The RTX 5090 Is Slowly Coming Into View
Published: 06-05-2024
Image Source: VallahExperte
While all we have are rumors at this point, the RTX 5090 is inevitable and it’s likely close. It’s unclear if the 5090 will be the first 50-series card to debut, but details about NVIDIA’s (likely) biggest gaming and enthusiast GPU are starting to come out of the woodwork.
Everyone expects the cards to officially arrive in October or November of 2024, and for the 5090 and 5080 to debut together.
That massive prototype 4090 Ti cooler above, posted by VallahExperte, was refuted by famed leaker Kopite7kimi, who claims that the 5090 will be a dual-slot, twin fan card in its vanilla Founder’s Edition form. That’s big if true, but what else are leaks suggesting about Nvidia’s upcoming apex GPU?
Rumored RTX 5090 Specs
The RTX 5090 is expected to use NVIDIA's flagship GB202 chip based on the Blackwell architecture. It could feature a 448-bit memory bus with 28GB of GDDR7 VRAM. A narrower bus than the 4090, but with faster memory and more total bandwidth at 1568 GB/s.
Maybe some of the best news is that the 5090 (and presumably other 50-series cards) won’t be using the infamous melting 12VHPWR connector. Instead it’s likely to be a new more robust 16-pin design.
When Do We Get the RTX 5090?
As we said, the rumors point to a release late in 2024, but it’s not actually clear if the 5080 or 5090 will launch first. The 5080 will obviously be a higher-volume seller, and even if the 5090 turns out to be a cut-down Blackwell die itself, it’s still likely to have lower volume production or more production issues than smaller chips in the family.
How Fast Do We Think the RTX 5090 Will Be?
The truth is no one knows, but based on the leaked specs, some people are expecting as much as a 70% leap on performance.
That would be one of the largest generational leaps in history for GPUs, but it’s hard to imagine this panning out in real-world performance, since most CPUs struggle to keep up with the 4090. Unless there’s a similar leap in CPU performance by the end of 2024, even such a huge leap would see limited applications.
How Much Could The RTX 5090 Cost?
NVIDIA is essentially in a position where it can charge whatever it wants for its apex card, but analysts seem to peg the MSRP anywhere between $1700 and $2500. As usual, that is pretty expensive for a consumer card, but a true bargain in the context of workstation cards. Especially with that large rumored VRAM allocation, which should make professional AI and similar workloads feasible.
We can’t wait to get our hands on the next generation of NVIDIA GPUs, and perhaps we’re even more excited about how big the leap will be for mid-range GPUs, or the jump for laptop GPUs, thanks to new power efficiency gains. Either way, it won’t be long now!
