I stumbled over a news related to a video on Youtube regardless if a leak, a rumor or just blatancy, it´s a realistic outreach for future. If they add DP2.1, then the PIMAX 12K may have an adequate driver available to release all expected features.
MLID is an inaccurate joke who’s been wrong many times, even about the recent RDNA 3! And deletes all his videos making wrong predictions.
They are not a reliable source.
Even a titan class card, if it did exist, would give you 10% more performance than a 4090 and probably at an even more ridiculous price premium. The 4090 is already ridiculously priced, we don’t need to encourage Nvidia more.
If they have headroom on the 4090 architecture, then I won’t be surprised if we see a 4090ti coming within the next few months. Only hope it has DP2.1
I have no issue with judgement it´s a pointless rumor only, yet. For me, a 4-slot design is just a ridiculous render gag and click bait, so I reviewed some comments only 
, but did not even start the video. 
But, there always is a grain of truth in it.
I believe, a full blown 48GB RAM maybe dual 4090 with internal NVLink card could be a real, but hidden for now roadmap topic for Nvidia. @VRGIMP27 this would be more than 10% , but I agree, 10% plus is not worth it.
Where we have now 4GB low level, and 24GB enthusiast level, 48GB professional level. we will see 24GB will become ubiquitous maybe in 4 years, maybe later.
What would 12K (2*6K) or true 2 * 8K Displays with fair 90° overlap (for realistic stereoscopic view) and 210° FoV horizontal and 150° FoV vertical require to run smoothly? Which for me is the “bionic” approach.
EDIT: typo
I am kind of curious what, if anything, NVidia may do with chiplets, for the next generation…
At least this time around, AMD apparently saw no benefit in having independent (…and maybe specialised), compute units, across which work could be distributed, and executed more or less asynchronously, but made another essentially monolithic chip, with only the memory moved out.
I am curious too… I bet, monolithic design comes to an end, as we already saw with AMD Zen. I wonder when qraphic units with chiplet design will be delivered to end customers. VR with high resolution is demanding, just a specific type of high performance computing. The same time, it is a consumer product. With Apples M1 we already see “chiplet” bonding, two M1 linked together, it is not a unique architecture. as well the cache is a potential candidate (see Zen), and 3D stacked Memory can be a part of it as well. This is what I guess their longterm agenda, both Nvidia and AMD.
I guess, we will see another one or two monolith generation, but in the back end, they prepare for chiplets.
EDIT
Just saw another news about this HPC development, this time from Intel, Intel Breaks Up Graphics Group, Raja Koduri Moves Back to Chief Architect Role, where they distinguish between consumer and
other industrial purpose, the call it data center market.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.