Leaked specs highlight big performance gap Updated: Sep 30, 2024 1:35 pm WePC is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more Table of Contents Table of Contents NVIDIA’s next-gen GPU lineup, specifically the RTX 5090 and 5080, codenamed “Blackwell” is rumored to have big specification gains upon launch. However, it seems like NVIDIA is trying to set a pronounced distinction between the two as the uncovered specs highlight a major performance gap between the two. The leaked specifications come from Kopite7kimi — one of the best leakers in the tech scene — who posted them on X. Possible NVIDIA GeForce RTX 5090 Graphics Card specs First on our list is the RTX 5090, the top GPU in the upcoming lineup, featuring the PG144/145-SKU30 PCB design and built on the GB202-300-A1 die architecture. For parallel processing capabilities, intensive computational power, and high-resolution gaming, the GPU will feature 21,760 FP32 CUDA cores instead of the total 24,576 cores and 170 Streaming Multiprocessors (SMs) out of the total 192 SMs. In terms of memory, the 5090 is supposed to be packed with a whopping 32 GB of GDDR7 memory operating across a 512-bit interface. Each GDDR7 memory module is likely to run at 28-32 Gbps, leading to 1.792 or 2.00 TB/s of total bandwidth. The card will also come with a thermal design power (TDP) of 600 watts, indicating NVIDIA’s focus on efficient power management even with high-performance components. NVIDIA GeForce RTX 5080 Graphics Card In contrast, the RTX 5080 is where the performance gap becomes clear. First off, the GPU will be based on the PG144/147-SKU45 PCB and will incorporate the GB203-400-A1 die. It will feature 10,752 FP32 CUDA cores, which is almost a 50% reduction compared to the 5090. The RTX 5080 will be paired with 16 GB DDR7 memory running on a 256-bit bus. Each memory module will run at 28 Gbps but will result in a lower memory bandwidth of 784. The card will be rated at 400W TDP, making it a more power-efficient option. The huge specifications gap between the two GPUs If these leaked specifications turn out to be true, the huge specifications gap can cause NVIDIA some trouble as it did before with the RTX 4090 and 4080. For context, the 4090 was equipped with 68% more CUDA cores than the 4080 which affected the overall performance of the two by a huge margin. Considering the performance distinction we can hope that NVIDIA comes with a balanced pricing strategy for their newer GPUs. However, these are just rumored specs for now and we can’t be 100% sure as early contig can change a lot before the launch. So, it is best to wait for the official announcement from NVIDIA themselves. If everything goes well, we can expect to see the next-gen GPUs together at the CES 2025 exhibition in early January next year.
