Nvidia GeForce RTX 5090 tape-out rumors: smaller than the 4090, 448-bit bus, monolithic die

Nvidia's RTX 5090 might feature a two-slot cooler, 448-bit bus, and 28GB of GDDR7 VRAM.

: Rumors suggest Nvidia's GeForce RTX 5090 will be smaller and more efficient than the RTX 4090, featuring a two-slot design and possibly less power consumption. The RTX 5090 is expected to have a 448-bit bus and 28GB of GDDR7 VRAM, offering significant performance improvements. Nvidia plans a late 2024 launch for the RTX 5090 and 5080, potentially facing stiff competition from AMD and Intel.

Nvidia's upcoming GeForce RTX 5090 graphics card is reported to be smaller than its predecessor, the RTX 4090, possibly indicating a shift in design philosophy towards more efficient power use and cooling solutions. According to leaks, the RTX 5090 will adopt a two-slot, dual-fan design as opposed to the three-slot cooler seen on the RTX 4090. This could point to a reduction in the GPU's power requirements, despite old rumors of a 600W card. The monolithic die approach indicates that Nvidia is maintaining a traditional single die design, differing from AMD's chiplet approach in its latest offerings.

The rumored specifications for the RTX 5090 include a memory interface of 448 bits and 28GB of high-speed GDDR7 VRAM, which would theoretically allow for up to 1.5 TB/s memory bandwidth, a 50% increase over the RTX 4090. The GPU is expected to utilize 14 of its 16 memory modules based on the new dense memory configuration. Such a configuration supports predictions of substantially enhanced performance capabilities over the current generation, potentially reinforcing Nvidia's position at the pinnacle of consumer graphics technology.

Looking forward, the GeForce RTX 5090 and its sibling, the RTX 5080, are slated for a late 2024 release with the 5080 potentially arriving earlier. They are set to compete with upcoming releases from AMD's RDNA 4 GPUs and Intel's Arc Battlemage GPUs, adding to the high stakes in the consumer GPU market. As these launch dates approach, the exact specifications and performance benchmarks will become clearer, providing a better understanding of how these cards will stack up against current and future competition.