s GeForce RTX 30 series was just unveiled last week but reports are coming that we might be getting even faster variants soon. The latest report comes from Kopite7kimi who has spilled the beans on what could be the specifications of the AMP’d up GeForce RTX 3060 graphics card along with a few tidbits that we have received for the memory boosted GeForce RTX 3070 and GeForce RTX 3080 models.

NVIDIA To AMP UP! Ampere Based GeForce RTX 3060, GeForce RTX 3070 & GeForce RTX 3080 To Tackle AMD’s RDNA 2 Graphics Card Lineup

The speculation of an AMP’d Up series goes all the way before the launch of the GeForce RTX 30 series cards when NVIDIA was still early in deciding the final specifications of its graphics cards. There’s a reason why we saw entries of several memory variants including 11/12/20/22 GB prior to the announcement day. As expected, some of these cards were finalized but the rest weren’t taken out of the equation, at least yet.

Bykski Launches GeForce RTX 3090 & GeForce RTX 3080 Reference & Custom FE Flavor Water Blocks

It’s simply because the other variants can be devised in a way to tackle whatever AMD has planned with RDNA 2. The immediate competitor which is also known as “Big Navi” is a complete unknown at the moment but NVIDIA has already positioned itself well over that by offering its Titan class RTX 3090 under the GeForce banner. That doesn’t mean we won’t be getting a Titan product with better specifications later on (up to 48 GB memory if NVIDIA wants to wait till Micron is ready with 16 Gb GDDR6X modules).

NVIDIA GeForce RTX 3060 (SUPER/Ti) Specifications

So what comes next? Surely there’s room for a lot of graphics cards in NVIDIA’s Ampere lineup even after the release of the GeForce RTX 3070 in October. We know that NVIDIA has the RTX 3060 that’ll come after the RTX 3070 but according to Kopite7kimi, the card may feature much better specs than previously expected. The naming is claimed to either feature the Ti or SUPER moniker but let’s just call the graphics card an AMP’d up variant of the original RTX 3060 design.

PG142-SKU20 GA104-200 8G GDDR6 4864FP32=RTX 3060Ti or Super, Thanks to VCZ.

— kopite7kimi (@kopite7kimi) September 8, 2020

AMD Ryzen 4000 ‘Vermeer’ Zen 3 Desktop CPU Lineup Could Feature 10 Core Flavors, New Boost & Infinity Fabric Design

In terms of specifications, the GeForce RTX 3060 is expected to feature the PG142 PCB and utilize the GA104-200 GPU core. This chip will feature 4864 CUDA cores in 38 SMs compared to the 5888 cores featured on the RTX 3070 which has 46 SMs enabled for its GA104 GPU. The full GA104 GPU is likely to feature 6144 cores or 48 SMs. The card is also said to feature 8 GB of GDDR6 memory which is a nice upgrade over the standard GeForce RTX 2060 graphics card but then again, this is more of a successor to the GeForce RTX 2060 SUPER which also features 8 GB memory and came with the TU104 core upgrade.

NVIDIA GeForce RTX 3080 / RTX 3070 AMP’D Up Cards?

Next up, we have the GeForce RTX 3080 and GeForce RTX 3070 variants with double the memory. Now these graphics cards have been in the discussion for a while and several OEM and AIBs have mentioned (and even confirmed these cards). The main issue is they don’t know when NVIDIA will be releasing them or will they keep them reserved for a later launch. It all depends on how good AMD’s RDNA 2 is.

While these two cards have the same names as the standard variants, don’t expect similar pricing at all. The GeForce RTX 3070 with double the memory (16 GB) could end up close to the pricing of the GeForce RTX 3080. The GeForce RT 3080 with 20 GB could end up between $799-$899. These two cards aren’t meant to replace the ones that are initially announced but rather offer a step-up in terms of memory and possibly even better core specs, at higher prices. The rumor that NVIDIA is going to adjust the prices of existing cards to make way for newer variants is untrue.

The other interesting detail we got is that the 16 GB RTX 3070 features the PG142-10 PCB. This is the same PCB SKU as the RTX 3070 and the rumored RTX 3060 which is mentioned above. From what I can tell, a PCB revision needs to be made for the card to support double the VRAM. There have to be solder joints for dual-VRAM capacity on the back of the PCB but that doesn’t seem to be the case for the RTX 3070 reference design. As such, NVIDIA could use the higher density GDDR6 modules for the 16 GB variant but that also costs a bit more to use.

The GeForce RTX 3080 could go two routes based on how well RDNA 2 performs. If 20 GB is required imminently, then NVIDIA could just slap in another 10 GDDR6X memory modules at the back of the existing PG133/PG132 PCBs and call it a day. The second and more expensive option is to wait for the higher density GDDR6X (16Gb) to arrive early next year and incorporate them but considering that wouldn’t result in a good margin for NVIDIA, it is highly likely that they would just go the dual-sided DRAM route like the RTX 3090.

NVIDIA is aware that AMD might try to leverage the higher VRAM capacity as a big marketing point for its RDNA 2 lineup which is why NVIDIA wants to offer its own higher VRAM offerings. The 20 GB RTX 3080 could be ideally positioned against a 16 GB card while a 16 GB RTX 3070 can be a good marketing standpoint against a 12 GB variant. That’s not to say that the lower-VRAM cards that are already out by then wouldn’t offer great performance but like I said, the amount of VRAM will be huge for marketing this gen.

NVIDIA GeForce RTX 30 “Ampere’ Graphics Card Rumored Lineup:

Once again, these cards are still in a rumored territory for now, and don’t expect any information from NVIDIA to come out at least until AMD is done with its launch. We will keep you posted if we hear any more details on the remaining Ampere (RTX 30) cards.