Fry Cry 6 crashes consistently with 24.6.1 and 24.7.1. Rolling back to 24.5.1 seems to have fixed it, and also performs noticeably better with AFMF and seems to have picked up a few FPS in Starfield too. Since these are both "AMD games" I would expect them to perform well and run stably.
I probably should have mentioned, this is on 7900 XTX and 7900 XT.
This is for AMD so that someone might look into it.
I have done this on 2 systems now, this process seems to work to fix it:
1. Downgrade to 24.5.1 with the Factory Reset box checked. Reboot if prompted, otherwise just continue.
2. Run Adrenaline and do whatever setup you would normally do in there for games, etc.
3. Reboot
4. Upgrade to 24.7.1, do only an upgrade. Do not uninstall previous drivers, do not use Factory Reset in the installer, do not use DDU.
5. Reboot when install is complete, before testing.
I have no idea why this would work, other than to speculate that something in the installer is borked and maybe an internal setting is getting set wrong in a clean install that would import a previous setting when doing an upgrade? I had major issues when installing "clean" after a factory reset (box checked in the installer), but it works fine as an upgrade from 24.5.1. I suspect 24.6.1 has the same issue, downgrading to that had no effect, but downgrading to 24.5.1 got rid of all crashes.
This is "so far", but the clean install would crash within 5-30 minutes, and I can now play it for up to 2 hours without a crash - I didn't test longer than that.
The crashes are random, don't worry. You will get them sooner or later even with this procedure.
I have tested all the 24 drivers so far and all had issues. The most recent stable drivers are 23.12.1.
Yeah, they came back. I actually got 3 hrs and 15 min of FC6 out of it without crashing tonight. Unfortunately I did a bunch of things so now I have to begin the process of not doing them one by one and try to figure out which, if any of them made it more stable.
I had done those before though and it still crashed, so I kinda doubt it was any of them. Tonight in particular, before playing I:
I'm leaning toward the VPN client as that has a direct effect on the networking part of the game - tomorrow I'll try it with FC6 removed from the split tunneling, then with split tunneling completely off if it crashes, and then with VPN disconnected as opposed to exiting (I really don't want to have to shut off VPN for the rest of my system).
Did you previously use a public preview or bèta driver?
If so please uninstall the preview driver prior to updating 24.6.x or 24.7.x
Check windowsupdates from safemode or cmd safemode.
No. And on top of it, I reinstalled Windows a week or so ago (for unrelated reasons). And yes, FC6 did crash before reinstalling Windows.
Oh well, so much for my downgrade then upgrade "solution". Longer testing still has it crashing periodically.
This DID fix the frequency of crashes after upgrading from 24.6.1 to 24.7.1, then it would crash in like 5 minutes of play. Now it's more "normal" for FC6, I get an hour give or take before the game crashes. If you go from 24.6.1 to 24.7.1 and any games suddenly crash more than they did, try it.
Reverting to 24.5.1 still has the issue, I either got lucky at first or just didn't test over enough time. So, back to 24.7.1 it is, if FC6 is going to crash anyway, I might as well be on the latest driver.
I don't remember this being an issue when I had my 6800XT (RDNA 2). The performance of RDNA 3 is great, but stability has taken a hit.
I have a similar issue since installing driver version 24.6.1
At first, i could play for a good while. Then my game started having random CTDs and freezes combined with AMD driver crashes. Then it got so bad that my PC even started having bluescreens.
and then it got to a point, where my PC would go into a bluescreen 10 seconds after booting. I was honestly scared that somehow my hardware got borked. I did SFC /scannow and since then didn't touch that game anymore. Seems to be fine for now. hoping for a fix...
That's bad... My issue was isolated to FC6. At this point, I am not sure anymore if it's a driver issue or a game issue. Ubisoft really left a lot of unresolved issues with that game, and now it's abandoned as far as development goes.... Very disappointing.
You do know starting the Game in ADMIN fixes this issue???
I have tried that before and it didn't change anything, but I'm not opposed to trying it again.
I shut off core isolation today and gave it a short test - too short to be anywhere near positive, but it was on a break from work so I left everything else running - 2 browsers with a zillion tabs open, Signal, Steam, VPN.... No crash, but I only played for probably 30 mins.
However.... I swear I must have picked up 15 FPS in any given scene, and this is with Adrenaline downclocking the GPU from stock settings and PBO completely off, 3 monitors on, and comparing to when I run it with no other programs else open. (I usually shut down the 2 side monitors after work so that frees up some GPU resources and 55-60w of power at idle.)
EDIT: Jeezus, I had a VM running on my system too and didn't even realize it until now.
I also should mention I went into the game files and ran it admin from there so if you want to try it in there and see if it works for you
Try turning off freesync, especially if you have multi-monitor setup.
Since 24.6.1 I've been experiencing black screen of death/restart and random FPS caps on specific games while it runs OK on some other games (albeitwith weired refresh caps here and there).
Turned off FREESYNC on the 7800XT and everything runs good, looks at the community posts before and it feels like the freesync bug had made a come back.
I think I have this figured out. If I'm right, and so far I have had zero crashes since changing this, including 4 straight hours today and about an hour last night, then it's fixed.
After a ton of trial and error and researching online, I think it's RAM speed.
Like I'm guessing a high percentage of other DIY Zen 5 users do, I have the Corsair DDR5-6000 30-36-36-76 2x16GB DIMM setup for 32GB. After finding a post from a guy on a different game's forum that said he lowered his RAM speed and his AMD driver timeout problem disappeared, I gave it a try.
I looked in hwinfo and it gave me this table:
So I used the values from the 2800 line (2800x2=5600). On Asus, you just switch to the Advanced view (F7), go to the "AI Tweaker" menu, set it to "EXPO I", then just below that there is a drop box for RAM speed, change it from 6000 to 5600. Then go into the DRAM timings submenu and change the timings from 30-36-36-76 to 28-34-34-71. That's it, exit and save.
If you prefer to set it 100% manually, do the above besides loading EXPO, then there is another setting for PMIC timings that "EXPO I" sets to "sync all PMICs", which I don't know where the heck it is, but you can find it with search.
Also turn on "memory context restore" in the 2 places it shows up, so it it POSTs in a few seconds instead of taking over a minute.
Non-Asus BIOSes should have a similar process, with just the specifics different.
Here's where it gets weird: after it seemed to fix the crashes, I wanted to see how much performance I was giving up from lowering the speed, so I tested and instead performance went UP! I first tested in CPU-Z* where it had the highest multi-threaded score ever on this system by about 200, and the single-core shot up to over 715 before settling down to 705 (which is good for a 7800x3d), and in Passmark it scored the highest CPUMark score this system has ever gotten, and the MemoryMark was faster by a few hundred than it was at 6000 on EXPO I. Test notes - PBO is disabled, and I have a -20 all-cores undervolt set.
If this continues to stay fixed, people that are off-the-hook as far as being any part of the problem include:
AMD
Asus
Powercolor
Ubisoft
Signal, Proton VPN, Steam - shutting them down definitely made it go longer between crashes, but probably not due to anything the programs were doing wrong - IMO they were probably just stressing the RAM more, leading to more frequent game crashes.
On-the-hook for this: Corsair, IMO for probably pushing its speed rating for the memory a bit too high. What this appears to be is really good 5600 DRAM with very tight timings, but it is less happy at the 6000 it's advertised as. But, I originally bought G-Skill and that wouldn't even POST with EXPO. So this is still possibly/probably the best RAM out there for AM5 anyway.
* Yes I know CPU-Z is a garbage benchmark I would never use it to compare different systems, but it's good for a quick check on the same system to see what the effect of changes is.
Just did another hour with no problems. This time I ran it under circumstances that would have almost definitely crashed previously - played after playing 4 hours this morning, did a bunch of other stuff on the computer, left it up all day with web browsers open, left all the background programs running, set Adrenaline to "Rage" mode (actually I did this earlier for the last hour of my 4 hr morning test), then just came back and ran the game. No crashes.
Well, #$$%$&^^$#%^&%^& E@W$@#@#$@# @##$^%$^Q% the problem came back eventually. So this time I turned off EXPO completely and just let it default to the JEDEC speed, and once again it stopped crashing.
I ran a few OCCT CPU+RAM and just CPU torture tests with and without EXPO and they definitely crash sooner and harder with EXPO on.
I think my RAM is going bad, even though it passed 8 passes of MemTest86 with EXPO on.
Another telltale sign is my CPU and RAM bench slower at 6000 than 5600 or even JEDEC 4800 @ CL40 on CPU-Z and Passmark's CPU and RAM benchmarks, which should NOT be the case.
I'm just gonna get 2 new sticks and see if that's it. This time I'm not going with the most aggressively timed 6000 RAM available, I'm going to get something more conservative but reliable like Crucial's 6000 CL36 Overclocking Pro sticks, Crucial is never the ultimate performer but always very reliable, and that's fine with me, memory doesn't make a huge difference in real world performance anyway.
Well, this just gets weirder. While waiting for Amazon to deliver the new RAM, I realized I had a CPU undervolt set and it's not right to blame the memory without testing it at full stock, so I loaded BIOS defaults and left RAM at JEDEC and ran an OCCT torture test. Passed that no problem, so I next tried EXPO I. When I first booted it up I ran a CPU-Z benchmark and it scored higher than it had previously at 6000. Scratched my head for a sec, then thought about what did I not do in the BIOS that I would normally do. Started the OCCT run, which it passed.
DING DING DING Memory context restore! So I did some searching and sure enough, found a topic about instability with it on, and come to think of it, I did learn about that reducing POST times and turned it on about when the 24.6.1 drivers came out. (The undervolt was something I just recently set while troubleshooting this issue, it was not the cause - FC6 seemed to crash less with it undervolted).
So I left it off and tried EXPO II. Passed torture test. Ran FC6 for a couple of hours, no crashes.
Wouldn't be the first time a "fix" turned out to be premature, so I'll be eyeing it for a few days and just have to put up with 60 second POST times. Too bad, I liked it going from 60 secs to like 5 secs, but I'll take stability over fast boot anyday.
Well, after a couple of days the timeouts came back to where I couldn't get more than an hour, ans usually much less than that without a crash. I swapped out the RAM, no help. No memory test shows any kind of error with the Corsair or the Crucial RAM, at JEDEC nor EXPO speed.
At this point, after probably over 20 hours working on this problem and looking into everything in my system, I am 100% positive there is nothing wrong with my PC. It passes any torture test I can throw at it, and does it without any readings that would raise any eyebrows (like temperatures, etc). I put the Cosair RAM back in because it's faster and there's no difference in stability.
For good measure I ran:
SFC /scannow
DISM /Online /Cleanup-Image /CheckHealth
DISM /Online /Cleanup-Image /ScanHealth
DISM /Online /Cleanup-Image /RestoreHealth
Then uninstalled 24.7.1 and installed the Preview Driver for AFMF 2 Driver Version 24.20.01.02.
No more crashes, at least for tonight. One different thing happened that I've never seen before - it froze for a second, like it does when there's a driver timeout, then recovered and just went on like nothing happened. @amd - was there anything snuck into this driver to address the timeout problem? This "recovering from an apparent timeout" is completely new behavior.
The crash came back. I did some more stuff, tried different RAM, ran more torture tests to be sure all was well. Game still crashed.
Ran some DSIM commands to make sure my Windows files were good. Uninstalled 24.7.1 and installed the AFMF 2 preview driver. Crashing stopped again for now. New behavior that I have never seen before - the game appeared to hang like when a driver timeout occurs, then it recovered and went on like nothing happened.
Now I'm wondering if there was something in the preview driver to address the timeout issue? I am not the only person that has complained about this in general, maybe the first to focus this much attention on one game, but I have read many similar complaints, so it's not an unreasonable thing to wonder about....
Finally got a crash out of it after 5 straight hours of leaving it up. Big improvement.
How did you fix the crash on 24.7.1?
Frequent crashes in Far Cry 6 with drivers 24.6.1 and 24.7.1 can be frustrating. Try rolling back to a previous stable driver version to see if that resolves the issue. Sometimes updating to the latest game patch or adjusting in-game settings can also help improve stability. If the problem persists, it might be worth reaching out to the support team for further assistance.