Less than 3 months have passed since the release of Nvidia’s RTX 4090 and already rumors are swirling about the next generation Nvidia GPUs; the inevitable RTX 50 series (or some call it the RTX 5000 series). Thanks to a hacker group in South America, who reportedly stole 1TB of Nvidia’s data for ransom, we get our first glimpse of what might be behind the curtain for the next generation RTX 5090s (5080s, 5070s…). You can read more about the alleged hack at Videocardz.
Table of Contents
New Codename: Blackwell
NVIDIA hasn’t officially announced any news about the RTX 50 series. However, data from the alleged hack suggest the next generation RTX 50 is codenamed, “Blackwell”. According to Wccftech.com, Blackwell could be either the consumer codename or the commercial codename. Just as RTX 40 “Ada Lovelace” is the consumer counterpart to the commercial codename “Hopper”.
Codename “Blackwell” is based on the last name of the first renown African American statistician & mathematician, David Blackwell (April 24, 1919 – July 8, 2010). He made significant contributions to game theory, probability theory and some other areas of math/statistics.
RTX 50 Series – 3nm (nanometer) Architecture
Sources around the web seem to agree, NVIDIA will use the 3nm process to build the Blackwell generation of GPUs. This is a notch down (lower the better) from Ada Lovelace which uses a 4nm fabrication process and, a GIANT leap from Ampere RTX 30s 8nm process. This is good news because, smaller architecture allows for more pixel crushing muscle on the same size GPU.
Nvidia isn’t in the same boat when it comes to other chip makers in the industry. In a nutshell, Nvidia is the only processor manufacturer sticking to a “monolithic” approach when building CPUs. All the other big players are moving to a modular CPU build using UCIe (Universal Chiplet Interconnect Express) to connect different chiplets to make a whole CPU. It’s similar to building a PC and adding, for example, a GPU to the motherboard. Chiplets tiles are chosen from 4 categories: compute, graphics, SOC and I/O tiles. They connect to the base CPU tile using the UCIe interface, thus allowing different combinations or “flavors” of a CPU. Why Nvidia isn’t currently moving in this direction is unknown, but we think at some point they will follow suit or do something similar.
Credit: ZMASLO via Wikimedia Commons
Biggest Performance Jump Between GPU Generations, EVER
Rumor has it, the RTX 50 will have the biggest generational performance gap of any Nvidia GPU. Thus far, highest performance gap we’ve seen is the jump from Nvidia 8600 GT to 9600 GT. Nvidia released the 9600 GT with an advertised performance gap between generations at 116 percent. The current benchmarks from UserBenchmarks.com show the gap around 182 percent. Can you imagine 1.82 times the performance of a RTX 4090? Mind-blowing, if you ask me. (Insert jokes here, about RTX 50 series cooling solutions)
Too Early To Say, But…
If these rumors are anywhere close to the truth, high FPS 8K gaming is right around the corner. 4K gaming will be yesterday’s 1440P gaming. Multi monitor setups in 4K will finally get a much-needed reprieve and, the RTX 50 may usher in the era of 8K gaming monitors.
I am currently running a RTX 3080ti with 3 x ASUS PG42UQs 4K OLEDs. Display resolution at 11520 x 2160 is a lot of pixels but, the 3080ti is okay with some tweaks here and there. Sometimes I lower the resolution to 7680 x 1440 which allows me to increase the quality settings quite a bit. While the jump to a RTX 4090 would be a big performance step for my rig, I am choosing to wait and see how the RTX 50 series sizes up. If the performance rumors are correct, it will be a worthwhile wait. Of course, I will always have the option to buy a RTX 4090 at a lower price, if I don’t like the RTX 50’s value proposition.
The release date for the RTX 50 series release date is rumored to be sometime around 2024, a long way away. Plenty of time for more rumors and teasers. We will keep you posted. Let us know in the comments below – Are you planning on upgrading to a RTX 40 or are you waiting; why?