God of War Gets FSR 2.0 Support in Latest Patch

[btarunr]

Sony today released the v1.0.2 patch for the PC release of "God of War," which adds support for AMD FidelityFX Super Resolution 2.0 (FSR 2.0). The latest performance enhancement by AMD lets you experience even higher image quality at a set performance level, or higher performance at a set quality level. It leverages temporal data to add detail to upscaled images, and in our testing with other games, was found to offer comparable quality and performance to NVIDIA DLSS 2.0. The latest v1.0.2 patch for "God of War" is now up on Steam.

Update 20:57 UTC: We've just posted our God of War FSR 2.0 Review.



View at TechPowerUp Main Site | Source

 

[ZetZet]

They're popping up quick, must be fairly easy to implement when going from 1.0 to 2.0

 

[Ferrum Master]

1080ti owners eyes are sparkling

 

[wolf]

Months back they mistakenly updated the game to show 2.0 in the menu's, I think the devs have been keen to add this since hearing about it.

Just tested a little now, quite close to DLSS over all, impressively so even. However some details are a bit more warbled/shimmery/crunchy , like Kratos/Mamir's beard, and finer details like the axe for example, among others. I also get a few more frames with DLSS @ 3440x1440 native.

I'd say for more casual gamers they'd be hard pressed to notice though, so pretty bloody good all things considered, and of course just impressive overall, especially for free.

 

[eidairaman1]

Months back they mistakenly updated the game to show 2.0 in the menu's, I think the devs have been keen to add this since hearing about it.

Just tested a little now, quite close to DLSS over all, impressively so even. However some details are a bit more warbled/shimmery/crunchy , like Kratos/Mamir's beard, and finer details like the axe for example, among others. I also get a few more frames with DLSS @ 3440x1440 native.

I'd say for more casual gamers they'd be hard pressed to notice though, so pretty bloody good all things considered, and of course just impressive overall, especially for free.

This will eventually overtake dlss, and dlss will be abandoned like physx, gsync, sli...

 

[ZetZet]

This will eventually overtake dlss, and dlss will be abandoned like physx, gsync, sli...

So you could say that Nvidia is paving the way and AMD is then improving the feature and making it open source. I actually wonder what would happen if AMD got ahead in the market, would they then try to innovate and make their features exclusive too?

 

[eidairaman1]

So you could say that Nvidia is paving the way and AMD is then improving the feature and making it open source. I actually wonder what would happen if AMD got ahead in the market, would they then try to innovate and make their features exclusive too?

Doubt it, SAM/BAR at 1 point was like SBA in the AGP days.

So you could say that Nvidia is paving the way and AMD is then improving the feature and making it open source. I actually wonder what would happen if AMD got ahead in the market, would they then try to innovate and make their features exclusive too?

And Also another point, AMD didn't have physx or their own version, the Havoc Physics Engine was able to be ran by cpus, rendering phsyx useless.

AMD did bring about Vulkan API though.

 

[wolf]

This will eventually overtake dlss, and dlss will be abandoned like physx, gsync, sli...

Maybe, or maybe something even better will, FSR 2.0 / DLSS 2.3 isn't where this branch of tech just stops getting worked on (especially ML) because FSR became decent and is open, especially since they're so similar to implement, there is little barrier to adding them both for the foreseeable future.

Do I think DLSS will be the premier video game reconstruction/upscaling technique in 20 years? of course not (I'd also doubt FSR will be), but it's been a great tech to 'ride the wave' of so to speak, and the wave continues for now.

EDIT: Also, not all those things are 'abandoned', but for one that definitely was, AMD (everyone...) abandoned it too. (multi GPU)

 

[ZetZet]

Doubt it, SAM/BAR at 1 point was like SBA in the AGP days.


And Also another point, AMD didn't have physx or their own version, the Havoc Physics Engine was able to be ran by cpus, rendering phsyx useless.

AMD did bring about Vulkan API though.

In the AGP days it wasn't even AMD, it was ATi wasn't it?

Yeah, but Vulkan is available for everyone too.

 

[Ferrum Master]

In the AGP days it wasn't even AMD, it was ATi wasn't it?

Yeah, but Vulkan is available for everyone too.

No... you probably don't remember, thus don't understand...

AMD did have a north bridge AMD761 arguably ULI/Ali? I was rocking a dual Socket A board with a pair of Applebread Durons.

, I was rocking ad

 

[watzupken]

Just curious, when is God of War part of Capcom? Isn't it a Sony title?

 

[Sithaer]

They're popping up quick, must be fairly easy to implement when going from 1.0 to 2.0

I would hope so.
This is good news for me cause DLSS was a selling point for me whenever I manage to upgrade to a current-ish gen card between AMD/Nvidia.
Now not so much, RT performance is still better on Nvidia's side but that I don't care much about tbh and not really worth it when I only buy budget-mid range cards anyway. 'looking at 6600 XT/3060 Ti performance level or so'

 

[looniam]

So you could say that Nvidia is paving the way and AMD is then improving the feature and making it open source. I actually wonder what would happen if AMD got ahead in the market, would they then try to innovate and make their features exclusive too?

it was ati/amd that introduced tessellation and nvidia turn that to their own exclusive. so i don't blame them to let nvidia drive. fwiw i'm ok w/NV throwing a ton of R&D money into something and have AMD come by a ~year later and make it for everyone.

seems to be working out ok so far . . .

 

[Chomiq]

Capcom?!

 

[birdie]

This will eventually overtake dlss, and dlss will be abandoned like physx, gsync, sli...

No.

Also PhysX and Gsync have not been abandoned and both are very well alive. SLI is dead but so is CrossFire - D3D 12 and Vulkan made them obsolete.

Lastly PhysX was made open [source] a long time ago, and it may run on the CPU. Sadly due to Epic the future of PhysX 5 (whose SDK is yet to be released) remains uncertain.

 

[Vayra86]

No.

Also PhysX and Gsync have not been abandoned and both are very well alive. SLI is dead but so is CrossFire - D3D 12 and Vulkan made them obsolete.

Lastly PhysX was made open [source] a long time ago, and it may run on the CPU.

PhysX GPU though is where the juice is at. Literally, if you want to calculate naturally flowing juice. Overall, strong physics engines in games definitely aren't taking off. But PhysX implementations most certainly do add something other engines haven't really shown us a whole lot of, especially not running so smoothly. We're not seeing much of it, if any, these days. It really is pretty much dead and CPU PhysX is a whole other ball game, again quite literally. I mean, the last games I remember seeing GPU PhysX in, was Warframe and Borderlands.

As for DLSS... unless Nvidia can magically undo their special hardware requirement, FSR will certainly take over and it will likely push DLSS in the same segment as Gsync right now: a high end segment where you don't really know what you're paying premium for, except minor details that are quite irrelevant but do exist so there is some idea of value. So again, like PhysX, not abandoned, but pretty much dead commercially and therefore eventually also on Nvidia's won't do list. DLSS will exist for as long as Nvidia commercially has a use for it, and this also applies to a per-game implementation.

 

[birdie]

PhysX GPU though is where the juice is at. Literally, if you want to calculate naturally flowing juice. Overall, strong physics engines in games definitely aren't taking off. But PhysX implementations most certainly do add something other engines haven't really shown us a whole lot of, especially not running so smoothly. We're not seeing much of it, if any, these days. It really is pretty much dead and CPU PhysX is a whole other ball game, again quite literally. I mean, the last games I remember seeing GPU PhysX in, was Warframe and Borderlands.

As for DLSS... unless Nvidia can magically undo their special hardware requirement, FSR will certainly take over and it will likely push DLSS in the same segment as Gsync right now: a high end segment where you don't really know what you're paying premium for, except minor details that are quite irrelevant but do exist so there is some idea of value. So again, like PhysX, not abandoned, but pretty much dead commercially and therefore eventually also on Nvidia's won't do list. DLSS will exist for as long as Nvidia commercially has a use for it, and this also applies to a per-game implementation.

1. DLSS SDK is completely free to use and NVIDIA provides extensive support for it.
2. DLSS uses AI image upscaling, so it can be used in situations when FSR 2.0 doesn't work, e.g. in more performant modes.
3. DLSS will always be faster than competing technologies as it uses specialized fixed-function hardware.

Intel's XeSS looks like a real competitor to DLSS but the company has yet to release it. If XeSS proves to be as good and as versatile as DLSS, this is what the industry could settle on, not FSR.

As for me personally, I'm looking forward to someone creating an API which encompasses FSR 2, DLS and XeSS, so, instead of implementing all three, developers could use this library instead as they all work very similarly.

 

[DeathtoGnomes]

They're popping up quick, must be fairly easy to implement when going from 1.0 to 2.0

There were several PRs saying how 1.0 api was a very easy implement, no reason 2.0 would be any different.

PhysX GPU though is where the juice is at. Literally, if you want to calculate naturally flowing juice. Overall, strong physics engines in games definitely aren't taking off. But PhysX implementations most certainly do add something other engines haven't really shown us a whole lot of, especially not running so smoothly. We're not seeing much of it, if any, these days. It really is pretty much dead and CPU PhysX is a whole other ball game, again quite literally. I mean, the last games I remember seeing GPU PhysX in, was Warframe and Borderlands.

As for DLSS... unless Nvidia can magically undo their special hardware requirement, FSR will certainly take over and it will likely push DLSS in the same segment as Gsync right now: a high end segment where you don't really know what you're paying premium for, except minor details that are quite irrelevant but do exist so there is some idea of value. So again, like PhysX, not abandoned, but pretty much dead commercially and therefore eventually also on Nvidia's won't do list. DLSS will exist for as long as Nvidia commercially has a use for it, and this also applies to a per-game implementation.

I believe DLSS is a software API, where as Gsync has the hardware requirement. I'm likely wrong, unless there is something that states DLSS cannot work without having Gsync.

 

[wolf]

no reason 2.0 would be any different

2.0 is completely different technologically to 1.0 and takes considerably more time to implement relative to 1.0. It essentially requires the same engine data, effort, etc as DLSS does, which is why we're very likely to see both side by side in many upcoming games for a while to come.

 

[HABO]

Just curious, when is God of War part of Capcom? Isn't it a Sony title?

Definitely , should be Sony

 

[ratirt]

DLSS SDK is completely free to use and NVIDIA provides extensive support for it.

Free to use? So can you use it on a 1080Ti not to mention 6600XT or Intel Arc graphics?
You cant say free to use since it is only for certain NV cards. Free implementation is more accurate. You can't use it freely if you dont have the hardware to back it up.
You have stated it clearly here as a specialized hardware.

DLSS will always be faster than competing technologies as it uses specialized fixed-function hardware.

As for me personally, I'm looking forward to someone creating an API which encompasses FSR 2, DLS and XeSS, so, instead of implementing all three, developers could use this library instead as they all work very similarly.

That would be FSR 2.0 Any card can use it so if you want this feature like DLSS to be implemented across the board it will be FSR not DLSS 2.0 since the latter one uses specific hardware.
If it turns out, DLSS does not require specific hardware to be utilized, than it literally means NV lied to their own customers having older gen graphics cards and not being able to use that feature which in their case would have been crucial. Obviously not a choice but rather a must for spending money for a new graphics to acquire the feature that could have worked on their older cards.
At least that is how I see it.

 

[DeathtoGnomes]

2.0 is completely different technologically to 1.0 and takes considerably more time to implement relative to 1.0. It essentially requires the same engine data, effort, etc as DLSS does, which is why we're very likely to see both side by side in many upcoming games for a while to come.

This is the same arguement when 1.0 came out, that its similar enough to DLSS to implement. I'll just disagree with this, for now.

 

[ratirt]

This is the same arguement when 1.0 came out, that its similar enough to DLSS to implement. I'll just disagree with this, for now.

That is something I have in mind as well. Always the same story. Considering devs and time they have to spend to implement something that is tailored only to certain audience, they would pick FSR over DLSS. Less time spent and everyone is covered. If the game is a hit that would mean further implementations like DLSS for instance as an exclusive to certain audience so they would know why they have paid so damn much for their cards.

 

[Hircine91]

They're popping up quick, must be fairly easy to implement when going from 1.0 to 2.0

Fsr 1.0 was fairly easy to integrate fsr 2.0 is alot more involved. FSR 2.0 is only integrated onto this game so quickly as the game already has dlss as fsr relies on allot of the same data sets such as motion vectors etc that dlss already requires. also it is likely that the studio had help from amd to integrated fsr 2.0, many other games its likely to be just as hard to integrate as dlss is

 

[Vayra86]

1. DLSS SDK is completely free to use and NVIDIA provides extensive support for it.
2. DLSS uses AI image upscaling, so it can be used in situations when FSR 2.0 doesn't work, e.g. in more performant modes.
3. DLSS will always be faster than competing technologies as it uses specialized fixed-function hardware.

Intel's XeSS looks like a real competitor to DLSS but the company has yet to release it. If XeSS proves to be as good and as versatile as DLSS, this is what the industry could settle on, not FSR.

As for me personally, I'm looking forward to someone creating an API which encompasses FSR 2, DLS and XeSS, so, instead of implementing all three, developers could use this library instead as they all work very similarly.

Right. Its free to use. Just like CPU PhysX. Too bad it won't run on that GPU though that is not an Nvidia branded one of the correct generation

What DLSS really is, just like XeSS if it also requires fixed function hardware (for the good version of it, mind), is an attempt to corner the market with a piece of hardware design. You are correct though, yes, the market could adjust to that. So far, Intel has released 0 discrete GPUs with XeSS in it, the latest news story was that it was in Dolmen-OOPS- never mind it wasn't... So far that ain't going places at all. And DLSS has a relatively small adoption market because you won't be seeing it on any consoles.

My bets are on FSR without a hardware requirement. Did you notice that Nvidia just announced a GTX 1630 sans fixed function hardware? Do you see where this is going? Its a commercial tool, a feature that they will make money out of, whereas FSR is just an option anyone could use. RTX is a way to get people to buy in higher tiers than they might need or want. So far, the bottom end is not served a single GPU that can do RT or DLSS even though it could use it most.

As for your last point, yes, so do I. Let's see where this goes, but Nv history so far doesn't point to that.. and Intel hasn't got any history to speak of other than being possibly even filthier than team green.