It's obviously completely wrong. I can get near double the PS5's performance on the 4070ti at PS5 matched settings (so way more at settings properly optimised for the Ada architecture) and the 4070ti is only about 30% faster than the 3080 in this game.
Ratchet & Clank technical analysis *spawn
- Thread starter Inuhanyou
- Start date
It's obviously completely wrong. I can get near double the PS5's performance on the 4070ti at PS5 matched settings (so way more at settings properly optimised for the Ada architecture) and the 4070ti is only about 30% faster than the 3080 in this game.
Flappy Pannus
Veteran
Looks like the Nvidia Directstorage controversy may soon be...gdeflated (sniper bullet pierces my apartment window)
Good, so it wasn't all Nixxes. AMD's drivers caused RT to crash and NVIDIA's drivers are causing performance issues with Reflex and DirectStorage.
Awesome news. So hopefully nothing fundamentally wrong with DS in terms of GPU performance impact after all. Can't wait to test this!
arandomguy
Veteran
What happens if you just toggle off Reflex currently in the option menu?
Benchmarking with Reflex on always had a caveat in that while Reflex lowers input lag it can actually drop performance.
Benchmarking with Reflex on always had a caveat in that while Reflex lowers input lag it can actually drop performance.
Last edited:
It seems to slightly increase performance for me.
Jensen lowers the scope from his eye, another day in the trenches of the GPU warz.
Flappy Pannus
Veteran
Should note that I missed the detail in the driver notes - Nvidia specifically refers to Directstorage performance in R&C with Reflex enabled.
I don't use reflex in this game as it has a slight performance hit, but DS still has a significant performance hit all the same, with or without it enabled. So no guarantee this next driver update will actually 'fix' the issue of DS causing GPU drops in general.
Nvidia open issues said:
I don't use reflex in this game as it has a slight performance hit, but DS still has a significant performance hit all the same, with or without it enabled. So no guarantee this next driver update will actually 'fix' the issue of DS causing GPU drops in general.
The NVIDIA rep specifically quoted this post though:
To which he replied that there is a fix. The implication is that this is a separate issue from DS with Reflex and that they will fix it in the next driver update. It could also mean just DS with Reflex though. He was asked to clarify but I don't think he's answered.
Certainly Reflex has a nasty performance hit with DS enabled. I just tested this in one area and had frame rate swing from ~80 to the low-mid 60's in a repeatable sequence with Reflex on while turning it off increased those lows by 10fps. That's without even having to reset the game. I didnt test much just now without DS (time constrained) so it wasnt clear whether performance was better again that the DS on/Reflex off results.
DavidGraham
Veteran
New patch is out, ray tracing on AMD hardware is now back up and running.
davis.anthony
Veteran
From the release notes
- Visual improvements to some ray-traced reflections and ray-traced shadows
I wonder if these are more artistic changes where the RT effects have altered a scene is a way that was not intended by the artists? Do you have a save file at this point by any chance to see if it it's any different now?
Flappy Pannus
Veteran
LOD bias bug after fresh boot...fixed! Well, mostly.
As you can see, no difference. And it's just not a case of the DLSS reset now still sticking with the wrong bias, as you can see here from previous patch to this one (you'll have to zoom in for the boot details but then it becomes readily apparent).
(Also ensured I wasn't forcing a negative lodbias through nvinspector before these tests)
The 'mostly' part? Still has the mode-switching ballooning your vram by ~1GB. So on a 12GB card if you play around with DLSS settings, even going from fixed to dynamic res, VH textures will start to choke your card until you reload. So the previous recommendation to dial in your settings and then don't play with them after a reboot still stands.
But no need for lodbias hacks or dlss resetting after loading anymore.
Also further testing, the water caustics issue has not been fixed yet.
As you can see, no difference. And it's just not a case of the DLSS reset now still sticking with the wrong bias, as you can see here from previous patch to this one (you'll have to zoom in for the boot details but then it becomes readily apparent).
(Also ensured I wasn't forcing a negative lodbias through nvinspector before these tests)
The 'mostly' part? Still has the mode-switching ballooning your vram by ~1GB. So on a 12GB card if you play around with DLSS settings, even going from fixed to dynamic res, VH textures will start to choke your card until you reload. So the previous recommendation to dial in your settings and then don't play with them after a reboot still stands.
But no need for lodbias hacks or dlss resetting after loading anymore.
Also further testing, the water caustics issue has not been fixed yet.
Last edited:
Awesome news! I Assume no significant performance impact that you've noticed?
Shame about the water caustic. I'm wondering if it's less a bug and more a deliberate implementation decision. Perhaps like AMD RT it wasn't working properly so they chose to leave it out at least for now. There is a water ripple texture applied so they've clearly gone to some effort to put something there in its place.
Flappy Pannus
Veteran
No additional performance hit from compared to resetting the lod bias in the previous patch, but there is a hit vs. the wrong lod before as you might assume, simply because it's drawing in more.
The water ripple texture exists in the PS5 version too, this is just missing the additional wave effect.
*Sigh*
Edit: Ok, it was my mistake. There definitely WAS issues texture streaming issues... but I forgot that I updated my BIOS and my RAM wasn't running XMP and my CPU OC wasn't there.. lol
There's no texture streaming issues anymore, and the performance is where it should be again.. my bad lol
Last edited:
DavidGraham
Veteran
The gap between AMD and NVIDIA in RT performance in Ratchet and Clank is significant, at 1440p native, the 3080Ti is 66% faster than 6900XT, at 2160p, it's 50% faster. So it shrinks with resolution. The 4070Ti is also 40% faster than 7900XT at both 1440p and 2160p.
gamegpu.tech
Ratchet & Clank: Rift Apart v1.808: PC Performance Benchmarks for Graphics Cards and Processors | Action / FPS / TPS | GPU TEST
In our latest review, we tested graphics cards and processors at maximum graphics settings to show you how the new patch for Ratchet & Clank: Rift Apart v1. 808 affected game performance with the intr
The gap between AMD and NVIDIA in RT performance in Ratchet and Clank is significant, at 1440p native, the 3080Ti is 66% faster than 6900XT, at 2160p, it's 50% faster. So it shrinks with resolution. The 4070Ti is also 40% faster than 7900XT at both 1440p and 2160p.
Ratchet & Clank: Rift Apart v1.808: PC Performance Benchmarks for Graphics Cards and Processors | Action / FPS / TPS | GPU TEST
In our latest review, we tested graphics cards and processors at maximum graphics settings to show you how the new patch for Ratchet & Clank: Rift Apart v1. 808 affected game performance with the intr
Not unexpected at the max RT settings but a comparison at console level RT would be really interesting (relative to no RT).
Flappy Pannus
Veteran
Nvidia Hotfix driver for DS in R&C does improve things, but not much.
