Leaked: Final specs for Nvidia’s subsequent-gen RTX 5080 and 5090 GPUs

Final week, the leaker identified as kopite7kimi — identified for his neatly suited predictions — revealed on X (formerly Twitter) the seemingly final specifications for Nvidia’s subsequent-gen RTX 5080 and 5090 GPUs.

The originate of the unusual 5000 series has also been scheduled, with Nvidia planning to unveil the graphics playing cards in precisely a few months at CES in January 2025. The first RTX 5080 and 5090 graphics playing cards are expected to be on hand for rob in February.

The greatest change compared to what we previously knew referring to the unusual high-of-the-line RTX 5090 model is an upgraded memory interface for a whopping 32GB of video memory, which is now 512 bits broad (as a change of 448 bits). Nvidia has also elevated the different of shader items to an improbable 21,760.

In actual fact, Nvidia can equip the GB202-300-A1 graphics processor with 24,576 shader items in fats growth, nevertheless the fats computing energy of the Blackwell chip is presumably reserved for expert playing cards (which Nvidia can promote at a vastly better tag point).

Nonetheless, it’s also seemingly that Nvidia will add an RTX 5090 Ti or Titan in the direction of the tip of the 5000-series product cycle — if the aggressive effort demands it, which isn’t too seemingly for PC graphics playing cards.

RTX 4090 vs. RTX 5090, compared

Alongside with the massive make greater in computing energy, the Total Graphics Vitality (TGP) — i.e., the utmost graphics card energy consumption at maximum utilization — has sadly also elevated from 450W to a crazy 600W in the unusual RTX 5090.

The technical specs of the Nvidia RTX 5090 promise a large make greater in performance compared to the already-very-hasty RTX 4090, which is why we take that the RTX 5090 will likely be an costly graphics card.

The RTX 4090 FE launched with an MSRP of $1,599 nevertheless in all equity necessary very no longer going to search out at that tag anymore. You’re looking out for to exhaust upwards of $2,000 or more for an RTX 4090 this exhibit day, so there’s no reason to interrogate the quicker and better-geared up RTX 5090 to debut with an MSRP below $2,000.

Since Nvidia has most steadily monopolized the graphics card market with a market portion of 88 p.c and is technically some distance earlier than their closing final competitor in AMD, Nvidia can nearly tag the 5090 nonetheless they wish. Prepare your self for a hefty tag make greater, provided that it’s going to be the “fastest graphics card on this planet.”

RTX 4080 vs. RTX 5080, compared

kopite7kimi also announced the expected technical specs of the RTX 5080, which comes with a smaller leap in performance over its predecessor. The different of streaming processors is going up by spherical 10 p.c from 76 to 84, and the memory connection is final the an identical (even though the 5080 is getting the quicker GDDR7 video memory).

That acknowledged, it’s sophisticated to estimate how noteworthy the RTX 5080 will in actual fact be. In purely technical terms, it has 10 p.c more shader items, nevertheless the unusual graphics chip architecture must even be more efficient and the quicker video memory must present extra performance. All in all, we interrogate an total performance bump of between 15 and 25 p.c.

When put next to its predecessor, the energy consumption of the RTX 5080 is only being bumped up by a moderate 50 watts. As for the expected tag of RTX 5080 playing cards? That’s serene up in the air.

What’s most necessary here is the chasm between the RTX 5080 and 5090, which offers almost twice as necessary computing energy. Nvidia has by no manner had this kind of sizable difference between an 80 and 90-series chip sooner than, which is reason to tackle the leaker’s figures with caution.

Further studying: The very best graphics playing cards for every funds

Creator: Michael Schmelzle, Head of Hardware & Checking out, PC-WELT