Whilst DLSS works well in a small number of titles and specific trained resolutions, CAS sharpening has demonstrated that you can get just as good results without dedicated hardware. I hope that the silicon space waisted on Tensor cores is used for more RT cores in future GPUs.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Nvidia DLSS 5 months on-a win or fail?
- Thread starter easyrider
- Start date
So no more ray tracing then? cause tensor cores run ray tracing and dlss
if they abandon tensor cores, they need to find a different way to do ray tracing
Nope, RT cores handle they ray tracing. It's been public knowledge since the launch, if not before.
Tensor cores are for deep learning and inference operations. So, limited use in a gaming card except for DLSS. There is a small chance they could be used for AI enemies, but I’d rather have more dedicated RT cores instead.
Google tensor core denoiser
Thank me later
Without the tensor cores, your rayvtraced game would output an image that looks like you took LSD and now all you see is fuzzy dots
Associate
- Joined
- 3 Nov 2010
- Posts
- 752
What Gpu do you have to give such words of wisdom. I assume it must be an rtx and that you indeed have first hand experience in such matters
What is it with you and needing to have first hand experience?
Imagine if something got a very bad review, would you still insist on buying the said item to try yourself so you can get “first hand experience”?
so i just tried this nvidia sharpen.
i compared 3840x1080 with sharpening filter 50% to 5140x1440.. so super ultrawide 1080p vs 1440p
kcd
and wow. the different is miniscule. This is going to provide so much life to graphics cards which can use it.
the two images looks basically identical but one was running at 80fps, the other at 50fps.
i compared 3840x1080 with sharpening filter 50% to 5140x1440.. so super ultrawide 1080p vs 1440p
kcd
and wow. the different is miniscule. This is going to provide so much life to graphics cards which can use it.
the two images looks basically identical but one was running at 80fps, the other at 50fps.
Whilst Tensor cores can in theory be used for denoising images it doesn’t work in real-time. No games have used it yet and are instead using traditional filters as part of the compute shader activity. Nvidia’s own frame breakdown analysis shows that Tensor cores are only being used for DLSS.
My graphics card is in my sig. It works perfectly for comparing screenshots of DLSS vs CAS. Don’t need pointless Tensor cores to do that.
From the top of my head Hardware Unboxed have said it sucks on more than one occasion. They have been around for a while and show objective comparisons between DLSS and using a lower resolution/sharpening and are not just making stuff up.
Thing is, obviously the quality of DLSS varies a lot from game to game. Some implementations are better than others. From what I have seen online Control has released with decent DLSS which is encouraging and bodes well for future 2020 titles I am interested in playing which will have RT/DLSS.
I just don’t get why people get upset and defensive about all this RTX stuff. If you are happy with being an early adopter and happy with an handful of games that are out that make good use of the tech, then great. But not everyone is the same and won’t have the expectations/requirements once people get this you won’t feel so hurt.
I am personally excited about RT/DLSS because of it being supported on quite a few games I am interested in playing next year. Could not give a rats arse about the handful of games released this year. By next year I am hoping they will get better with it all and have much more RT cores on the 3000 series.
I will admit, the DLSS on Control looks pretty good, certainly the best implementation.
However, I still stand by that I'd always prefer to have either:
a) A generally better performing card in the first place for the 50% cost increase.
b) Option of a non RTX card without the price hike.
Instead of what we have now with a poorly implemented answer to a question nobody asked, paying 50% more for the privilege.
However, I still stand by that I'd always prefer to have either:
a) A generally better performing card in the first place for the 50% cost increase.
b) Option of a non RTX card without the price hike.
Instead of what we have now with a poorly implemented answer to a question nobody asked, paying 50% more for the privilege.
Associate
- Joined
- 3 Nov 2010
- Posts
- 752
What is it with you and needing to have first hand experience?
Imagine if something got a very bad review, would you still insist on buying the said item to try yourself so you can get “first hand experience”?
ok then its like someone without a HDR TV and giving an opinion on how bad HDR is based on screen shots they have seen rather than actually experiencing it first hand. So yes its pretty relevant in my book. Reviews are also very subjective
Well spotted Sherlock
Why is it that those who think RTX is market ready tend to fall in to the group that doesn't understand the tech, or sits 7' from the screen?
Yea, I think you just don't want to see what I am sayingok then its like someone without a HDR TV and giving an opinion on how bad HDR is based on screen shots they have seen rather than actually experiencing it first hand. So yes its pretty relevant in my book. Reviews are also very subjective
What you are saying is reviews are subjective, other people who own an rtx card who says the same are subjective, but if sunshine here says it is the best thing since slice bread then that is objective. Sounds about right?
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,080
- Location
- In a house
It doesn't look too bad in SOTTR either to me, yeah looks a bit softer (noticed when running the benchmark), but actually just playing the game, i doubt id notice tbh.
ok then its like someone without a HDR TV and giving an opinion on how bad HDR is based on screen shots they have seen rather than actually experiencing it first hand. So yes its pretty relevant in my book. Reviews are also very subjective
If the 2080 Ti had arrived without the extra silicon to do RTX it would sell for about £650 to £700 (same as 1080 Ti price point).
Question to you...
Is having RTX on a 2080 Ti worth the extra £400 over the cost of a card without it?
Well spotted Sherlock
Why is it that those who think RTX is market ready tend to fall in to the group that doesn't understand the tech, or sits 7' from the screen?
It is a good spot, makes those opinions or observations less factual and more of fiction.