Rumor: Nvidia RTX 5090 could pack 24.5K cores, 32GB GDDR7 VRAM, and 3GHz boost clocks

In July, it was reported that the next Nvidia flagship GPU (RTX 5090 or 5090 Ti) will likely have a 512-bit memory bus, which means 32 GB of VRAM. Today, it was revealed that Nvidia is targeting an overall 1.7x performance boost for its next-generation "Blackwell" architecture GPUs. New leaked specs of the flagship RTX 5090, codenamed GB202, have surfaced, showing Nvidia's bold targets if achieved.

According to a post by known leaker Panzerlied on the Chiphell forums, Nvidia aims for a 50% increase in CUDA cores for the RTX 5090, bringing the total to a massive 24,576. That's over 8x the cores of Turing-based RTX 2080. Memory bandwidth is also slated to receive a 52% bump through use of 32Gbps GDDR7, up from GDDR6X on the current RTX 4090.

Clock speeds could see one of the biggest generational jumps with a projected 15% increase. This would see the RTX 5090 boost up to 2.9GHz, easily surpassing 3GHz during gaming workloads. The existing RTX 4090 maxes out at 2.52GHz.

To further boost performance, Nvidia reportedly plans to triple the L2 cache to 128MB - far exceeding AMD's 96MB on the RX 7900 XTX. This additional caching aims to maximize the higher memory bandwidth enabled by GDDR7.

If Nvidia can deliver on these ambitious targets, the RTX 5090 would provide over 70% greater performance than the RTX 4090 based on specs alone.

NVIDIA RTX 5090 vs. RTX 4090

• 50% more cores than the RTX 4090

• 52% increase in memory bandwidth

• 78% larger L2 cache (likely 128MB)

• 15% higher frequencies, pushing boost clocks over 3.0GHz

• Around 70% higher overall performance

While unconfirmed, the leaked specs so far point to a new Blackwell-based GPU codenamed "GB202" that could push gaming performance to new heights. It remains to be seen if Nvidia can maintain a reasonable 3-slot form factor for such a powerful chip.

Source: Chiphell forum (Chinese) and VideoCardz

Source