Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
Man these 546.31 drivers are worse for me. Latency has doubled using them. The 546.33 drivers may of had issues but performance for me was good. Well only if using my set framerate limit to 1/2 refresh. I did a DDU uninstall and oddly enough sometimes you just have to install a driver twice. So about to try installing again.
OK that didn't work either. So went back to 546.33 and yup had to use 1/2 refresh for framelimit and now back to smooth playing. Guess i go back to 546.29 and see how that is. This is so odd.
546.29 Frame Generation is ON. Vsync On in control panel. This is a stuttering mess. Only fix is setting in game framerate limiter to 1/2 refresh. for me that's 60/2=30. Runs very smooth that way for me. Frametime is doubled in Menus as it should be as framegen doesn't work there and we have framelimit set to 30. So feels sluggish cause it is. So working as it should.
546.31 Hot Fix driver that specifically fixes Vsync. Frame Generation is ON. Vsync On in control panel. OK well it actually does fix the Vsync issue. You don't have the stuttering mess. However it's absolutely horrible at sending a signal to the monitor or in my case my LG TV. Using Nvidia's FrameView i'm getting up to 600ms Latency. This is 4x what the other 2 drivers are getting. Average was 340ms or so. More then 2x what the other 2 drivers are getting. Not to be confused with frametime btw as they are not the same thing. So for me this driver is NOT to be used.
546.33 Frame Generation is ON. Vsync On in control panel. This is a stuttering mess. Only fix is setting in game framerate limiter to 1/2 refresh. for me that's 60/2=30. Runs very smooth that way for me. Frametime is doubled in Menus as it should be as framegen doesn't work there and we have framelimit set to 30. So feels sluggish cause it is. So working as it should.
I find it odd how going back from 546.33 to 546.29 fixes issues for some people when they both act the exact same for me. But this is very common. We all got different hardware. I know a lot of people really like the 546.31 Hotfix driver. However i'm betting it's not people playing cp2077 forcing Vsync when using FrameGen. Horrible driver for that. 546.29 probably the best choice atm and with the instructions i gave earlier if needed.
Doing a VSync test with Frame Gen using 546.29 drivers below as an example with my i7 2600k, ASUS P8z68-V Pro Gen 3, ddr3 2133 ram + Rtx4080. Noting the Min fps especially at 52.69fps is pretty good. With the average being 59.95fps
https://steamcommunity.com/sharedfiles/filedetails/?id=3125491819
No worries, I haven’t setup properly. I think Gsync/Freesync/adaptive sync is considered the same as not using Vsync. They all allow your monitor to update at a variable rate instead of a fixed 60 times per second (for 60Hz). Sorry you’re having issues with the driver.
That build is interesting, how does one end up with a 4080 paired with Sandy Bridge i7? Aren’t you constrained by your PCIe bandwidth (and your CPU)? Not meaning to be disrespectful, I appreciate your help with my questions, and your GPU is better than mine (I only have a 4070).
I had a gtx970 in it previously. Basically i bought an 13600, Gigabyte rtx4080 and Z790 Aorus Elite AX mobo, Corsair ddr5, Seasonic 1k psu, new case, 2 radiators and 6 140mm fans, water pump and tank. I just never got around to finishing the build. I still need hard water tubes and fittings. Probably get a pci-e riser card for gpu so it's vertical and not horizontal. The water block for the RTX4080 is so heavy. So started buying parts Oct 2022-current. They were just collecting dust. So put the new psu in old system along with the rtx4080 when i got it.
Yes bandwidth severely hampered. pci-e 2.0 only. I have an nvme samsung 980Pro 2tb in a pci-e slot via an adaptor and using a modded bios just for that. cpu and mobo only have 16 pci lanes. nvme benches about 1,500mb/s only if i have gpu use 8 lanes and nvme use 4 lanes. 4 left with nothing to do. But then gpu is hampered a bit. Or like i'm using it now gpu uses all 16 lanes and nvme shares 1 of them with the gpu but can only bench at 417mb/s. If i set it to use x4 mode the gpu goes to x8 mode. Pretty sad for an nvme that can and should do 7,000mb/s. So yes hugely constrained here. I'm shocked how well it actually works with the limitations i have. Amazing results in cp2077, Starfield, BG3. or most any game that has frame generation. Hogwarts did decently. But Elite Dangerous=no framegen=forget it 37fps on planets. Playing Red Dead Redeption 2 atm and no framegen. So having to lower the cpu specific settings in which i've found 3 to maintain 60fps and i have.
I'm just lazy in my retirement. And no game yet has forced me to finish my build. RDR2 forcing me to lower cpu related settings may get me off my but though. I don't lower crowd control the cpu setting in cp2077 as i don't need to. It's all good. And results without Vsync are shockingly good below. Ultra settings+Frame Gen but no dlss or ray tracing. Just a cpu test more or less.
https://steamcommunity.com/sharedfiles/filedetails/?id=3111438775
Oh, so it's just a temporary holdover for the time being, that's understandable, especially since you're next build is using water cooling. I don't know much about proper water cooling, I have an AIO for my 5800x3d but that's not really the same thing at all. I've always been curious about water cooling but it seemed expensive and complicated when I looked into it before I built my PC. Thanks for explaining. And hey, if your current setup is working, no need to rush the build if you're still sourcing parts and your current one gets the job done well enough for you to be satisfied.
On my PC, with a 5800x3d + 4070, Path Tracing at 2954x1231 (77% of my monitor's native resolution of 3840x1600), performance is only slightly worse than Psyco RT mode. I get about 80-90fps using DLSS Balanced + FG.
I set it at that resolution because it's 3.6 million pixels, roughly the same as 1440p. Native resolution of my monitor is 6.1 million pixels, which is less than 2160p/4k but still high enough to cause problems in some games (especially using path tracing).
https://steamcommunity.com/sharedfiles/filedetails/?id=3121414560
https://steamcommunity.com/sharedfiles/filedetails/?id=3118125014
Still i think it's pretty amazing. Note overdrive does better i think because it put dlss in auto so may have swapped from quality to balanced at times i'm guessing. I think i did a Ultra, no upscale test , Ray tracing, path and reconconstruction lemme check. Hm don't see one. I'm betting the gpu hits 100% now. Guess i run another bench. And ah Ray reconstruction only works with dlss enabled.
https://steamcommunity.com/sharedfiles/filedetails/?id=3126384799
cpu only hit 78% gpu 88% i just ran out of bandwidth. So not playable at 50fps. Would be so much worse in game in Dogtown especially. Interesting test in that it clearly shows my bandwidth limitations.
I alternate back and forth about whether the path tracing mode is worth using or not. I’ve been mostly playing that way lately, but changing back to standard RT ultra mode still looks really great too. Path tracing is cool and I’m glad my PC can do it, but I wonder if someone changed my settings to regular RT and didn’t tell me, would I even notice that I wasn’t using it?
Merrydoc, What exactly do you mean by "when i switch my monitor"? You mean literally switching focus to another monitor or moving the game screen from 1 monitor to another? Have you tried disabling fullscreen optimizations? C:\Program Files (x86)\Steam\steamapps\common\Cyberpunk 2077\bin\x64\ cp2077.exe properties/compatibility/ check disable fullscreen optimization hit apply then ok. Always worth trying that but usually to remove stuttering in games. If you're already using it try unchecking instead and see if that fixes the crashing.
dpi scaling=same results (no worries here)
Windows Game Mode Enabled=same results or just possibly a minor gain (may as well turn it on)
Disable Fullscreen Optimizations=same results (not needed)
Disable MPO Multiplane Overlay=PC latency doubles in borderless/full (don't disable MPO)
It may be ok to disable MPO with new drivers. Note that nvidia is who put out the .reg files to disable MPO when using their drivers years ago. I'm pretty sure they fixed that a while back. But now disabling it makes it worse. Seems very odd to me. Check my next post for another test i did.
33% increase in FPS for CP2077 with HPET=Disabled
https://steamcommunity.com/sharedfiles/filedetails/?id=3127323775
https://steamcommunity.com/sharedfiles/filedetails/?id=3127322051
I have absolutely no idea why it's a 33% increase. I mean it's always worked wonders for me. But by that much never. I mean it's been a while since i've tested it but yikes. I tested in game as well and yup quite a big difference. Also note i have the gpu at max performance as i wanted to see what it looks like in the Bench Results Screen.
In the Bench Results Screen
HPET=Enabled 55fps and cpu at 61.5%
HPET=Disabled 97fps and cpu at 78%
So for some odd reason when HPET=Enabled it's reserving some cpu? A difference of 42fps or 130% increase with HPET=Disabled. This is so odd. Maybe it doesn't mean anything. Maybe the screen or i should say the load is stuck at the last frame rendered in the benchmark. I've no idea here. Maybe it's just some Sandy Bridge mobo cpu weirdness going on. In games i play i tend to get fps = up to a 9700k running at stock speeds at times if they're not optimized. But i also have another thing i do/disiable. And i'll not share that as it's up to another 13% gain for me. And most people can't disable it. Only older mobos/cpus can do it. Or i should say the older it is the more helpful it is. And it's an Intel thing only. It's also unsafe i think may be a fair word. That's 3 big hints there. Hm i may be wrong just checked the website and maybe newer ones can benefit, again Intel. But ya the newer the cpu the less helpful it is until 8th Gen+ no help at all. My 2600k is already overclocked to 4.3ghz from 3.8ghz as an fyi and you bet that helps. But not what i'm talking about.