32GB GDDR7, 20,000+ CUDA cores and 600W power

Some interesting features of NVIDIA’s next-generation GeForce RTX 5090 graphics card have been leaked: There is a huge difference between RTX 5090 and RTX 5080 graphics cards. The new RTX 5090 has more than 20,000 CUDA cores, 32 GB GDDR7 memory and will consume up to 600 W of power.

In a new post on . It is located on a 512-bit memory bus and consumes up to 600 W of power.

This means the new GeForce RTX 5090 will have an incredible memory bandwidth of 1.79 TB/s or 2.0 TB/s, depending on the bandwidth of the GDDR7 memory: 28 Gbps GDDR7 on a 512-bit memory bus, 1.79 The faster 32 Gbps GDDR7 on a 512-bit memory bus, which will result in TB/s, will increase memory bandwidth by 2.0 TB/s. They’re crazy, totally crazy.

We were expecting a 448-bit memory bus in previous rumors, but the massive 512-bit memory bus means NVIDIA is really aiming for an ultra-high-spec design for the GeForce RTX 5090. The possible 2.0TB/s memory bandwidth for 4K 240FPS games would be insane.

Kosta was also presented with NVIDIA’s new GeForce. The RTX 5090 may ship with not one but two 16-pin PCIe power connectors, which will likely require a new PSU.

NVIDIA GeForce RTX 5090 details so far:

20,000+ CUDA cores: With the rumored 21,760 CUDA cores, we’re expecting a major upgrade over the 16,384 cores inside the RTX 4090. Not only more CUDA cores, but also upgraded Blackwell GPU architecture cores over the existing Ada Lovelace GPU cores inside the RTX 4090.

32 GB new generation GDDR7 memory: NVIDIA is the first company to use the ultra-fast GDDR7 memory standard; The flagship GeForce RTX 5090 is pegged to pack an incredible 32GB of GDDR7 memory. This is another major upgrade over the RTX 4090 with 24GB of GDDR6X memory.

Up to 2.0 TB/s memory bandwidth: We hear that the RTX 5090 will feature a massive 512-bit memory bus, allowing for an incredible memory bandwidth ceiling of up to 2.0TB/s if the RTX 5090 uses 32Gbps GDDR7 memory modules. Another huge upgrade over the RTX 4090, which has 1.0 TB/s memory bandwidth in comparison.

600W power: Another big upgrade for the RTX 5090 is 600W of power, which is 150W more power on top of the RTX 4090’s 450W TDP. We’ve seen dedicated models and overclocked RTX 4090 cards with up to 600W+ power, but the RTX 5090 will be yet another level on top of that. We hope we see 600W+++ special AIB models of the RTX 5090.

4K performance: NVIDIA’s next-gen premium gaming GPU in the GeForce RTX 5090 should have 50-70% more performance over the RTX 4090, especially at high-end 4K resolution. 4K 120FPS gaming will be a much easier feat for the RTX 5090 than the RTX 4090. 4K 240FPS is coming this year and in 2025, so the RTX 5090 will be the GPU of choice for 4K 240FPS games. future.

VIEW GALLERY – 2 IMAGES

RT performance: This is where the biggest performance improvements of the next-generation Blackwell-based GeForce RTX 50 series GPUs will come: ray tracing. You can expect a pretty big 2-3x performance increase using RT, and in some cases 4x or more compared to the RTX 5090 and RTX 5080 and their RTX 40 series counterparts. Add the new DLSS 4 on top and you’ll get great RT performance.

DLSS4: This is probably my personal favorite part of the excitement around NVIDIA’s next-gen GPUs, the AI-powered upgrade with next-gen DLSS 4 running only on the new Blackwell-based RTX 50 series GPUs. We should expect higher image quality than what DLSS 3.x offers on a new RTX 5090 or RTX 5080 graphics card, and new levels of performance with DLSS 4 enabled.

power efficiency: NVIDIA’s current-gen GeForce RTX 4090 can use anywhere between 450W and 600W of power depending on model and overclock, but the new GeForce RTX 50 series GPUs are more than the flagship RTX 4090 and RTX 4080 SUPER Graphics cards that will offer more performance per watt. Another exciting part that will be solved in the near future.