Random black/gray/orange/green.... screen when playing games AMD Gigabyte RX 480

[abaximan]

Hello,

When gaming my computer freeze/bsod and I have to hard reset rig. While ago I couldn't even reset and I had to turn PC off and turn it on again to restart, but I'm not really sure what changed, but I think, it was due changing memory xmp setting in UEFI on or of. I did more or less everything suggested on various forums, but I just can figure out what is wrong with my system.

It looks that undervolting GPU helps just a bit, but sooner or later issue return. Sometimes I can play for hours, but then again sometimes freeze happen just after few minutes game play. It doesn't matter if I play or game is just idle or if I'm in the game menu.

I managed to log few such events with info HWiNFO64 (https://drive.google.com/drive/folders/1vNPxk5iF8Ov0Hjf6XlE53ve4DgwZzEAJ), but I can't really see anything wrong with temperatures of CPU or GPU. Voltage on GPU looks stable too, but I'm not expert so I could be wrong, but generally those numbers look OK. I also ran RealBench v2.56 for errors, but according to software all is good, although I can't figure out why it's showing only 4GB vram, so i'm not sure how to test entire 8GB vram.

All drivers are updated to latest, have latest bios on GPU and mobo, clean win10 were installed just along with setting up new rig.

I also installed different older drivers (went back to 01-2016), tried tu install older Vulcan Api. Earlier was running on stock settings, but I did tinker a bit because of obvious reasons - mostly to try to figure out If OC or different volting would do any difference. I tried like every bench test and stress test I could find, but it looks like thing happens only when I'm running games. Only once I succeed to recreate BSOD with 3DMark (FIre, but like I said stress test doesn't do much harm.

I tried GPU on other PC and there was no issues. Secondary PC was running game for few hours with no crash. I also tried different RX480 (Radeon™ RX 480 G1 Gaming 8G), which is basicly same - it's just OC. Everything was great, except this card was really hot while the one, presumably faulty never heat this much.

I also removed RAM and try to run with only one stick ...same thing happened. In about a week, I'm getting different RAM, because I'm suspecting it could be memory issue. I also read few post where there is mention of BSOD, G.SKILL memory, Asus mobo and RX480 with same issue as I have, so I'm starting to suspect it could be RAM compatibility issue with my mobo.

I'm sure it's not about PSU, because I tried 3 PSUs and same thing happen.

I was thinking maybe could be some UEFI setting, but I did default and nothing happen.

My specs:

ASUS MB STRIX Z370-H GAMING, LGA 1151, DDR4, ATX

CPU Intel 1151 Core i7 8700K 4.2GHz Box 95W - Coffe Lake

Cooler Intel/AMD SCYTHE Mugen 5 Rev.B SCMG-5100

GPU GIGABYTE RX480 (GV-RX480WF2-8GD)

DDR4-16GB 3000MHz CL16 KIT (2x 8GB) G.Skill 16GB Aegis (F4-3000C16D-16GISB)

PSU Gigabyte B700H 700W

Monitor AOC Q3279VWF connected via DP (HDMI or DVI have same problem)

WIN 10 EDU


I need t omention one more thing. Before I upgraded this rig, RX480 was used with overclocked Intel Quad Q9650 with DDR2 8GB and I could run almost anything on high setting on 22" 1080 monitor. I never had issues with gaming, but there was always some weird behaviour around. Desktop image went out, then gpu driver reset, even I was just using word or internet browser. Second weird issue was, that monitor lost connection when it waked from sleep and I had to turn it on and off to run it. I always had connected LED TV too, but recently I have issues with dual screen or extended screen too. When switching I got only black screen on both screens.

I did probably try almost anything, I found on forums. Any help or suggestions would be greatly appreciated. If changing RAM won't help,

Hopefully some had same experience and could give right solution, but there is so many post about this same issues, I have doubts.

 

[scout_03]

post picture if the sensor page of hwinfo scv is hard to follow also did you memtest each ram module also you tell gpu is overclock set it back to default he other thing you did check gpu in another system and did try many gpu in this one since you have more then one pci-e slot on motherboard did you try the gpu in them and do you have this color issue with onboard .

 

[abaximan]

post picture if the sensor page of hwinfo scv is hard to follow also did you memtest each ram module also you tell gpu is overclock set it back to default he other thing you did check gpu in another system and did try many gpu in this one since you have more then one pci-e slot on motherboard did you try the gpu in them and do you have this color issue with onboard .

HWinfo is hard to follow indeed, that's why I use GenericLogViewer. I'm not sure how to screen shot entire sensor page. There is awfully a lot of data there for just one screen shot.

Hmmm ... I believe my post already answer most of your questions, but I guess I can do that again.

Memtest was done - no error. Both modules together and separately.

You read wrong....GPU is not OC. It was just undervolted in profile for each game, because that help for some reason. Not always though. GPU was always crashing on stock settings, that's why undevolting.

Not exactly sure what were you asking about PCI-e and GPU, but I'll try to answer anyway

Just for clarification ... I have 2 PCs. One at work and other for personal use. So let's call it HOME and BUSINESS.

PC HOME have issues with Radeon™ RX 480 WINDFORCE 8G.
PC BUSSINES work great, but in there I have installed Radeon™ RX 480 G1 Gaming 8G.

Radeon™ RX 480 WINDFORCE 8G was tested on my PC BUSSINES and one other system. It was working with no issues. As I already wrote - this GPU was also working with no mayor issues on my old system.

I tested Radeon™ RX 480 G1 Gaming 8G from PC BUSSINES on my HOME rig and everything was working fine.

I also tested both PCI-E slots, but that doesn't really change anything.
.
I'm guessing "colour issue" is RGB on mobo? If that is correct, no I have no RGB issues.

 

[scout_03]

so the rx 480 from office work on both system but the one from home system does not so could you keep the office one in home system to game with it and see if random color will be back and use the home one in office that will permit to test both gpu under load after you will have to lok at video cable and monitor on home pc .

 

[abaximan]

There has been some development. Now I'm 99% sure culprit is GPU. Different RAM didn't change anything, but after testing RAM in touched card and it was really hot. I almost burned my fingers, although HWInfo last record showed 76 °C. So I'm guessing there must be issue with sensors, maybe fan. I also remembered, that fans of other card I was testing with (RX 480 G1 Gaming 8G), was much nosier under load. So I decided to leave computer case open and do some extensive gaming for a weekend and bam ... it worked ... well more or less. All games worked well enough under stock settings. It did BSOD once and just before that game start stuttering so I alt+tab to HW monitor an temp was 80 °C (wattman was showing 72 °C) and second later PC went BSOD. So i'm going to RMA gpu, hopefully they found it's faulty and I can get new one.

 

[scout_03]

wow a 72 c on load with hwinfo that is too high what was the fan speed you had for it since i use a r9 380 and max it goes on full load is 60 c .

 

[abaximan]

I RMA my RX480 and changed it for RX580. There was obviously GPU issue, but I guess more powerful new rig revealed this. After long testing, I finally manage to pinpoint problem. It was not overheating. Well yes temperatures are high, but GPU shouldn't BSOD when GPU hit 80-81°C. And that exactly what was happening to my RX480. I also changed housing with COOLER MASTER 690 III and added max number of posibble pwm fans so that should take some heat off, since AMD gpu's tend to produce more heat.

Anyway .. since I was looking for solution all over internet and I've seen there is much ppl with similar problem, I suggest first check at what temp bsod happen. RX580 or RX480 should not bsod even when it heat even to 90°C. So anything else is probably false gpu.