Yeah it’s been a while since I checked the installer but I think there’s some separate packages for Turing and Ansel now that you mention it over the earlier GPU models, would have assumed it would be installed but inactive until a compatible GPU was detected but the way the driver installer has all these packages it’s probably separated entirely and the INF controls what’s installed for a particular device or group here.
EDIT:
https://us.download.nvidia.com/Windows/461.09/461.09-desktop-win10-64bit-international-dch-whql.exe
https://us.download.nvidia.com/Windows/461.09/461.09-desktop-win10-64bit-international-whql.exe
Should start using these when I post the drivers come to think of it, both the DCH and older version.
Hmm Shadow Play and Experience are actually included, camera tool and others.
And there’s NVNGX been a while since I looked through these files.
EDIT:
Had GPU availability been better (Been a thing at all really. ) I’d probably be on a 3080 now the 10 GB VRAM shouldn’t be a major issue for 2560x1440 and the cost for the later revealed and to be released 3080 Super and 3080 Ti variant would probably have been a bit extra in terms of performance and value.
Curious about the 3080 Ti in particular the 3080 S is likely a 20 GB variant but the Ti should be somewhat faster although info so far points to it being somehow squeezed between the 3080 and 3090 still on a 320-bit bus (Thus 20 GB not 24 I guess.) and slightly slower overall speed and bandwidth.
There’s like a 5 - 8% gap unless you give the 3090 a bit more power or use a custom variant that provides that to begin with.
Kinda like AMD’s 6900XT which only Power Color provides a 3x 8-pin for but AMD still enforced a cap so that’s ~300W instead of the 450W it should be able to deliver.
Power Colors response to AMD capping the 6900XT Red Devil to that was probably something like this:
EDIT:
Cards respond well to overclocking but you get 6 - 8% because of the bios and driver caps instead and there’s some overrides for how the GPU handles and scales to voltage changes so yeah.
(Oh yeah that’s 300% gain or more over the 5700 clock scaling … if you omit the actual final percentage so it kinda looks impressive with that 2% versus 6% ha ha. )
GeForce Experience with experimental features probably gets a bit more out of it’s overrides as a result and NVIDIA while strict as well has lifted things a bit more for the custom 3080’s and 3090’s which also show how these scale when not power limited or throttled.
(Well not limited as much as the default at least.)
EDIT:
Entertaining as that is and how it compares AMD also loses a bit from AMD’s own reference pricing being very generous but the custom cards have smaller margins so with prices and such being normalized it’d still be close to the NVIDIA GPU’s so there’s little incentive for a 6800XT over a 3080 as a example.
Reference 6800 had a nice advantage for it’s pricing (Less so for the custom cards once again.) but the 3070’s are close and the 3060’s are likely going to be cheaper so while there’s a bit of a performance gap there’s also a bit of a difference in cost here until AMD’s own 6700 which might have problems against the 3060 depending on how AMD scales this.
Basically at reference pricing AMD has a good bit of performance at a nice lower price though there are also features from NVIDIA and functionality to consider.
With that gone and it now mostly being custom variants this is almost removed entirely.
(Plus overall limited availability though that affects NVIDIA also due to overwhelming demand.)