Stock Vega 64 keeps crashing during games on default settings.

[iBoolGuy]

Hello, good day, so I've had my Vega 64 since launch, and it's been playing games like a beast ever since. I've had countless problems with it, I think I faced every GPU problem (especially software related) with this card, it gave me nightmares, but regardless tho, when it worked, it worked really good.

One thing I want to note before going on onto my problem, I gotta say that, I have tried and worked around most of the problems I've had with the cards (I have 2), things like: slightest overclock on the CPU causes instability in any game, my AVR setup with the home theater and PC was a pain to work with, having dual monitors (technically 3, since the AVR through HDMi is considered a display), problems with the settings of the Vega itself, whether trying to overclock, underclock, playing with the fan curve, it was just always a pain in the ass. But, in AMD's defense, *most* of those bad days got resolved throughout various updates, they fixed A LOT of stuff!

2 Things remain unfixed though, and those are what I need help with. So, sometimes, if my screens turn off after 5 minutes say, and I come back to them and touch the mouse, they will turn on, but the GPU drivers will display a notification saying there was some link issue blahblahblah AND, this is the problem, the GPU settings were reset to default! Or, restarting the computer itself sometimes will reset the GPU profiles/settings.
Now, I don't care about those link issues, because absolutely everything else in the PC remains normal, other than resetting the GPU global profiles that I set. 
Why that's bad?
The second thing remaining unfixed, the default profiles for my GPUs are simply not able to apparently keep the cards cool. I start up my PC, undoubting and unsuspecting Bool opens up any given game, goes through the menu, waits for the game to load, map and gameplay load up, I'm in the world of the game now, give me 10-15 seconds of moving around excited to play, and boom, the GPU crashes.
The crash goes as follows: all displays shut off, the tach/load LEDs on the Vega all shut off, the fans on the card turn to 4900RPM. It's like you'd plugged out the GPU from the system while it's running basically...

 

So what I have to do from there is, stay calm and collected, not lose my s**t and break anything, restart the PC, go to radeon settings, load up a global GPU profile that slightly tweaks the fan curve, instead of the max default at around 1800-2000 RPM, I allow it to go to max. And then I run my games and play for hours without issues. (There are issues, playing with fullscreen, then alt-tabbing and going to another display is prohibited, that crashes the game (either Apex Legends or Fortnite, those are the 2 I been playing), but disregard those issues, I forced myself to get used to switching to windowed fullscreen whenever I want to alt-tab, and it's been fine.

 

So, what do you suggest? I heard that the Vega 64's stock BIOS settings have a safety threshold on temps at 85 degrees C, which my card apparently reaches in ease and proceeds to crash.

Is there a SOLUTION? Or am I better off sticking to my "workarounds"?

[191x7]

I would like to ask you to try something. One card at a time.

Take the second card out. Test the first one heavily.

Take the first out, put the second in and test that one.

If both cards work fine on their own, test a card in the second slot (but only a single card).

Your issues might be caused by Crossfire, the motherboard slot, one of the cards being faulty or maybe, just maybe, the PSU.

[iBoolGuy]

11 hours ago, 191x7 said:

I would like to ask you to try something. One card at a time.

Take the second card out. Test the first one heavily.

Take the first out, put the second in and test that one.

If both cards work fine on their own, test a card in the second slot (but only a single card).

Your issues might be caused by Crossfire, the motherboard slot, one of the cards being faulty or maybe, just maybe, the PSU.

Hey bruverino, thank you your suggestion, I have a couple things to reply to:
I've always used only one card, the second was always plugged in, but no Crossfire, never, besides testing, so the second card is plugged in, and its main use was its HDMI port, to take in the AVR. Also mostly these days, to avoid these issues, I disable my secondary displays when I game, all of them, to eliminate issues, but the profile reset and heat crashes still happen here.

I also lived back in the day like 3-4 months with only one card plugged in, I don't exactly remember the details, but I remember the conclusion, it was that there was no difference in issues, bugs, crashes, freezes, hangs, BSODs, all of it, if it was going to happen, it showed that it still happened with one card. But again, a lot of this drama got resolved with updates, except for heat crashes and profile resets!

Also, whenever I open up my case to add something or change something, and when I take the cards out, I put them back in random order, so they must've swapped spots a lot!

Testing a card in the second slot? hmmmm... never even thought of that, interesting, but I have one more thing to add so hold up!

Again, to resummrize, I never use crossfire, my motherboard is a Z370 Gaming Pro Carbon AC, I would doubt anything at this point in my life, so will test that out! Lastly, the PSU was brand new, HX1000i!

Now the puzzle gets umm.. more confusing.... so today, 6 hours ago, I received my new 5700 XT. I waited so long for MSI to drop their Gaming X model, and that's what I got, I love MSI and their cards. Interestingly, this card, out of every single other AIB 5700 XT model, is the beefiest, heaviest, most ridiculously cooled card. In reviews, your typical average 5700 XT, would run at around 70 C, after an hour loop on Forza Horizon (4,5,6,7? not sure which one, but the latest), and the fans (of a typical average 5700 XT) would be spinning at an average of around 2000RPM. In the same channel under the same testing methods, the Gaming X 5700 XT from MSI, kept the card at 70C, with only 1250RPM.
It's really really efficient and silent, and that's the main reason I was interested in it!

During my time, I took out the Vega 64 in the second slot (it's been there for really long), and I opened it to change the thermal paste, and boy oh boy was it dry as hell over there, I reapplied the stuff. Then plugged the 5700 XT at the first slot, and the 64 in the second one (just out of being used to doing so, thinking of it now, I should've plugged the 5700 XT alone). Anywho, I start my PC, and I'm idling, doing nothing, was busy with a new keyboard playing music in the background on the PC, only Groove music running, I have an old app to monitor GPU temps, which apparently doesn't work for the 5700 XT. The Vega 64 was sitting at 30C the entire time, so I was expecting the same from the 5700 XT.
Until I went to the drivers, checking the settings of it and the stats and everything, and I FOUND THE MOTHEREFFER SITTING AT 75C!!!!!!! The fans were at 0RPM, and the card is on a death wish to fry itself to death on its first day! I maxed the fans to make it chill, and came here to cry out loud.........

[Zando_]

Sounds like funky drivers. Are you making sure to run a WQHL version? They're usually more stable, but I have found the Adrenaline 2020 WQHL drivers are actually still unstable on my RVII (I still get random blackscreens with those and the latest 20.1.1 drivers). The only ones to be fully stable are Adrenaline 2019 WQHL drivers from mid last year. 

[iBoolGuy]

21 minutes ago, Zando Bob said:

Sounds like funky drivers. Are you making sure to run a WQHL version? They're usually more stable, but I have found the Adrenaline 2020 WQHL drivers are actually still unstable on my RVII (I still get random blackscreens with those and the latest 20.1.1 drivers). The only ones to be fully stable are Adrenaline 2019 WQHL drivers from mid last year. 

I never really tried that, I'm always at the latest drivers, and they really have resolving a lot of the annoying issues.. apart from mine apparently lol

[Zando_]

1 minute ago, iBoolGuy said:

I never really tried that, I'm always at the latest drivers, and they really have resolving a lot of the annoying issues.. apart from mine apparently lol

Yeah the Optional ones are hit or miss. Sometimes they fix stuff, other times they're absolute garbage. It's Radeon's biggest struggle right now. The hardware is there, just their drivers are really inconsistent. Once they get that fixed up, they should be in a much better spot. 

[iBoolGuy]

1 hour ago, Zando Bob said:

Yeah the Optional ones are hit or miss. Sometimes they fix stuff, other times they're absolute garbage. It's Radeon's biggest struggle right now. The hardware is there, just their drivers are really inconsistent. Once they get that fixed up, they should be in a much better spot. 

That is very true, I'm hopeful only because they show they've actually been working on them, things have been generally getting better.

With that said, I was just playing fortnite, and the card was silent, I put the Tuning settings for the card on Automatic, under automatic you can choose 'Underclock', 'Overclock', 'blabla', 'bla'. And there is 'Default', which is the natural default setting. I left it on that, and played for 40 minutes, opened MSI Afterburner, it was 93C..... I shut down the game, restarted the PC, and maxed the fans, then left the fans to idle, after 3mins of full RPM. Now it's idling at around 75C!!!

One thing yall may be able to help with, guide on upgrading the BIOS on those things? I checked the BIOS versions on the Vega cards, and it's older than the release date for the card. BIOS date from 2017! Those could definitely do with an update!

[Zando_]

1 hour ago, iBoolGuy said:

One thing yall may be able to help with, guide on upgrading the BIOS on those things? I checked the BIOS versions on the Vega cards, and it's older than the release date for the card. BIOS date from 2017! Those could definitely do with an update!

Eeeeeh flashing the BIOS is something I'd avoid unless it's an RX 5600 XT with the oof one. Risk of bricking your card isn't worth it.

As for the temps, your card seems to be being funky with the fan profile, tried setting a manual one to have the fans come on earlier?

[iBoolGuy]

On 1/24/2020 at 3:22 AM, Zando Bob said:

Eeeeeh flashing the BIOS is something I'd avoid unless it's an RX 5600 XT with the oof one. Risk of bricking your card isn't worth it.

As for the temps, your card seems to be being funky with the fan profile, tried setting a manual one to have the fans come on earlier?

Yeah.. after I looked into BIOS updates, seems like it's something you wouldn't normally do as in Motherboard BIOS flashing, only in critical cases as you mentioned, where there's a major shitshow that you gotta fix or smthn.
Apart from that, yes, I just started the PC, started room temps, dead cool, low 20C, and the fan profile is active and adjusted, the fans kick at 35% on 40C, and here, I'm typing this, and the temps are still climbing, it's 56C now.

 

[Zando_]

1 minute ago, iBoolGuy said:

Apart from that, yes, I just started the PC, started room temps, dead cool, low 20C, and the fan profile is active and adjusted, the fans kick at 35% on 40C, and here, I'm typing this, and the temps are still climbing, it's 56C now.

Just sitting at the desktop with your browser open? What are voltages and clocks like? Can you visually confirm the fans are actually spinning? 

[iBoolGuy]

On 1/24/2020 at 5:06 PM, Zando Bob said:

Just sitting at the desktop with your browser open? What are voltages and clocks like? Can you visually confirm the fans are actually spinning? 

Yes! Just sitting on desktop, with nothing open, with Instant Replay off, and absolutely zero GPU load, it would start, say, from 30C, keep climbing until the 75C, and stays there!
The voltages on the GPU core is 725mV while I'm writing this, nothing else open, the 725mV seems to be %75 of the way (based on the little speedometer graph), and in game it goes to like 1200mV. But, as the driver shows, it never goes below 725mV. The clock speeds are normal while idling (I suppose), they are very low, 50-150MHz. But the weird part about clock speeds, is the max boost, in every 5700 XT review I've seen, the card comfortably reaches 2050MHz and stays there almost all the time, only going down to like 1950MHz, my card goes down to the 1700's.. maxing out at like 1800's.
 

The fans are audibly and visually spinning.

One thing I did, I adjusted the fan curve, underclockd and undervolted the GPU, just to be safe. I started playing Fortnite, the fans weren't maxing out, my fan curve didn't push them to the max. So they were spinning at around 1800RPM, I opened the overlay information, the GPU Junction Temp was around 110C, and the GPU 'CURRENT' Temp (not sure what they mean by current, but I guess it's the regular temp we all know?) was around mid 80's C. I lost my shit... and allowed the fans to go to max, and continued playing, for the next 3-4 hours, the Junc temp was mid 90's, and GPU CURRENT temp was 75-85. So I guess it's... fine like this?

But I mean... the shitty situation, referring to many reviews, this card should run at %100 load, at 70C, and only 1300RPM fan speed.

I already submitted a ticket reporting an issue to AMD. But based on my past 3 experiences, they haven't replied once! So I guess as I mostly go about life, I only got "hope" on my side!

And btw, I tested with the 5700 XT alone, the Vega 64 being there or not made absolutely no difference at all! so there's that...