Victoria 3 | 12th/13th/14th Generation Intel Processors

[TheOriginalKage]

Hi, sorry for the late reply, I originally tried inputting the settings you had done on your 5 Dec comment, however that resulted in a hard crash. You said to play around with the numbers until I find something that works, but in all honestly I understand so little of what this all means that I just feel lost. Here is the current settings, if you have some recommendations on what I should try next that would be much appreciated.

Motherboard:
MSI MPG Z690 FORCE WIFI (MS-7D30)


View attachment 1070417

Try scaling back the AVX2 offset to 8.0x, then run the AVX stress test in XTU for 5 minutes to see how stable the system is.

 

[TheOriginalKage]

Anyone getting this/solved this for a i9-14900K? It works for a while (30 minutes to an hour) then crashes frequently afterwards. I've tried the BEX64 fix as that seemed to clash, I've ruled out other things. It just seems to really hate life. I tried with XTU and other things, but it seemed to make it worse.

I just got an Alienware Aurora R16 with a i9-14900K, and it ran Vic3 for several hours on 5 speed without a crash. I DID get crashes, but there was literal hours between them, and only in the late game (after 1900). I've actually done 3 full playthroughs on Vic3 with this system & CPU. Can you run a AVX stress test in XTU for at least 5 minutes without a crash or it telling you there is a failure?

 

[TheOriginalKage]

When will Paradox patch the game to resolve this? The game worked fine for me for two years until their update this year, and with the same hardware I'm now told it's the Motherboard's fault. Surely they can simply change back whatever they changed that makes these crashes happen?

I have not attempted fiddling with the CPU settings because that's not something I'm comfortable doing. I don't understand the "why" to the whole thing and don't want to damage my computer. I started the process but as soon as windows started warning me off using extreme tuner and the advice was to go into my BIOS settings and change things, I saw nah. That's kinda silly for a strategy game with middling graphics that works fine on my 3 year old macbook without issues.

If it's intel's problem, why are they not issuing a recall to resolve it? I still don't understand why this isn't Paradox's fault. But for the update, the game worked fine on my exact same system.

Paradox started using AVX calls on Vic3 1.5 and recent versions of CK3. Making AVX calls can speed up programs but they also push the CPU further. This is why both Intel XTU and various Intel motherboard vendor BIOS have AVX offsets, to lower the clock speed for AVX calls on CPU which are already pushing the limits.

In most cases, this would not be an issue. However, several motherboard vendors also undervolt the CPUs (I know ASUS and MSI motherboards do this, ASUS using the AI Tweaker) which reduces the power available to the CPU to lower wattage and heat. This works fine most of the time, until new additional calls are being made to the CPU, which will drive it over the edge and cause an undervolt crash.

Here is a basic test to see if your issues are AVX related: in Intel XTU, run a AVX stress test for 5 minutes. If you can do this without getting a fail, then your issue is not AVX related. If you DO get a fail, then you will get crashes in other things which call AVX... like Victoria 3.

What I can tell you is that I just got a new system with a i9-14900k with undervolt protection enabled. The CPU is getting the right amount of power and voltage. And I can play for Victoria 3 for several hours in a stretch (the longest being 5 hours) without a crash. You could try enabling undervolt protection on your system and seeing if it contains the crashes.

 

[TheOriginalKage]

I believe a critical step for everyone having issues with Victoria 3 is to enable undervolt protection for your platform/motherboard and disable overclocking, and see if you are still getting crashes.

 

[Abaddon55]

Try scaling back the AVX2 offset to 8.0x, then run the AVX stress test in XTU for 5 minutes to see how stable the system is.

Here are the AVX and AVX2 test results

 

[Generic_Name]

Hey there, Kage, just wanted to stop by to thank you for providing the solution that single-handedly fixed my incessant crashing. Saved me a great deal of sanity.

 

[Zimmerson]

So. I have done even more digging. I no longer believe the problem is the game, at least it's not just the game. At least for me, on a i9-13900k, the issue is a combination of the AVX commands being called by the game which is pushing these CPUs harder in ways that they have not before, modern motherboards from Asus (and possibly MSI and others) helpfully optimizing the CPU for you by undervolting it (sometimes drastically), and the system being overly optimistic about how well the CPU being able to perform at high clocks with multiple cores at these settings.

I now have a stable setup for Victoria 3 and my i9-13900k which allows me to play for literally hours without a crash while also not being hot as hell. I am as shocked you are.

First, the motherboard: I noticed that my Asus motherboard (Z790-P Wifi) had SVID Behavior set to Auto by default. This setting runs the CPU over 0.2 volts lower than when setting it to Intel fail safe default. It was a diference between 1.64v for the Intel default and 1.38v or so for Auto. This is a pretty strong undervolt! I ended up using Worse Case Scenario settings which put the CPU at 1.47v. I really wanted to make this work as otherwise, left at these dafaults, the CPU would both thermal and power throttle when running something similar like the Intel XTU benchmark, eating 300-305W. With the Worse Case Scenario setting running around 1.47v, my CPU tends to run around 240-250W in the same benchmark.

Second, the CPU settings: I used Intel XTU for this part, as it allowed me to tune to my CPU. What became apparent after multiple stress tests is the CPU cannot handle this undervolted setting while running all the core at full bore. The default settings assumed that it could handle 5.5Ghz on all 8 P Cores and 4.3Ghz on all 16 E Cores at the same time for a period of over a minute. Well, it cannot. Esepcially true when AVX workloads are added the mix. I attached a screenshot of the tweaks I did in XTU for my processor. In short, I created as curve to lower the clock speed as more cores were enabled, and I did this for both the P and E cores. This has allowed me to run longer (5+ minute) stress tests for both regular and AVX workloads and prevent thermal throttle. This is important, as the AVX workload runs the CPU harder, pulling more power and causing more heat... and crashes. I am also using a strong AVX offset of 10. It's probably aggressive but its stable for me and with it and the core changes I can run a 6 minute long AVX stress test with no crashes and no thermal thottling. If you have a better CPU cooler and/or you do not mind some occasional thermal throttling, you could probably more the core ratios up when multiple cores are called.

With these settings I have a CPU which still turbos when I need it to turbo, a CPU that pulls much less wattage and generates much less heat, and most importantly for this forum allowed me to play Victoria 1.5.10 without crashing every 5-10 minutes.

You would need to play around in XTU to see what is stable for your CPU. Unless your CPU has absolutely won the silicon lottery, I am betting it is almost impossible to overclock these CPUs and also have stable game.

Been having issue with my 14th Gen CPU (Intel-I9 14900k) And I've tried copying the screenshot you posted on what you've changed. Only issue is - replicating what you've done flags up that "Your system wil not be able to recover to a valid state automatically"

What on earth do I do? This is my first time doing this thing and I'm extremely afraid i'll brick my CPU.

Core Voltage for "Multiple" is also greyed out. Not sure why.

update:

Thought it'd go bang but I went for it anyway. Seems to not be crashing so far which is nice. I may push the stats closer to the default settings to see where it dies, optimising what I can get out of my CPU.

 

[Mouseklip]

So. I have done even more digging. I no longer believe the problem is the game, at least it's not just the game. At least for me, on a i9-13900k, the issue is a combination of the AVX commands being called by the game which is pushing these CPUs harder in ways that they have not before, modern motherboards from Asus (and possibly MSI and others) helpfully optimizing the CPU for you by undervolting it (sometimes drastically), and the system being overly optimistic about how well the CPU being able to perform at high clocks with multiple cores at these settings.

I now have a stable setup for Victoria 3 and my i9-13900k which allows me to play for literally hours without a crash while also not being hot as hell. I am as shocked you are.

First, the motherboard: I noticed that my Asus motherboard (Z790-P Wifi) had SVID Behavior set to Auto by default. This setting runs the CPU over 0.2 volts lower than when setting it to Intel fail safe default. It was a diference between 1.64v for the Intel default and 1.38v or so for Auto. This is a pretty strong undervolt! I ended up using Worse Case Scenario settings which put the CPU at 1.47v. I really wanted to make this work as otherwise, left at these dafaults, the CPU would both thermal and power throttle when running something similar like the Intel XTU benchmark, eating 300-305W. With the Worse Case Scenario setting running around 1.47v, my CPU tends to run around 240-250W in the same benchmark.

Second, the CPU settings: I used Intel XTU for this part, as it allowed me to tune to my CPU. What became apparent after multiple stress tests is the CPU cannot handle this undervolted setting while running all the core at full bore. The default settings assumed that it could handle 5.5Ghz on all 8 P Cores and 4.3Ghz on all 16 E Cores at the same time for a period of over a minute. Well, it cannot. Esepcially true when AVX workloads are added the mix. I attached a screenshot of the tweaks I did in XTU for my processor. In short, I created as curve to lower the clock speed as more cores were enabled, and I did this for both the P and E cores. This has allowed me to run longer (5+ minute) stress tests for both regular and AVX workloads and prevent thermal throttle. This is important, as the AVX workload runs the CPU harder, pulling more power and causing more heat... and crashes. I am also using a strong AVX offset of 10. It's probably aggressive but its stable for me and with it and the core changes I can run a 6 minute long AVX stress test with no crashes and no thermal thottling. If you have a better CPU cooler and/or you do not mind some occasional thermal throttling, you could probably more the core ratios up when multiple cores are called.

With these settings I have a CPU which still turbos when I need it to turbo, a CPU that pulls much less wattage and generates much less heat, and most importantly for this forum allowed me to play Victoria 1.5.10 without crashing every 5-10 minutes.

You would need to play around in XTU to see what is stable for your CPU. Unless your CPU has absolutely won the silicon lottery, I am betting it is almost impossible to overclock these CPUs and also have stable game.

Your CPU walk around fully fixed my CTD problem with both Victoria III AND CKIII. I went from consistent issues to the point of not playing at all, to completely fine games with the usual extremely rare CTDs which I'm fine with. My only other posts link my CTD logs and saves if anyone wants to reference. I searched my CPU online and the forums and the symptoms aligned.

i9-13900k

Thank you x100

 

[MadlockUK]

My CPU seems to work early game but anything late game (around 1870s onwards) it seems to crash more and more regularly as the USA. I've got it working early and mid game but going into the 20th century is next to impossible. Any ideas on why that is? I've done all the BEX64 fixes and everything again.

 

[AndrewT]

My CPU seems to work early game but anything late game (around 1870s onwards) it seems to crash more and more regularly as the USA. I've got it working early and mid game but going into the 20th century is next to impossible. Any ideas on why that is? I've done all the BEX64 fixes and everything again.

Your issue is likely to be unrelated to the one discussed here, as that problem causes crashes early on.

Whether game bugs, or some local issue, is hard to say. From a save that triggers this crash for you, can you duplicate the issue with it on another comnputer?

 

[Hjoril]

TheOriginalKage

So. I have done even more digging. I no longer believe the problem is the game, at least it's not just the game. At least for me, on a i9-13900k, the issue is a combination of the AVX commands being called by the game which is pushing these CPUs harder in ways that they have not before, modern motherboards from Asus (and possibly MSI and others) helpfully optimizing the CPU for you by undervolting it (sometimes drastically), and the system being overly optimistic about how well the CPU being able to perform at high clocks with multiple cores at these settings.

I now have a stable setup for Victoria 3 and my i9-13900k which allows me to play for literally hours without a crash while also not being hot as hell. I am as shocked you are.

First, the motherboard: I noticed that my Asus motherboard (Z790-P Wifi) had SVID Behavior set to Auto by default. This setting runs the CPU over 0.2 volts lower than when setting it to Intel fail safe default. It was a diference between 1.64v for the Intel default and 1.38v or so for Auto. This is a pretty strong undervolt! I ended up using Worse Case Scenario settings which put the CPU at 1.47v. I really wanted to make this work as otherwise, left at these dafaults, the CPU would both thermal and power throttle when running something similar like the Intel XTU benchmark, eating 300-305W. With the Worse Case Scenario setting running around 1.47v, my CPU tends to run around 240-250W in the same benchmark.

Second, the CPU settings: I used Intel XTU for this part, as it allowed me to tune to my CPU. What became apparent after multiple stress tests is the CPU cannot handle this undervolted setting while running all the core at full bore. The default settings assumed that it could handle 5.5Ghz on all 8 P Cores and 4.3Ghz on all 16 E Cores at the same time for a period of over a minute. Well, it cannot. Esepcially true when AVX workloads are added the mix. I attached a screenshot of the tweaks I did in XTU for my processor. In short, I created as curve to lower the clock speed as more cores were enabled, and I did this for both the P and E cores. This has allowed me to run longer (5+ minute) stress tests for both regular and AVX workloads and prevent thermal throttle. This is important, as the AVX workload runs the CPU harder, pulling more power and causing more heat... and crashes. I am also using a strong AVX offset of 10. It's probably aggressive but its stable for me and with it and the core changes I can run a 6 minute long AVX stress test with no crashes and no thermal thottling. If you have a better CPU cooler and/or you do not mind some occasional thermal throttling, you could probably more the core ratios up when multiple cores are called.

With these settings I have a CPU which still turbos when I need it to turbo, a CPU that pulls much less wattage and generates much less heat, and most importantly for this forum allowed me to play Victoria 1.5.10 without crashing every 5-10 minutes.

You would need to play around in XTU to see what is stable for your CPU. Unless your CPU has absolutely won the silicon lottery, I am betting it is almost impossible to overclock these CPUs and also have stable game.

I'm having a hard time following what to do. I'm sorry if it's been explained but everything I've tried by following hasn't changed anything. I've done the test after changing the AVX settings in Intel XTU....and nothing seems to have changed and my computer still thermal throttles me. As for the first part I don't understand how to get to that in bios I'm not seeing SVID behavior

 

[MadlockUK]

Your issue is likely to be unrelated to the one discussed here, as that problem causes crashes early on.

Whether game bugs, or some local issue, is hard to say. From a save that triggers this crash for you, can you duplicate the issue with it on another comnputer?

So it seems to grow in entropy, so I'll get the odd crash at the beginning, but as it progresses, it crashes more and more frequently. I'll see if I can get it to work on an old laptop on a particularly cold night to see if it'll replicate it.

I find the monthly ticks are when it crashes as per below, the spike on the right is when I would typically see a crash especially if certain ticks like enacting laws or mass migrations I notice seem to make the CPU really jump.

 

[killerdrath]

I was wondering if anyone could lend me a hand with this, I am unable to download the XTU so if someone could just give me a few pointers on what to change in the bose setting that would be very helpful.

 

[AndrewT]

I'm having a hard time following what to do. I'm sorry if it's been explained but everything I've tried by following hasn't changed anything. I've done the test after changing the AVX settings in Intel XTU....and nothing seems to have changed and my computer still thermal throttles me. As for the first part I don't understand how to get to that in bios I'm not seeing SVID behavior

I was wondering if anyone could lend me a hand with this, I am unable to download the XTU so if someone could just give me a few pointers on what to change in the bose setting that would be very helpful.

@TheOriginalKage Are these things you can help these users with, please? It's well beyond me!

 

[Tautdan]

@TheOriginalKage Are these things you can help these users with, please? It's well beyond me!

Is this something that can be fixed within the code of the game?
As again this issue is only experienced with Vic3 and CK3 which is unfortunate.

 

[AndrewT]

Is this something that can be fixed within the code of the game?
As again this issue is only experienced with Vic3 and CK3 which is unfortunate.

I think only by not using AVX at all, which I'm sure is not going to happen. As @TheOriginalKage found, the base issue is in the CPU and bios settings for it. I suspect it probably does affect other games that use AVX, but I guess not all for some reason.

 

[jpd]

Yup. The way I understand it is this: The CPU's are configured with a certain power consumption in mind. When you drive the CPU above a certain level of utilization, the power can't keep up and the CPU becomes unstable. AVX instructions seem to fall into this category of CPU utilization.

That's why disabling some (E-)cores works. That's also why lowering the clock for AVX instructions works. Both measures decrease the amount of power the CPU needs to draw because it's total workload per unit of time is being lowered.

In terms of electrical engineering: A Watt is defined as Voltage times Amps. Amps is what ultimately drives all electrical activity inside the CPU. If you do more work, you need more Amps to do it (watch the Apollo 13 movie for an explanation when they try to figure out the electronics restart protocol prior to re-entry). Given that the voltage is usually a constant (220V, 110V, 5V, 12V, etc. You get the idea), requiring more Amps means that the total value of Watt goes up.

But what if the total Watt is limited by an upper ceiling? Well, the Amps requirement remains, so the only way to adjust the formula (watt = volt * amps) to make the equation work is to lower the voltage. And, in a CPU, that's bad. CPU's use voltage internally to represent values. All values are stored in sequences of 0's and 1's. It's called a binary representation because of this. A low voltage means a '0', and a high voltage means a '1'. But what happens when the overall voltage drops because we're drawing too many amps? Some '1''s are getting misinterpreted as '0''s. When that happens to data, you merely get wrong answers from calculations. But when it happens to code, the CPU starts executing wrong or unintended instructions (because of the misinterpretations), or try to load data from, store data to or even call routines from wrong (or worse: non-existing) memory addresses. And it's this latter category that ends up in crashes to desktop.

All mitigation measures discussed in this thread to date attempt to ensure that the total amps drawn per unit of time stays within the max power wattage available without causing the voltage levels to drop below the minimum levels where misinterpretations start to happen.

 

[KOCTA]

Hello there; Same issue - applied different workarounds - nothing works for me
I think the problem in game and developers should fix it, since paradox support forwarded here, because of script system error:

[20:12:02][jomini_script_system.cpp:262]: Script system error!
Error: Failed to scope to state region by key 'STATE_NASSAU'
Script location: file: common/customizable_localization/99_ru_custom_loc.txt line: 164059

 

[AndrewT]

Hello there; Same issue - applied different workarounds - nothing works for me
I think the problem in game and developers should fix it, since paradox support forwarded here, because of script system error:

[20:12:02][jomini_script_system.cpp:262]: Script system error!
Error: Failed to scope to state region by key 'STATE_NASSAU'
Script location: file: common/customizable_localization/99_ru_custom_loc.txt line: 164059

Your problem is almost certainly unrelated to the issue discussed in this thread. Which is not a game bug.

You are free to report that error in the Bug Reports forum if you feel that is causing your problem. If however you feel technical support could help you, make a new thread in this forum. Thanks.

 

[AndrewT]

So. I have done even more digging. I no longer believe the problem is the game, at least it's not just the game. At least for me, on a i9-13900k, the issue is a combination of the AVX commands being called by the game which is pushing these CPUs harder in ways that they have not before, modern motherboards from Asus (and possibly MSI and others) helpfully optimizing the CPU for you by undervolting it (sometimes drastically), and the system being overly optimistic about how well the CPU being able to perform at high clocks with multiple cores at these settings.

I now have a stable setup for Victoria 3 and my i9-13900k which allows me to play for literally hours without a crash while also not being hot as hell. I am as shocked you are.

First, the motherboard: I noticed that my Asus motherboard (Z790-P Wifi) had SVID Behavior set to Auto by default. This setting runs the CPU over 0.2 volts lower than when setting it to Intel fail safe default. It was a diference between 1.64v for the Intel default and 1.38v or so for Auto. This is a pretty strong undervolt! I ended up using Worse Case Scenario settings which put the CPU at 1.47v. I really wanted to make this work as otherwise, left at these dafaults, the CPU would both thermal and power throttle when running something similar like the Intel XTU benchmark, eating 300-305W. With the Worse Case Scenario setting running around 1.47v, my CPU tends to run around 240-250W in the same benchmark.

Second, the CPU settings: I used Intel XTU for this part, as it allowed me to tune to my CPU. What became apparent after multiple stress tests is the CPU cannot handle this undervolted setting while running all the core at full bore. The default settings assumed that it could handle 5.5Ghz on all 8 P Cores and 4.3Ghz on all 16 E Cores at the same time for a period of over a minute. Well, it cannot. Esepcially true when AVX workloads are added the mix. I attached a screenshot of the tweaks I did in XTU for my processor. In short, I created as curve to lower the clock speed as more cores were enabled, and I did this for both the P and E cores. This has allowed me to run longer (5+ minute) stress tests for both regular and AVX workloads and prevent thermal throttle. This is important, as the AVX workload runs the CPU harder, pulling more power and causing more heat... and crashes. I am also using a strong AVX offset of 10. It's probably aggressive but its stable for me and with it and the core changes I can run a 6 minute long AVX stress test with no crashes and no thermal thottling. If you have a better CPU cooler and/or you do not mind some occasional thermal throttling, you could probably more the core ratios up when multiple cores are called.

With these settings I have a CPU which still turbos when I need it to turbo, a CPU that pulls much less wattage and generates much less heat, and most importantly for this forum allowed me to play Victoria 1.5.10 without crashing every 5-10 minutes.

You would need to play around in XTU to see what is stable for your CPU. Unless your CPU has absolutely won the silicon lottery, I am betting it is almost impossible to overclock these CPUs and also have stable game.

Seems we have public acknowledgement of your findings, excellent work!

Some Intel 13th and 14th Gen Core chips keep crashing

Oodles of problems with data decompression and suchlike, depending on system settings

www.theregister.com

Is your Intel Core i9-13900K crashing in games? Your motherboard BIOS settings may be to blame — other high-end Intel CPUs also affected (Updated)

Apparently a 4096W power limit might be a bad idea.

www.tomshardware.com

To everyone experiencing shader compilation "Out of video memory trying to allocate a rendering resource" errors :: Nightingale General Discussions

To everyone experiencing shader compilation "Out of video memory trying to allocate a rendering resource" errors Your CPU is faulty and has Core errors. It's not your GPU, its your CPU Crashing with this sort error message (and others, including straight up CTDs) when performing shader...

steamcommunity.com