Just only a few days left and NVIDIA’s #UltimateCountdown will hit 00:00:00. But some people are just too excited and can’t hold it anymore. Recently, specifications of the upcoming GeForce RTX 3090 and RTX 3080 were leaked. According to Videocardz, the upcoming RTX 30 series cards will feature 2nd gen Ray Tracing cores, 3rd gen Tensor Cores, and will support PCIe 4.0; including DisplayPort 1.4a and HDMI 2.1. These cards are based on NVIDIA’s Ampere architecture with a 7nm GPU. Check out the rest of the specs below.
NVIDIA GeForce RTX 3090 Specifications
Let’s start with the successor of the RTX 2080 Ti, which is the NVIDIA GeForce RTX 3090. Based on the previous leak, the GeForce RTX 3090 is one big graphics card that will occupy three-PCI slots. It features a GA102-300 7nm GPU and will have 5248 CUDA cores. The boost clock speed is said to be 1,695 MHz and is paired with a 24GB GDDR6X memory.
Micron already spilled the information regarding the memory. The RTX 3090’s memory clock speed is said to be 19.5Gbps and a memory bandwidth of 936 GB/s, running on a 384-bit memory interface. It’s thermal graphics power or TGP is 350W.
NVIDIA GeForce RTX 3080 Specifications
The GeForce RTX 3080 will feature a GA102-200 GPU, more or less a cut-down variant of the GA102-300. It features 4352 CUDA cores with a boost clock speed of 1,710 MHz. However, it looks like there is a significant difference between the memory size of the RTX 3090 and RTX 3080.
The memory size of the GeForce RTX 3080 is only 10GB GDDR6, with a memory clock speed of 19Gbps, and a memory bandwidth of 760 GB/s running on a 320-bit memory interface. Its thermal graphics power is not that far from the RTX 3090, as its TGP is 320W.
NVIDIA GeForce RTX 3070 Specifications
The GeForce RTX 3070 is also said to be announced together with the RTX 3090 and 3080. But some reports are claiming that it will be available towards the end of September instead. Specs-wise, the RTX 3070 features a GA104-300 GPU with 8GB of GDDR6 memory. Memory clock speed is 16Gbps and it has a bandwidth of 512GB/s running on a 256-bit memory interface. Its TGP is said to be 220W.
However, the number of CUDA cores and its clock speeds are still unknown at this time. Also, notice that it will use the older GDDR6 memory, instead of the faster and newer GDDR6X memory. I’m just guessing here, but by the looks of it, there might be a significant gap between the performance of the RTX 3080 and RTX 3070. We’ll find out that soon.
Below is a table comparing the specifications of the RTX 3090 and RTX 3080 vs the RTX 2080 Ti and RTX 2080.
|Specifications||RTX 3090||RTX 3080||RTX 2080 Ti||RTX 2080 Super||RTX 2080|
|Boost Clock||1695 MHz||1710 MHz||1635 MHz||1815 MHz||1800 MHz|
|Memory Size / Type||24GB GDDR6X||10GB GDDR6X||11GB GDDR6||8GB GDDR6||8GB GDDR6|
|Memory Clock||19.5 Gbps||19 Gbps||14 Gbps||15.5 Gbps||14 Gbps|
|Memory Bandwidth||936 GB/s||760 GB/s||616 GB/s||496 GB/s||448 GB/s|
|Price (At launch)||$1400 (Rumor)||$800 (Rumor)||$999 / $1,199 (FE)||$699||$699|