A user on the Hardwareluxx Forums confirmed the existence of the NVIDIA SKU Ampere GA102-250 GPU, which was previously expected to be featured on the GeForce RTX 3080 Ti video card. It was detected when the forum member disassembled his GeForce RTX 3090, just to reveal that the GPU had initially been labeled as GA102-250 SKU instead of GA102-300, which is the official SKU used in the RTX 3090.
NVIDIA GeForce RTX 3080 Ti GPU rumors, the Ampere GA102-250, located on a GeForce RTX 3090 graphics card
The NVIDIA GA102-250-KD-A1 GPU existed only in the rumor, but now we can confirm that this is really a real chip. Initially planned for the GeForce RTX 3080 Ti video card, it looks like NVIDIA decided it could use an additional cut-off SKU to power the next flagship, while this SKU can be used for the GeForce RTX 3090 video card. You can see clearly that the GA102-250-KD-A1 GPU label has been cut and has instead been reused as SKU GA102-300-A1. What is interesting is that the GA102-250-KD-A1 was expected to feature 78 SM units enabled for a total of 9984 CUDA cores, while the RTX 3090 with its GA102-300-A1 GPU packs 82 SMs or 10496 cores.
It is possible that NVIDIA has decided to redirect SKU GA102-250 to GA102-300 due to the scarcity of its cutting-edge graphics cards. We cannot say with certainty whether the 250 variant offers the same performance / power limits as the 300 variant, but given that it has a lower naming scheme, it can be expected to be a lower compartment than the standard GA102-300 SKU . The GPU was also made during the 49th week of 2020, which runs from late November to early December. Again, we don’t know how far NVIDIA is redirecting its GA102-250 chips to GA102-300, but as they exist and that is out now, we can expect more users to find out for themselves.
NVIDIA’s GeForce RTX 3090 was identified using the GA102-250 GPU, the formerly supposed SKU for the RTX 3080 Ti. (Image credits: HardwareLuxx)
The NVIDIA GeForce RTX 3080 Ti, on the other hand, has undergone a change in specifications and will now use the Ampere GPU SKU GA102-225-A1. The following are the specifications expected from the next flagship of the RTX 30 series.
NVIDIA GeForce RTX 3080 Ti 12 GB ‘rumored’ graphics card specifications
The NVIDIA GeForce RTX 3080 Ti 12 GB FE (Founders Edition) graphics card must feature the PG132-SKU18 PCB design and the GA102-225-KD-A1 graphics core. The GA102-225 GPU has also changed since the last time we saw them and will now feature 10240 CUDA cores in a total of 80 SM units. Since the main core configuration is completely new, leaked clocks and TMU / ROP counts must previously be discarded.
As for memory, the card will have 12 GB of GDDR6X memory. Unlike the RTX 3090’s 19.5 Gbps speeds, the GeForce RTX 3080 Ti retains the same memory speeds as the RTX 3080 at 19 Gbps. As we are getting 12 GB of memory, the NVIDIA GeForce RTX 3080 Ti will use a 384-bit bus interface, which equates to a total bandwidth of 912 GB / s. Therefore, although we see a reduction of 8 GB VRAM compared to the previous specification, the larger bus interface should generate a larger memory bandwidth.
The TGP for the board is set to be the same as the RTX 3080 at 320 Watts. This is definitely necessary to power the extra cores, so NVIDIA may have to optimize the clocks here a bit. The NVIDIA GeForce RTX 3080 Ti is expected to launch in mid-April for a suggested price of around $ 999. It is rumored that the card will be launched in May.
NVIDIA GeForce RTX 30 Series ‘Ampere’ graphics card specifications:
| Graphics card name | NVIDIA GeForce RTX 3050 | NVIDIA GeForce RTX 3050 Ti | NVIDIA GeForce RTX 3060 | NVIDIA GeForce RTX 3060 Ti | NVIDIA GeForce RTX 3070 | NVIDIA GeForce RTX 3070 Ti? | NVIDIA GeForce RTX 3080 | NVIDIA GeForce RTX 3080 Ti? | NVIDIA GeForce RTX 3090 |
|---|---|---|---|---|---|---|---|---|---|
| GPU Name | Ampere GA107 | Ampere GA106? | Ampere GA106-300 | Ampere GA104-200 | Ampere GA104-300 | Ampere GA104-400 | Ampere GA102-200 | Ampere GA102-225? | Ampere GA102-300 |
| Process Node | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm | Samsung 8nm |
| Matrix size | TBA | TBA | TBA | 395.2 mm2 | 395.2 mm2 | 395.2 mm2 | 628.4 mm2 | 628.4 mm2 | 628.4 mm2 |
| Transistors | TBA | TBA | TBA | 17.4 billion | 17.4 billion | 17.4 billion | 28 billion | 28 billion | 28 billion |
| CUDA Cores | 2304? | 3584? | 3584 | 4864 | 5888 | 6144? | 8704 | 10240? | 10496 |
| TMUs / ROPs | TBA | TBA | 112/64 | 152/80 | 184/96 | 192/104? | 272/96 | 320/112? | 328/112 |
| Tensioner / RT cores | TBA | TBA | 112/28 | 152/38 | 184/46 | 192/48? | 272/68 | 320/76? | 328/82 |
| Base Clock | TBA | TBA | 1320 MHz | 1410 MHz | 1500 MHz | TBA | 1440 MHz | TBA | 1400 MHz |
| Boost Clock | TBA | TBA | 1780 MHz | 1665 MHz | 1730 MHz | TBA | 1710 MHz | TBA | 1700 MHz |
| FP32 Compute | TBA | TBA | 12.7 TFLOPs | 16.2 TFLOPs | 20 TFLOPs | TBA | 30 TFLOPs | TBA | 36 TFLOPs |
| RT TFLOPs | TBA | TBA | 25.4 TFLOPs | 32.4 TFLOPs | 40 TFLOPs | TBA | 58 TFLOPs | TBA | 69 TFLOPs |
| Tensor-TOPs | TBA | TBA | 101 TOPs | 129.6 TOPs | 163 TOPs | TBA | 238 TOPs | TBA | 285 TOPs |
| Memory capacity | 4 GB GDDR6? | 6 GB GDDR6? | 12 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6X | 10 GB GDDR6X | 12 GB GDDR6X | 24 GB GDDR6X |
| Memory Bus | 128 bits | 192 bits? | 192 bits | 256 bits | 256 bits | 256 bits | 320 bits | 384 bits | 384 bits |
| Memory Speed | TBA | TBA | 15 Gbps | 14 Gbps | 14 Gbps | TBA | 19 Gbps | 19 Gbps | 19.5 Gbps |
| Bandwidth | TBA | TBA | 360 Gbps | 448 Gbps | 448 Gbps | TBA | 760 Gbps | 912 Gbps | 936 Gbps |
| TGP | 90W? | TBA | 170W | 200W | 220W | 250W? | 320W | 320W | 350W |
| Price (MSRP / FE) | $ 149? | $ 199? | $ 329 | $ 399 US | $ 499 US | $ 599? | $ 699 US | $ 999? | $ 1499 US |
| Release (availability) | 2021? | 2021? | February 2021 | December 2020 | October 29 | May 2021? | September 17 | April 2021? | September 24 |