NVIDIA’s RTX 50 series may be the company’s first MCM-based mainstream GPU – if recent reports are to be believed. Not only that, but Blackwell’s engineering could deliver the biggest performance leap in company history. NVIDIA recently confirmed that the Blackwell GPU architecture will be released in 2024. Since that was the very initial grumble, this post is quite a rumor, so keep those beans handy.
NVIDIA RTX 50 GPUs feature a new SM design and noise reduction accelerator
Now before I begin, I would like to point out that the Blackwell architecture is planned as a successor to the Hopper architecture and that a leak using Blackwell and RTX 50 interchangeably could mean several things. The parallel version of the Hopper in the consumer segment (RTX 40) was codenamed Ada Lovelace, so it’s possible Blackwell could also get its consumer counterpart. Alternatively, it’s possible that Blackwell could actually be a consumer codename (though admittedly, that’s less likely) and the data center side of things would get a new name. Previous reports indicated that Blackwell GPUs will be manufactured according to TSMC’s 3nm process.
The rumor comes from RedGamingTech and they have received some new information regarding the architecture. First, Blackwell will feature an all-new SM chassis. Given the underlying microarchitecture, the switch to the MCM design is being made with Blackwell – not surprising. Also, Blackwell will take advantage of a super fast bus that will connect SM and various chiplets. The noise reduction accelerator will also be part of the ray tracing pipeline (modern path tracing setups don’t actually follow the full sequence, they do it partially and the slider takes care of the rest) significantly improving RT performance.
There still doesn’t seem to be any word on the specs although the source notes that there are several Blackwell GPUs being considered and that much of the binning will depend on how AMD’s current RDNA 3 and future RDNA4 offerings perform. Finally, RGT leaves us with the following teaser: “The biggest performance jump in NVIDIA history.”
NVIDIA Hopper was the world’s fastest 4nm GPU at launch and the world’s first with HBM3 memory. It specs higher than the NVIDIA RTX 4090 (which has 16,384 CUDA cores) with a net total of 18,432 CUDA cores. Blackwell will provide significant generational improvements over Hopper (as has always been the case). Four NVIDIA Blackwell GPUs have already been confirmed in a previous leak.
The architecture of NVIDIA Blackwell was based on David Blackwell who was an American statistician and mathematician who made significant contributions to game theory, probability theory, information theory, and statistics. He was also the first African American inducted into the National Academy of Sciences. Blackwell continues NVIDIA’s trend of naming major architectures after prominent computer scientists and mathematicians and not much is known about Blackwell’s architecture at this point. It’s very likely that we will eventually be introduced to the parallel naming (like Hopper’s Ada Lovelace) that will complete NVIDIA’s first MCM duo.
#Rumor #NVIDIA #RTX #Series #Blackwell #GPUs #Deliver #Biggest #Performance #Leap #NVIDIA #History