Summary
Nvidia’s GeForce RTX 50-series of GPUs are the talk of the town currently, as they are the upcoming graphics cards that will take GPU hardware and software to the next level. TheRTX 40-series has already established how incredibly powerful Nvidia’s technologies areand the RTX 50-series will be no exception. There is a lot of information about the RTX 50-series of GPUs, some confirmed by the company itself but most of it churning in the rumor mill. So, here’s everything we know so far about the Nvidia RTX 50-series GPUs.
Nvidia GeForce RTX 50-series GPUs: Architecture and Improvements
The RTX 50-series of GPUs are based on the new Blackwell architecture, the successor to the Ada Lovelace architecture thatpowered the RTX 40 series. This time around, Nvidia is not using a new TSMC 3nm-class node but has gone with a higher-performing version of the 4N node – TSMC 4NP. Still, the company has managed to upgrade most of the technologies including the number of transistors, deeper AI integration, fifth-generation NVLink, and more CUDA cores, among other improvements.
NVIDIA GeForce RTX 4090 Founders Edition
NVIDIA’s top-of-the-line GPU, the GeForce RTX 4090 Founders Edition, contains 16,384 CUDA Cores. It can support up to 8K gaming and allows for actual, smooth ray-tracing gameplay. The Founders Edition card is also an excellent GPU variant in terms of cooling, VRMs, and value.
Nvidia GeForce RTX 50-series GPUs: Rumored Specifications
The GeForce RTX 50-series of GPUs is expected to follow the same naming convention as the RTX 40 series, which means we can expect the RTX 5050, RTX 5060, RTX 5070, RTX 5080, and the RTX 5090 at least as part of the initial launch. All the 50-series cards are expected to come with the 16-pin 12VHPWR connector that was present on the higher-end RTX 40 series GPUs. These new cards have internal codes: GB202, GB203, GB205, GB206, and GB207. These GPUs are also said to come with DisplayPort 2.1 ports, something that the RTX 40-series GPUs were missing butAMD’s Radeon RX 7000 series included.
The top-tier RTX 5090 is expected to come with a 512-bit memory and support 32GB GDDR7 VRAM modules that run at 28Gbps. In comparison to the RTX 4090, it is said to come with 192 streaming multiprocessors (128 on the RTX 4090), 192 Ray Tracing cores (128 on the RTX 4090), 24,576 CUDA cores (16,384 on the RTX 4090), 768 Tensor cores (512 on the RTX 4090), and a 2.9GHz boost clock (2.52GHz on the RTX 4090). If these specifications are true, the RTX 5090 will be about 1.7 times faster than the RTX 4090.
With the rumors about the Nvidia RTX 5090 out of the way, it is also expected that some of thelower-end GPUsin the upcoming lineup will stick with 8GB VRAM. With the industry, including AMD, moving towards higher and higher VRAM, it’s interesting that Nvidia is still offering GPUs with 8GB VRAM. It is also possible that like the RTX 4060, the company may release a 16GB variant of the 8GB GPU down the line.
CUDA cores
RTX 5050
2560
20
GDDR7 8GB 128-bit
RTX 5060
4608
36
RTX 5070
6400
50
GDDR7 12GB 192-bit
RTX 5080
10752
84
GDDR7 16GB 256-bit
RTX 5090
24576
192
GDDR7 32GB 448-bit
As per the aforementioned rumored specifications, the gaps between these GPUs will be filled by “Ti” and “Super” models that are also expected to come with GDDR7 memory. Looking at the previous generation, there will likely be an RTX 5060 Super, RTX 5060 Ti, or both. We can expect an RTX 5070 Ti/ Super/ Ti Super, and an RTX 5080 Super. With the gap between the RTX 5080 and the RTX 5090 being quite significant, it wouldn’t be surprising if Nvidia decided to release an RTX 5080 Ti or Ti Super this time around.
Nvidia GeForce RTX 50-series Laptop GPUs: Rumored Specifications
Of course, the RTX 50-series desktop GPUs will have theirlaptop counterpartsas well, but even less is known about the laptop versions. They are also expected to come with similar upgrades as the desktop version and the laptop version of the RTX 5080 may feature 16GB of GDDR7 VRAM. It is also believed that the GB207 (possibly the RTX 5050), which is the slowest of the bunch, may be present first in laptops and then make its way to desktops. It should be noted that the RTX 4050 is a laptop-exclusive GPU.
Nvidia GeForce RTX 50-series GPUs: Launch Timeline and Pricing
Just to be clear, Nvidia has not shared any information on the launch timeline for the RTX 50-series GPUs. As per the latest rumors, the new GPUs for desktops will likely make an appearance at CES 2025, but there is a possibility that one of the high-end cards, either the RTX 4080 or the RTX 4090, may be unveiled towards the end of this year. It is also unclear which GPUs will make their way to the public as part of the initial launch.
As for the laptop GPUs, they are expected to be launched early next year and all laptop manufacturers should have options for the public available for purchase by the first quarter of 2025. AMD is also looking to launch its Radeon RX 8000 series GPUs for desktops early next year, so Nvidia may capitalize on the empty end of 2024 and break cover for one or both of its top-tier GPUs in the RTX 50-series.
As of now, there is no information on pricing but the top-tier RTX 5090 can be expected to cost somewhere in the $1800 to $2000 range, considering the RTX 4090 launched at $1600 but is now available at around the $2000 mark, if not more.
FAQ
Q: Will the 5090 be better than the 4090?
The Nvidia RTX 5090 is expected to be 1.7 times more performant than the RTX 4090, as per the rumored specifications. It will likely use more power and cost more than the RT 4090 as well.
Q: How much RAM will an RTX 5090 have?
The RTX 5090 is rumored to come with 32GB GDDR7 VRAM.
