NVIDIA’s upcoming RTX 40 series has gathered all the hype in the community. From unconfirmed specs to release dates, we encountered all forms of information overload.
Twitter user @kopite7kimi suggests new specifications for the GeForce RTX 40 series.
RTX 4090
The RTX 4090 is the flagship GPU from the RTX 40 range; it would feature 16384 cores instead of 16128, as rumored earlier. Hence, the Streaming Multiprocessors are also upgraded to 128.
The expected TDP is 450W, and the graphics card would support GDDR6X memory. The memory capacity and clock speeds are 24GB and 21 Gbps, respectively.
The RTX 4090 Ti variant will likely use the AD102 GPU with 144 cores. The AD102 has a larger L2 cache, so we expect a similar speed as that of the Radeon RX 6000 series.
RTX 4080
The RTX 4080 is based on the AD103 GPU, featuring 10240 CUDA cores and 16 GB memory capacity. The graphics card has GDDR6 memory support and will be launched a month after the flagship RTX 4090.
However, the memory capacity will likely change since 18 GB GDDR6 memory will be slower than the RTX 3080, which features 19 GB GDDR6 memory.
RTX 4070
The RTX 4070 will have a memory capacity of 10 GB, not 12 GB. Hence, the memory bus interface is also reduced to 160-bit from 192-bit. The GPU will feature 7168 CUDA cores. The supported AD104 GPU will have 512 disabled CUDA cores.
Here’s an overview of the rumored specs for RTX 40 series:
| RTX 4090 | RTX 4080 | RTX 4070 | |
| Architecture | Ada (TSMC N4) | Ada (TSMC N4) | Ada (TSMC N4) |
| GPU | AD102-300 | AD103-300 | AD104-275 |
| Board Number | PG139-SKU330 | PG139-SKU360 | PG139-SKU341 |
| CUDA Cores | 16384 | 10240 | 7168 |
| SMs | 128 | 80 | 56 |
| Memory Support | GDDR6X | GDDR6 | GDDR6 |
| Memory Capacity | 24 GB | 16 GB | 10 GB |
| Memory Bus | 384-bit | 256-bit | 160-bit |
| Memory Speed | 21 Gbps | 18 Gbps | 18 Gbps |
| TDP | ~450W | ~420W | ~300W |
| Release | Sep-Oct | Oct-Nov | Nov-Dec |
The RTX 4060 will use the AD106 GPU and is expected to have a TDP of over 220W. NVIDIA will be launching the graphics card in January next year.
About Nvidia
NVIDIA Corporation is an American multinational technology company incorporated in Delaware, based in Santa Clara, California. They design graphics processing units (GPUs) for the gaming and professional markets, as well as system on chip units (SoCs) for the mobile computing and automotive market.
Best known for the “GeForce” lines of GPUs, they are a direct competitor to AMD’s “Radeon” series. NVIDIA has also expanded its offerings with its handheld game consoles Shield Portable, Shield Tablet, and Shield Android TV and its cloud gaming service GeForce Now.
No Comments on NVIDIA’s RTX 4090, 4080 & 4070 Models Launching from September Onwards