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
The author of this topic has marked a post as the answer to their question.
Discussions Rules and Guidelines
Report this post
Limit your FPS to 30 too if not done yet, beyond that it'd crush for me too. Its not really pc optimized
I have the game in fullscreen mode.What's the difference between fullscreen and HDR mode (Disabled)?
Technically, any boost clock above 1733 MHz is overclocked for a GTX 1080. Unless it's a founder's edition directly from Nvidia, it's probably factory overclocked.
I also have a 1080 and I run it at 2025 MHz core using a V/F curve, but not all chips are created equal. Some aren't going to be 100% stable at their factory clocks, let alone a manual OC like mine.
So do you know what clock yours is at when running the game and have you tried manually backing it off closer to 1733 MHz to see if that stabilizes it?
I actually have no idea on what Mhz my card runs. Where do i see it and how can I "down clock" it? ty for your help
edit: MSI Afterburner says Base 1265 but the white arrow is on 1750
Yeah that's really not terribly high, but still... the way Nvidia Boost 3.0 works, it may be trying to push the clocks higher than even the vBIOS boost clock (the one set by manufacturer) if it's cool enough. People run into this a lot when using liquid cooling blocks on GPUs in particular, because Nvidia's Boost 3.0 algorithm sees the card is only 40-50C and tries to ramp up the clocks too aggressively.
If you detatch the graphs at the bottom, you can actually see what the card is doing over time (or just use the OSD). But more or less, you can just run the game then mouse over and look at the details for the peaks on the graph for clock speed and see what it was doing when/if you get an ERR12 crash.
I had a friend who got this constantly on his GTX 1070 Ti and we had to manually downclock it from the factory 1783 MHz it would run at 65C to around 1650 MHz, 33MHz below even reference spec for a 1070 TI, to get it to run MHW stable. Unfortunately, the game came out after these cards were tested and I think some chips will run into instability at what the factory thought was good based on the games from 2015/2016 that most of the GTX 1000's were likely tested with when the factory clocks and AIB cooling configurations were going through their paces.
The error, in and of itself, is a TDR, which means the card or drivers locked up and did not respond within 2 seconds, so the driver was closed by Windows and restarted. It's the same as a D3D_DEVICE_HUNG or "The D3D device was removed" error in other game engines. TDR means "Timeout Detection and Recovery" and is put there to prevent what used to happen in games when a GPU was running into instability as just frozen hard locks (video frozen or black with continued or repeating audio, requiring a restart).
It can be caused by a buggy driver sometimes, but IMO, as prolific as it's been in MHW it's obviously caused by the card's core clock being unstable when handling the game's D3D11 calls. The game is just particularly good at exacerbating any even minor instability in GPU chips if you ask me. Some higher-end (binned) chips handle it fine like mine does, while others, like my friend's, may not even be stable at what is considered the reference clock for the chip.
Edit: Since you have Afterburner already, I can tell you some easy ways to muck around with the clocks.
With AB open, hit CTRL+F to open the V/F curve window. This is essentially the frequencies available to the card and the corresponding voltages it will use to run those clocks.
Now, there are two ways to approach this, you can lower the clock speed or you can increase the voltage the card draws at a given clock. Don't worry. You really can't do anything wrong here. The card is going to have a voltage limit set by the vBIOS and you can't go beyond that even if you want to, so killing your card through overvolting it isn't possible.
Just to test, I suggest clicking on the dots along the curve until you find the one that is 1733MHz or just below that. With that dot selected, if you hit CTRL+L, you will see a yellow vertical dotted line appear on that dot. That's telling you that the frequency lock is enabled. The card is now unable to go above or below that frequency, but will stay right there. Now test the game and see if you get an ERR12. If you still do I would move the dot to the right, increasing the voltage given to the GPU when it runs that frequency and test with the game again. More voltage means more heat, so ideally, you want to find the sweet spot where it's just enough voltage to run that clock stable and no more.
This is essentially just like CPU overclocking except much safer. When a frequency in a CPU OC is unstable, we keep increasing the voltage until it is either stable or we reach the safe voltage limits of the CPU. Same theory applies to GPUs. If a GPU is unstable at a given clock, you have two choices: drop the clock or increase the voltage. Luckily, modern GPU's have protections so that you cannot overvolt them so play around with it. If you have to bump up the voltage a bit to get it stable, that's fine. It just means that the factory V/F curve stored in the vBIOS was a little too aggressive on the clock vs the voltage it's feeding the GPU and it needs a little manual tweaking.
Once you understand this, conceptually, you can then move all the higher clocked dots you see to the right of the one we locked a bit further right, giving them a tad bit more voltage, and if you want to be thorough, test them each using the frequency lock (or just turn it off and let the GPU use the unlocked curve) to ensure they are stable.