Okay, card has a zero rpm function, so may not be the fans. Although an increase of only 18-28% fan speed at 73C just doesn't seem right, I mean it will have to go from 28 to 100% in the space of 15-20c ? RPM seems to check out though so the rpm% thing is a bit strange.
According to Saphire: According to Sapphire, the pair of 95mm fans on the Dual-X cooler have an 85 percent longer life than traditional fans and repel dust. What's perhaps more unique is that Sapphire's partners will replace a fan if it goes bad, rather than having the user send back the entire card. "To resolve a specific fan issue, Sapphire’s Quick Connect System provides a quick and easy solution. If there’s a fan issue detected, the user doesn’t need to return the entire card—Sapphire’s partners will send out a replacement fan directly to the customer or the retailer," Sapphire says.
Just checked all the readings, and other than missing the VRM temperature sensor, everything else seems fine. If it were overheating somehow, it would have to be the memory causing the crashes, as the VRM would likely do nothing other than shutting the system down at a dangerous temperature. Not seeing any GPU memory errors in the log. My first guess would be that it's probably just driver related, and newer drivers will improve compatibility with some games. Although the fact that it doesn't crash with the fans at a higher setting would indicate something temp related. So, first suggestion would be to flip the GPU bios switch, and also check your PSU 12v voltage in your system bios - the hwinfo64 reading on the PSU 12v rail is perfectly fine. The only vaguely suspicious thing I see related to the PSU is that the GPU VRM 12v rail drops from 11.8v all the way down to 11.5v under load, which could indicate a slight drop in power but these things should be fine within a 5% range and it's just a software reading. perhaps confirming the number of 12v rails on your PSU and ensuring the PCI-e connectors are correclty fitted might help.
Alright, I'll try to run the secondary BIOS. EDIT: Switching the BIOS didn't change anything, still crashed after 5min of witcher 3 this is my psu https://www.techpowerup.com/reviews/Antec/HCG-520/2.html I believe I only have 1 12V rail. I checked the BIOS for the voltage and here's what I got:
I don’t see a picture. That’s a very good PSU, and meets the recommended minimum, but how long have you had it for? 5 years in, and PSUs start to lose their efficiency and borderline power delivery can cause crashes under load. But again the fact that increasing the fan speed makes your card stable would suggest it’s probably not the PSU, although power usage will be slightly lower if the card is running cooler.
I don't know why my images don't appear correctly, you can open it by right clicking and open in new tab. Yes, I've had this PSU for a while now, 6 years~ish I'd say. When I have time I'll try to switch this PSU for another more recent one in another computer. Meanwhile, I guess I'll just let my custon fan curve on. Thanks for all the help!
The 12v seems fine in the Bios, so nothing noticeably wrong with the PSU. Switching to the more recent one is worth a shot.
Alright, so I tried the other PSU, for a moment I though it fixed the problem, and then it crashed haha, I guess it took longer to crash because this time around my case was open, so the temperatures were lower than usual. Here are the logs with old and new psu: https://files.fm/u/q8y3vnp7 So I guess I should just return it ?
Hmm wonder if the cooler connects properly to the VRM's though yes if these hit the thermal threshold (Varies from card to card.) I was under the impression the GPU would just shut down. There was a user wondering about higher than average temps and then a careful check showed that the vapor chamber had warped and likely been overlooked during assembly though a RMA took care of it. For the drivers that's these for now. https://www.amd.com/en/support/kb/release-notes/rn-rad-win-18-11-1-rx590 And from a quick check in the INF files I'm not finding a 590X entry in 18.11.2 and I did hear there were some issues on Linux with the GPU but that's also driver related and doesn't affect Windows. EDIT: Ah there's that hardware issue I mentioned. https://www.reddit.com/r/Amd/comments/9i71jh/xfx_vega_56_blower_fin_stack_vapor_chamber/ Although I don't know if Sapphire uses vapor anymore other than the limited versions of their GPU's or if it's just a standard heatsink though if it's not properly seated or there's gaps that could be a problem whether it's the heatsink or something with the thermal pads and such. Unless it's purely software and a driver issue.
As long as HWiNFO or GPU-Z aren't updated for the rx590 i wont be able to check the VRM temps to know whether it's a misplaced heatsink or it's the driver.
Right that would need to be updated to get the sensor info, although I thought GPU-Z was updated with 590 support as of it's latest version but maybe it's not getting all the data then. https://www.techpowerup.com/download/techpowerup-gpu-z/ Well it's updated pretty frequently so can't be too long until the next version comes out. EDIT: Ah I'm being dumb I think, as a non-reference board it could also be using other components and that could need a update too before it can be supported.
Ye I have the last version, unless I have an option to enable to make the vrm temps appear? I checked the settings but didn't find anything
I've had another look at the log files, and the only thing I can see is that the GPU VRM 12v rail is dropping to 11.3v under load, with a potentially high range of 0.6v. The GPU VRM 12v rail will usually scale with temperature and load. So, in theory, if the VRM were getting too hot, it could affect the 12v rail and therefore power delivery stability. We have to keep in mind that these are software readings, so I can't say much about sensitivity or accuracy. Of course, being able to see the actual VRM temperature would be very helpful here. Regardless, if increasing fan speed results in no crashes, it almost certainly must be a heat related issues, which could be caused by the driver/bios or the heatsink placement. It might be worth contacting Sapphire support, as they may have had other people with similar issues and may provide a solution.
I've always found it much better to use a custom fan profile with my Nitro+ starting 20% fan and going up from there unless you're wanting silence as their fan profile is just rubbish
Ye definitely, it's just it's supposed to work properly with their default fan profile at the very least.
Just to get that off the list, did you reinstall windows when upgrading from the previous GPU because DDU won't always do the trick.