The current GeForce RTX 2000 series based on Nvidia’s Turing architecture was launched in September 2018. Almost exactly two years later, the time has now come for the RTX 3000 series of gaming graphics cards based on the new Ampere architecture. As usual, the high-end cards in the lineup – the GeForce RTX 3090, RTX 3080, and RTX 3070 – are launched first and appear to greatly enhance performance in the high-end GPU market.
For the Ampere generation, Nvidia has adopted seven-nanometer (nm) manufacturing technology and a new GDDR6X memory standard. The Ampere-based GPUs also come with second-generation raytracing cores and third-generation Tensor cores. Another detail worth noting is that Nvidia has taken the step over to the PCI Express 4.0 interface.
RTX 3090, 3080 & 3070 Specifications
| RTX 3090 | RTX 3080 | RTX 3070 | |
|---|---|---|---|
| GPU Chip | Ampere GA100 | Ampere GA103 | Ampere GA104-300 |
| CUDA Cores | 10,496 | 8,704 | 5,888 |
| GPU Clock | 1.4 GHz | 1.44 GHz | 1.5 GHz |
| Max. Turbo | 1.7 GHz | 1.71 GHz | 1.73 GHz |
| VRAM | 24GB GDDR6X | 10GB GDDR6X | 8GB GDDR6 |
| Memory Bus | 384-bit | 320-bit | 256-bit |
| Shader TFLOPS | 36 | 30 | 20 |
| RT-FLOPS | 69 | 58 | 40 |
| Tensor-FLOPS | 285 | 238 | 163 |
| MSRP | $1,499 | $699 | $499 |
According to Nvidia, Ampere almost doubles the performance per watt compared to its predecessor, and the RT (ray-tracing) cores likewise offer significantly better raytracing performance.
GeForce RTX 3090 – The Ampere Titan Equivalent
At the very top of the hierarchy is the equivalent of a Titan card but with a new naming scheme. The Geforce RTX 3090 comes with no less than 24 GB of video memory (VRAM) operating at 19.5 Gbps, a whopping 10,496 CUDA cores, and a maximum turbo boost frequency of 1.7 gigahertz (GHz).
According to Nvidia’s own data, the Geforce RTX 3090 will offer considerable improvements in “RTX On” performance compared to the previous generation’s flagship. During the launch event, Nvidia had this monster of a card pushing ray-traced games in 8K resolution at 60 frames per second. It also has the capacity to burn a really large hole in your wallet thanks to its $1,499 price tag.
GeForce RTX 3080
Right below the flagship, we find the GeForce RTX 3080, which is considerably less costly and uses a slightly trimmed version of the RTX 3080’s graphics processor. It still includes a very respectable 8,704 CUDA cores and has a maximum turbo frequency of 1.72 GHz. Again according to Nvidia, the RTX 3080 will offer up to double the performance compared to the RTX 2080 at a $699 price point.
Nvidia has also cut the amount of VRAM down to 10 GB of GDDR6X. There is however speculation that partner manufacturers might offer versions with additional VRAM (up to 20 GB).
An early hands-on preview of the RTX 3080 from Digital Foundry does not quite manage to confirm the doubled performance over the RTX 2080, but gains of around 80% are still very impressive – and the largest generational performance leap for many years.
GeForce RTX 3070
The GeForce RTX 3070 will launch a bit later than the 3090 and 3080, in October. With its 5,888 CUDA cores along with eight gigabytes of GDDR6, it should greatly outpace the current high-end favorite, the RTX 2070 Super, but will be priced about the same at $499.
thanks for sharing this helpful informative article with us.