NVIDIA Quadro RTX (Ampere) To Feature The Full Fat GA102 GPU With Over 10,000 Cores & 48 GB of GDDR6 Memory
by Hassan MujtabaNVIDIA has recently unveiled its consumer-targeted GeForce RTX 30 series Ampere cards and soon, the company will be unveiling its prosumer and workstation targeted Quadro RTX graphics cards in the Ampere lineup too.
NVIDIA Quadro RTX (Ampere) Graphics Card With Full Fat GA102 GPU & 48 GB of GDDR6 Memory Spotted
The NVIDIA Quadro RTX lineup was first introduced with Turing GPUs and will soon be upgraded to the new Ampere graphics architecture. In a tweet by Moore's Law is Dead (via Videocardz), he says that the picture is for a next-generation Quadro RTX graphics card which features the full-fat GA102 GPU.
The leaked image that we are looking at is from the NVIDIA Control Panel which lists down several key specifications of the card itself. It is not stated what the card would be called but according to Kopite7kimi, the card could probably go by as Quadro RTX A6000 which means there will be room for the higher-end Quadro RTX A8000 in the future.
The specifications list down 10752 CUDA Cores which are arranged in 84 SMs. This is the full-fat Ampere GA102 GPU we are looking at which should be faster than the GeForce RTX 3090 since it packs 2% more cores however it would mostly depend on the clock speeds the card runs at. The Quadro RTX A6000 operates at a boost clock of 1860 MHz which should equal around 40 TFLOPs of Compute horsepower.
In terms of memory, the card will feature a total of 48 GB of GDDR6 memory running across a 384-bit wide bus interface. This should equal a total bandwidth of 768 GB/s. Now the bandwidth is still lower than the NVIDIA GeForce RTX 3090 which should be hitting almost 936 GB/s since it utilizes GDDR6X memory dies. It is definitely an interesting choice and could mean that we might end up seeing the GDDR6X implementation next year with a higher-end Quadro SKU since higher-density GDDR6X chips will be going in mass production in early 2021. There are also talks about a new Titan variant coming soon but we don't have any specifics.
NVIDIA Quadro RTX Series Graphics Card Lineup:
| Graphics Card | Quadro GV100 | Quadro RTX 5000 | Quadro RTX 6000 | Quadro RTX 8000 | Quadro RTX (Ampere) |
|---|---|---|---|---|---|
| GPU | Volta GPU | Turing GPU | Turing GPU | Turing GPU | Ampere GPU |
| GPU Process | 12nm | 12nm | 12nm | 12nm | 8nm |
| Die Size | 815mm² | 545mm² | 754mm² | 754mm² | 628mm² |
| GPU Cores | 5120 Cores | 3072 Cores | 4608 Cores | 4608 Cores | 10752 Cores |
| Tensor Cores | 640 Cores | 384 Cores | 576 Cores | 576 Cores | 656 Cores |
| Boost Clock | 1.62 GHz | 1.81 GHz | 1.77 GHz | 1.77 GHz | 1.86 GHz |
| Single Precision | 16.66 TFLOPs | 11.2 TFLOPs | 16.31 TFLOPs | 16.31 TFLOPs | 30 TFLOPs |
| Ray Tracing Spec | N/A | 6 GigaRays/Sec | 10 GigaRays/Sec | 10 GigaRays/Sec | TBA |
| VRAM | 32 GB HBM2 | 16 GB GDDR6 | 24 GB GDDR6 | 48 GB GDDR6 | 48 GB GDDR6 |
| NVLINK VRAM | N/A | 32 GB With NVLINK | 48 GB With NVLINK | 96 GB With NVLINK | 96 GB With NVLINK |
| Memory Bus | 4096-bit | 256-bit | 384-bit | 384-bit | 384-bit |
| Memory Bandwidth | 870 GB/s | 448 GB/s | 672 GB/s | 672 GB/s | 768 GB/s |
| TDP | 250W | ~160W | ~200W | ~225W | TBA |
| Price | $9000 US | $2300 US | $6300 US | $10000 US | TBA |
| Launch Date | 2018 | Q4 2018 | Q4 2018 | Q4 2018 | Q4 2020? |
We also don't know whether NVIDIA would utilize its GA102 or GA100 GPU for use in its flagship Quadro and Titan graphics cards based on the Ampere architecture. We will keep you posted if we hear more about the Quadro RTX Ampere line of graphics card but do expect news in the coming month.