AMD FSR ,Extended Life for your NVidia 10 series GPU

DanteErodov

Disciple
Let's see how it compares to DLSS 2.1. Anything close to DLSS & better than TAA upscaling would be good enough due to the higher adoption of this tech.
 

princeoo7

On a Journey called Life :P
Skilled
Let's see how it compares to DLSS 2.1. Anything close to DLSS & better than TAA upscaling would be good enough due to the higher adoption of this tech.
its better than then performance of 2080Ti DLSS and bit less then 3080 DLSS as they are second gen and FSR is first gen. I get around 144fps on 2k in warzone. make this constant and I am happy with FSR implementation.
 

DanteErodov

Disciple
its better than then performance of 2080Ti DLSS and bit less then 3080 DLSS as they are second gen and FSR is first gen. I get around 144fps on 2k in warzone. make this constant and I am happy with FSR implementation.
I am more interested in the quality comparison vs DLSS or existing TAA upscaling & the corresponding performance gains. Getting good image quality with 15-25% increase in performance is good enough in my books specially when compared to significantly lower quality with higher performance gains. Lets wait for Digital foundry analysis videos for that.
 

Stronk

Adept
any thing which is non proprietary is bound to succeed .
Unfortunately, that's not always the case :(
Lookat TressFX, AMD's equivalent of Hairworks. It both looked and performed better than Hairworks in the same demo (or game? I think I'm forgetting) on AMD as well as Nvidia cards, whereas Nvidia looked worse and had huge performance hit on both Amd and Nvidia, tho ofc not as big of a hit as Amd. The performance comparison is with respect to both technologies being switched off. So only time (and ofc money behind the scenes and under the table :p) will tell.
 

princeoo7

On a Journey called Life :P
Skilled
I am more interested in the quality comparison vs DLSS or existing TAA upscaling & the corresponding performance gains. Getting good image quality with 15-25% increase in performance is good enough in my books specially when compared to significantly lower quality with higher performance gains. Lets wait for Digital foundry analysis videos for that.
That is what I said. its around 75% in terms of performance and somewhat less in terms of quality to dlss2.1. this was what leak suggested. but the results are more rock solid then expected since then. I think it will not disappoint us. especially when Nvidia don't even support their old cards for this tech.
 
Finally it's coming! Nvidia brings Gsync for Nvidia cards - amd brings freesync for all cards then Nvidia brings AI upscaling for Nvidia cards and AMD brings it for all cards, again!

As much as I love AMD, their prices have never been competitive in India. I mean how many of you would buy a 6700xt for 46k over 3060ti for 36k? I'm pretty sure no one would, it's just poor value.
 
Unfortunately, that's not always the case :(
Lookat TressFX, AMD's equivalent of Hairworks. It both looked and performed better than Hairworks in the same demo (or game? I think I'm forgetting) on AMD as well as Nvidia cards, whereas Nvidia looked worse and had huge performance hit on both Amd and Nvidia, tho ofc not as big of a hit as Amd. The performance comparison is with respect to both technologies being switched off. So only time (and ofc money behind the scenes and under the table :p) will tell.
so you came with one example which no one cares .

my exampl: open source USB vs Firewire/Thunderbolt
 

Stronk

Adept
Unfortunately the image quality at 1440p quality mode doesn't look that promising IMO. And this is the mode I think most people will be looking forward to the most, specially the ones with budget GPUs. 1080p modes would be significantly worse due to obvious reasons.

Take a look here -> https://images.anandtech.com/doci/16723/AMD_FSR_Example.jpg
To be fair even the image on the left looks terrible to me lol. For some reason that game has always looked pretty bad to me - soft background details and excessively sharpened character details, which is visible here as well
 

DanteErodov

Disciple
To be fair even the image on the left looks terrible to me lol. For some reason that game has always looked pretty bad to me - soft background details and excessively sharpened character details, which is visible here as well
Right, left hand side is also pretty bad but if you look at some of the static object likes the pillar on either side or the textures on the path way, you'll notice that the image with quality mode is a lot more blurry. Then again this is just the first example & I am sure this will vary more in different games & hopefully things will only improve in the future. Looking forward to June 22nd to see more information around this.

What I am worried about is this -> "In our pre-briefing with AMD, the company did confirm that FSR is going to be a purely spatial upscaling technology; it will operate on a frame-by-frame basis, without taking into account motion data (motion vectors) from the game itself.", this mean that it works similarly to how DLSS 1.0 used to work & we know how it ended up looking.
 
Last edited:

SunnyBoi

Branch Manager
Patron
Lookat TressFX, AMD's equivalent of Hairworks. It both looked and performed better than Hairworks in the same demo (or game? I think I'm forgetting) on AMD as well as Nvidia cards, whereas Nvidia looked worse and had huge performance hit on both Amd and Nvidia, tho ofc not as big of a hit as Amd. The performance comparison is with respect to both technologies being switched off. So only time (and ofc money behind the scenes and under the table :p) will tell.
On that note, what happened to Mantle?

FSR looks exciting, lets wait and watch on how effective it is. Especially on older midrange cards.
 
Unfortunately the image quality at 1440p quality mode doesn't look that promising IMO. And this is the mode I think most people will be looking forward to the most, specially the ones with budget GPUs. 1080p modes would be significantly worse due to obvious reasons.

Take a look here -> https://images.anandtech.com/doci/16723/AMD_FSR_Example.jpg
To quote from the article:
Given NVIDIA’s experience with spatial-only upscaling, I’m concerned that AMD is going to repeat NVIDIA’s early mistakes. Spatial is a lot easier to do on the backend – and requires a lot less work from developers – but the lack of motion vector data presents some challenges. In particular, motion vectors are the traditional solution to countering temporal artifacting in TAA/DLSS, which is what ensures that there are no frame-by-frame oddities or other rendering errors from moving objects. Which is not to say that spatial-only upscaling can’t work, only that, if it’s competitive in image quality with DLSS, that would be a big first for AMD.
It is extremely unlikely it will be anywhere close to DLSS 2.0 which uses temporal upscaling with AI training and deep learning. Note that DLSS 1.0 which itself was considered poor was based on spatial upscaling with neural network models. Technically, this doesn't even match up to DLSS 1.0, so it would be a miracle if AMD can even meet DLSS 1.0 quality without any sort of machine learning. If the still images look bad, one can only imagine moving objects to be significantly affected without temporal data. It seems they are simply targeting a frame rate boost at different resolutions at the expense of quality.

This is certainly good for old card owners as it provides some sort of upscaling. However, hope that it doesn't come at the expense of DLSS implementation in games since this will have a lot of catching up to do over the next few years.
 

DanteErodov

Disciple
To quote from the article:

It is extremely unlikely it will be anywhere close to DLSS 2.0 which uses temporal upscaling with AI training and deep learning. Note that DLSS 1.0 which itself was considered poor was based on spatial upscaling with neural network models. Technically, this doesn't even match up to DLSS 1.0, so it would be a miracle if AMD can even meet DLSS 1.0 quality without any sort of machine learning. If the still images look bad, one can only imagine moving objects to be significantly affected without temporal data. It seems they are simply targeting a frame rate boost at different resolutions at the expense of quality.

This is certainly good for old card owners as it provides some sort of upscaling. However, hope that it doesn't come at the expense of DLSS implementation in games since this will have a lot of catching up to do over the next few years.
Again, it is too early to form any opinion although I am also a little bit skeptical regarding how it will work. Regardless, if they are able to improve the quality even compared to the native reduced render resolution with equivalent performance increase, it will still be a big win specially if it provides better results than TAA upscaling. And I am sure AMD wouldn't be showcasing this too much if there were no improvements.
 
Top