The alleged specs of among the first Nvidia GeForce RTX 40-series graphics playing cards have appeared on-line. The supply, MEGASizeGPU (opens in new tab), as we speak supplied some particulars concerning the RTX 4090 reference design. Nonetheless, the most important shock was his assertion that two RTX 4080 reference designs had been reportedly brewing in Jensen Huang’s inexperienced cauldron. Earlier than occurring, please do not forget that this rumor might or will not be correct, so add a touch of salt to the broth.
Some reckon that Nvidia’s Ada Lovelace structure GPUs for shoppers might begin to trickle into public view (formally) as quickly as this month. However, first, it could start with high-end launches and releases, similar to it did with the Ampere technology. So it’s with little shock that the primary assured leaks of tech specs rising by way of the Twitterverse, different social media, and tech boards concern purported merchandise just like the GeForce RTX 4090 and RTX 4080.
MEGAsizeGPU’s assertions concerning the RTX 4090 are fairly easy. He says the RTX 4090 allegedly comes with 24GB of GDDR6 and a reference design PCB constructed from 14 layers. The Nvidia RTX 3090 Ti FE board featured a 12-layer design to place the PCB layers spec into perspective. Nonetheless, top-end fashions for essentially the most strenuous calls for (just like the Ok|NGP|N version) got here with 14-layer PCBs to facilitate better energy focus.
The Ada Lovelace specs Tweet will get edgy the place the RTX 4080 is anxious. MEGAsizeGPU claims there are already two “utterly completely different” designs within the pipeline; a 12GB mannequin with ten-layer PCB, and a 16GB mannequin with 12-layer PCB. Each use GDDR6X just like the RTX 4090. Nvidia has put out SKUs which range solely (or primarily) by reminiscence quota in earlier generations, but it surely is likely to be extra logical if this lower-spec GPU is an RTX 4070 (Ti). MEGAsizeGPU later added that the 12GB pattern had a 192-bit reminiscence interface. If we observe by means of with the speculation that the weakest GPU highlighted is the RTX 4070, as it’s too completely different to market as an RTX 4080, that additionally strains up higher with our earlier protection.
In case you are frightened that one of many first high-end RTX 40 playing cards to be launched would possibly come packing a 192-bit reminiscence interface – like earlier gen 060 tier playing cards – please examine our incessantly up to date Nvidia Ada Lovelace every thing we all know function. Briefly, concerning the reminiscence bus, Nvidia is beginning to use a big L2 cache with RTX 40-series playing cards, which can assist make up for uncooked reminiscence bandwidth of their focused gaming resolutions.