nVidia's gaslighting us - the '3' part of DLSS 3.x *is* the frame insertion - upscaling and Reflex were/are already a part of DLSS 2.x - all they've committed to is continuing to revise the upscaler. Personally I don't believe that such a significant change to DLSS would have been engineered *without* backwards compatibility - nVidia's simply decided that doubling the frame rates of the 20xx and 30xx series isn't in their financial interest.DLSS 3 will include enhancements to existing DLSS, these enhancements will work on 'all' RTX cards. It is the specific frame insertion feature (that many are sceptical about anyway) that will be exclusive to 4000 series.
4080 and 4090 include a much nicer encoder which is now dual core and natively supports AV1, so seems Nvidia recognised their huge filesize problem with the added AV1 support. Media I think arent talking about this anywhere, only know its updated due to the Nvidia slides. They always just seem obsessed with rasterization performance.
4080 16 gigs version has a nice profit margin on the VRAM, extra $300 for an extra 4 gigs VRAM although they also giving it 4090 cooler spec and some small amount of extra cores.
Yes - the AV1 encoder is nice to have - 40xx is a decent die-shrink and refresh of Ampere and seems to deliver the kind of performance you'd expect from the changes/improvements - it's never going to be 2-4x the speed of a 3090 though except in extremely edge cases.
The 12GB is a straight rip off - it's a xx70 series card that nVidia's selling at near double the MSRP it should be. The only card with any real value in the lineup is the 4090 unfortunately.
Last edited: