Nvidia’s current RTX 4000 generation is far from fully released, but the first rumors about the successor RTX 5000 are already doing the rounds. They are from the Youtube channel RedGamingTech and should be used with caution. Exciting and partly also plausible key data are in it.
The Performance plus compared to the current Ada Lovelace generation should be with the upcoming Blackwell graphics cards at a proud 100 to 160 percent. However, these values have to be considered with restrictions for several reasons, from the factor
rumor not to mention.
According to the leaker, they apply to both gaming and professional cards from Nvidia, although the performance increase in games should be in the lower range (i.e. roughly 100 percent). In addition, it is completely unclear to which comparison exactly they refer.
Is it performance with ray tracing or without? Is DLSS included in a possibly new, even faster generation or not? And which hardware exactly is being compared here? Regardless, the specifications mentioned read quite impressive.
Nvidia RTX 5000: That should be the key data
In addition to more general information such as a
massive remodeling of architecture and one
new structure of the streaming multiprocessors, RedGamingTech also provides supposedly more solid data on the RTX 5000 generation. They are as follows in relation to the gaming top model:
- Manufacturing at TSMC in 3nm
- monolithic design (one chiplet instead of multiple chiplets)
- 384-bit (1,728 TB/s) GDDR7 memory
- 18,432 shader units
- 96 MB L2 cache
- over 3.0 GHz GPU clock
For comparison, a few key data of the current top model RTX 4090: It relies on slower GGDR6X memory with lower memory bandwidth, fewer shader units (16,384), less L2 cache (72 MB) and lower clock speeds in the 2.5 GHz range.
The RTX 4090 thus achieves a performance increase of around 60 to 70 percent compared to an RTX 3090 Ti. However, if DLSS 3 or the frame generation, which is only supported with the RTX 4000 generation, is added, the performance increase increases to percentage values in the three-digit range.
How plausible is the information?
In principle, the information given sounds quite realistic, even if we see a big question mark behind the performance plus mentioned at the moment.
On the other hand, it is no secret that new chip generations are being manufactured by TSMC in 3nm. Nvidia is also (still) doing very well in the gaming sector with the monolithic design, GDDR7 memory is expected for the RTX 5000 and RX 8000 anyway and the growth mentioned shaders, L2 cache and clock should be within the realms of possibility for the Californians.
But that does not mean that it will actually happen, especially since the release of the new Nvidia generation is still in the distant future. According to the current status, the Blackwell models can be expected at the end of 2024 at the earliest.
Accordingly, we have to be patient for a long time before we can talk about officially confirmed data instead of rumours. After all, manufacturers are reluctant to let their cards be looked at at an early stage.
How do you rate the current rumors about the new RTX 5000 graphics cards? Quite plausible and exciting, completely uninteresting since it’s just rumors anyway, or something in between? Feel free to write it in the comments!