Question about GPU clocks and crashes
Question about GPU clocks and crashes
Hello, I own a Nitro+ 9070 XT card. Over the past few months I experienced some crashes in games, so I wanted to verify whether it’s typical for the GPU to operate above its advertised 'Boost Clock'. The card lists a boost clock of 3060MHz online, but during testing—such as with Marvel Rivals and certain benchmarks—I’ve observed it reaching 3200+MHz. Is this behavior expected? Could it lead to crashes? I’ve encountered crashes on BF6 returning to desktop or even full PC restarts, along with issues in Unreal Engine 5 (Rivals/Oblivion Remaster). I’m aware that these titles and UE5 have known problems. Checking EA forums or Reddit shows the BF6 crash is a widespread concern, and there are reports about needing to disable features like Lumen in UE5 to avoid issues. This seems frustrating despite having high-quality components.
I’m not attributing the problem solely to the GPU; I just want to confirm if this is normal. I’m currently running it on stock hardware without any overclocking or undervolting. I use a +10% power setting, but now it’s back to zero while still clocking at 3200MHz in Marvel Rivals. Temperatures are within acceptable ranges—game temps around 46°C and hot spots near 69°C. Other games may generate more heat. I’m not planning major tweaks, but should I adjust the clock offset to -200MHz or lower the power level? Does this seem like typical behavior? I usually cap FPS at my monitor’s 180-190 range.
Other updates I’ve made:
- Installed new RAM; memtest86 showed some errors but passed overall. A RMA with Corsair replaced the faulty memory.
- Upgraded PSU from an EVGA 850W ATX 3.0 to a quieter 1000W Platinum ATX 3.1 with a dedicated 12V 2X6 power cord. Noted transient spikes over 550W reported on HWinfo, suggesting my original unit might be insufficient.
- Purchased a new M.2 SSD and switched all games to a secondary drive running Gen3.
- Reinstalled Windows 11 completely.
I haven’t played many games yet to confirm stability. I intend to keep AMD metrics overlay and HWinfo64 active during Rivals and Battlefield 6. In previous crashes—before the RAM and PSU upgrades—I didn’t see any warnings in the event viewer, and my AMD adrenalin settings remained unchanged or didn’t trigger a crash notification. This makes me think the issue might be related to game code rather than hardware. For Rivals/Oblivion, crashes often appear with a UE5 crash box displaying errors like windows_shipping.exe or similar messages. Despite reinstalling Windows 11, I haven’t seen any crashes during testing.
Current specs:
- Monitor: 1440p at 180Hz
- CPU: Intel i7-12700KF
- RAM: Corsair 2X16 5200 DDR5 XMP enabled at 5200/1.25V (new from Corsair RMA, passed memtest)
- Motherboard: MSI MAG B660M Mortar Wifi DDR5 (latest BIOS)
- GPU: Sapphire Nitro+ RX 9070 XT
- Storage: Main drive is a WD black Gen4 M.2
- Game Drive: Crucial Gen4 M.2
- OS: Windows 11, freshly installed (all updates)
Begin by checking Reliability History/Monitor and Event Viewer.
These tools might show error codes, warnings, or events just before or during crashes.
Reliability History/Monitor is easier for users and can highlight patterns.
Event Viewer takes longer and needs more effort to interpret.
You can click on logged entries for extra information, though it may not always be useful.
Share your findings accordingly.
I haven't used it before, but I'll try anyway if crashes continue. The situation remains unclear. It's as if you can check the bucket or ID, search online – it seems connected to my driver or the game itself… but is it due to my RAM or XMP settings? Hopefully, it would be something straightforward like "Your game crashed because your GPU clocked too high because of cool temperatures and it ran too hard."
I participated in a Marvel Rivals match and recently played around 40 minutes of Fortnite. The GPU clock varies, it isn’t stuck, but during gameplay it’s between 3200-3400 on STOCK settings. Is this typical or just a lucky run? I might have crashed before because of my PSU or something similar. Checking my original post, I asked Sapphire tech support who confirmed it’s not normal—cards usually run higher than the boost clock, but overclocking software can push it further. They noted that with overclocking tools, it shouldn’t reach such high levels without them.
Many people do overclock their cards to reach similar performance, so I shouldn’t need to return the card. My temperatures during Fortnite were around 40-55°C at 98% GPU load, showing about 200W of power usage. The spikes above 500W were probably temporary.
Here are the images I shared:
https://imgur.com/a/vhsv5vB
View: https://imgur.com/a/vhsv5vB