New RTX 2080 Ventus 8G Crashes With Multiple Games, no GPU O/Cing..

[Sithtiger]

Hi, I upgrade from an MSI GTX 1070 Gaming 8G, 4th gen i7 4790k O/C to an Asrock Z390 Phantom Gaming SLI/ac, 16GB of G.Skill DDR 4 PC3200. Using XMP profile and slightly O/Ced i5 9600k to 4.6GHz. Anyway, it was rock solid with O/Ced 1070 Core +185MHz memory +475MHz. Anyway, everything was until I started testing it on games. I uninstalled the newest driver. Since the 1070 and 2080 use the same one, but use DDU. Bought MSI RTX 2080 Ventus 8G. Installed 417.71 again. Oh, I have a Dell SE2717Hr 27 inch that only has two outputs. A VGA and an HDMI. On my 1070 I'd use the HDMI to port to the TV and DVI-D using an HDMI cable with an HDMI to DVI-D adapter. Now, I have to use HDMI to Display port cable. No added adapter, just the input to the computer has a Display Port connector. Anyway, I didn't O/C it at first, played BFV for a few hours, no problem. Then I played Warhammer 40000K Retribution and it crashed immediately. I thought it was the game. Then I tried playing SWTOR, Star Wars: The Old Republic and it ran fan for a while, but it crashed. When it crashes it's a hard crash and causes a complete reset. If I use Kombustor, it crashes after about 5 seconds...hard crash...no BSOD.

One thing that I've noticed though is when my core clock can go up to 2045....something like that, memory 70000. Its boost clock is only supposed to be 1710. I know it can go higher, but why is it auto Overclocking when I don't have anything running. No Afterburner, nothing? It usually hangs around 1800 to 1860, but why? So when I set Afterburner Core to 0 and memory 0 (meaning I want it to run at stock speeds) it still O/Cs there too. I've tried doing that while setting a high custom fan curve. When it crashes, it's usually in the low 60c, sometimes lower. Would anyone happen to know why this is happening?

Oh, I don't know if this means anything, but when I opened up the packing, the box itself was in perfect condition, but there was no plastic wrap on it and no tape on either side. It looks brand new, but they ALWAYS have tape on them at least... I'm worried this was a return or something. Anyway, if anyone has any idea, I'd be grateful! Thanks in advance!

 

[Sithtiger]

Bottom line is I think it's overclocking and I don't have Afterburner running, I've disabled it. How is it auto O/Cing past 1710MHz up to 200MHz sometimes by itself and how do I stop it?

 

[VoodooNinja]

Bottom line is I think it's overclocking and I don't have Afterburner running, I've disabled it. How is it auto O/Cing past 1710MHz up to 200MHz sometimes by itself and how do I stop it?

My MSI RTX2070 Ventus came wrapped in plastic cellophane. Your card might have been a return. It's normal for the card to boost oc to those numbers. Although mine a 2070, it hits that 1800 to 1850 oc too.

 

[noko]

A return with a modify bios?

DDU again and load drivers to ensure no resident settings in registry or drivers are being used regardless if OCing software is open or not.

If still crashing, return it and get your money back is my recommendation.

 

[Maddness]

First thing i would do is put your whole system back to stock settings. No overclock at all, this means CPU and memory. I have seen stable PC's crashing with new cards before.

Also what is your PSU as you never listed it, because you say are saying it is just rebooting without bsod, this can be a power issue.

Start there and let us know how you get on.

 

[Sithtiger]

My MSI RTX2070 Ventus came wrapped in plastic cellophane. Your card might have been a return. It's normal for the card to boost oc to those numbers. Although mine a 2070, it hits that 1800 to 1850 oc too.

Yeah, that's what I thought. I think it's a return as well. I wouldn't be concerned about the numbers if it wasn't crashing, but it is when it's pushed.

A return with a modify bios?

DDU again and load drivers to ensure no resident settings in registry or drivers are being used regardless if OCing software is open or not.

If still crashing, return it and get your money back is my recommendation.

Yeah, gonna try a few things. Agreed. I'm gonna uninstall the drivers with DDU and then reinstall with the drivers just before that. I think I'll uninstall Afterburner as well too. Thanks for the help!

First thing i would do is put your whole system back to stock settings. No overclock at all, this means CPU and memory. I have seen stable PC's crashing with new cards before.

Also what is your PSU as you never listed it, because you say are saying it is just rebooting without bsod, this can be a power issue.

Start there and let us know how you get on.

Yeah, gonna try that. Gonna uninstall Afterburner for now. It's not starting up, but still. If it still crashes, I'm gonna try put the last drivers before these on. If it's still happening, then I'm gonna return it. As I said earlier, there was no cellophane on the package and no tape on the box at all! The card looks new, but I'm betting someone else had this problem and this was happening and returned this card. They checked it, found it was fine and put it back on the shelf. I did notice a bit of dust on the underside of the card, but that could have been from my HSF. I did a cleaning before that though, so I'm not sure if it was some residual dust or what. When it crashes, I looked at the logs with GPU-Z and temp isn't that high, so I don't get it. My PSU is a Cooler Master HX 850 Modular PSU. I've had it a while, but I have a 1070 O/Ced decently and had no problems whatsoever. Here we go. I'm hoping it's drivers, but you're right, an O/Ced CPU and/or RAM could cause this too, as well as a PSU, but I think the PSU is OK, but I need to check it with a multimeter.

Thanks for replying!!! I'll keep everyone updated as to what's going on. There are not too many things to check here, so I should be able to test all this in the next 12 hours easily.

 

[Sithtiger]

So, I tried everything I stated trying above, no dice! Same problem. I came across some other threads regarding the RTX 2080 Ti and it sounds like the same exact problem! I know 2080's had a few problems too, but not nearly as bad as the 2080s. That and I believe this one was returned. It doesn't crash on every game, or even some bench tests are fine. I tested Firestrike and Firestrike Extreme under Stress Testing and it was fine. Same with Sky Diver is it and one other one....all fine. However, if I Load Kombuster and try GF OC Test (GL4) that test causes it to reboot usually under 2 seconds, however, I had it running sometimes, but just when I think the problem is gone or fixed, the same thing happens!

Something I didn't check and that's to make sure the card is seated and the PCI-E cables. Someone else also asked if each PCI-E cable had its own cable. So, a separate cable for each plug. I don't think that matters, but I'll give it a shot. I'm gonna go back on drivers once more. I don't think that's It but wanna try and see. It would be nice if it was just one of the cables was not plugged in all the way. I highly doubt it as I'm overly careful. Almost to a compulsive point, but I will check because stranger things have happened before. I think the card suffers from the same fate of most RTX 2080 Ti owners and I'm guessing a few 2080 users as well.

I forgot to mention as well that it has a very loud coil whine, which I can put up with. My 1070 had it, both my 970's had it. Anyway, with the 2080, just before it crashes, the coil whine gets noticeably louder and more intense. Here's a log file from GPU-Z. It's not the temp or fan speed. The core clock gets high sometimes, but when playing BFV, it's fine. So far, it's never crashed, but then again I've never played it on this card for more than an hour I believe. Anyway, here's a log file from CPU-Z

 

[Sithtiger]

OK, so I was able to get Newegg ship out a new GPU first. It will be here Monday. So I put my 1070 back in and ran those tests. It had a pretty bad coil whine too and actually sounded very similar to the 2080, but it didn't crash on any test or game, not once! One thing I noticed while testing. I was using GPU-Z to monitor temps. On the 2080, if I switched from the bench test or game to look at the telemetry, it just kept on running, however doing the same thing with the 1070, when I'd switch to GPU-Z, the GPU would ease the workload down. The 2080 wouldn't do that. It kept on working. It's as if the accelerator on a car was stuck and just kept running after you take your foot off of it.

Also, I think the other problem is the same as the 2080 Ti's have or had....the GDDR6 memory is the culprit. It's mostly the same things I was experiencing. At first, Nvidia said it was just the FE cards, but AIB's have it too. 2080s aren't affected nearly as bad, but it affects some. I wish I had taken a screenshot of the specific memory manufacturer, but this link from Forbs talks more about the problem too. God, I hope the replacement doesn't have the same problem! I also mentioned that mine had no cellophane and there was no tape on the box.

 

[Sithtiger]

Got my new card today. They sent it overnight. I have a request to ask of anyone that would do it and that is if they could download MSI Kombustor. I'll link it so you can download it if you don't have it. The test that always kills it is called GF OC Test (GL4). Just click 'Run Stress Test Settings panel' button. My 1070 didn't crash at all It says this test fails when OC settings on GeForce Fermi & Kepler. It says this test fails when OC settings are too high. The thing is, I'm not even running Afterburner. It's not even setup. If it was the only test crashing, while all games were fine, I'd be fine, but the first GPU crashed crashing on this, it wouldn't be a big deal, but it crashes on certain games when it shouldn't crash at all. So the new card does the same thing as the first one did with this GF OC Test (GL4), however, it did NOT crash playing Warhammer 40,000K: Retribution. As I said it was fine on that but had r before on the first 2080 and occasionally it ran it OK, but it did seem like it was about to crash on Warhammer 40K, but hasn't so far. The real tests. That is the coil whine was fairly bad but hadn't crashed yet. It also crashed playing SWTOR (Star Wars: The Old Republic). An older game that came out around October 2011. That crashed after about 30 to 40 minutes and I was just standing by the GTN. Why oh why would it crash there.

Someone mentioned...maybe another forum. asked if my PSU was working OK. A valid question, but it's a very good question. Wanted to know about my PSU. So I have a Corsair HX 850 Modular PSU. It is giving enough power out, especially on the 12 volt rails because the 1070 was O/Ced as well as my CPU and RAM. The 1070 never crashes on anything. When the 2080 crashes, there's no BSOD, it jussfdaat reboots which sounds like a PSU problem but I think it's fine, because I put my 1070 back and I ran all the same tests on the 1070 without incident at least 3 times each or more.

So, right now, those are the two tests or bench test that causes spontaneous reboots with no BSOD or even artifacts, just reboots. I'll let you know what happens with this one. BTW, this one so far only reboots using MSI Komustor GF OC Test (GL4), Warhammer 40k. I'll try some other tests and see what happens. Thanks in advance guys!

 

[Sithtiger]

I'm not done with the card yet, but so far it seems amazing. I was wrong about one test though. The GF OC Test (GL4). The test is fine, but the mistake was mine. I will explain in a second, but the bottom line is the GPU was bad. It still crashed 9 times out of 10 on Warhammer 40K Retribution. The first card that is. The 2nd card has been solid with everything I 've thrown at it. Doesn't crash with SWTOR and it doesn't crash Warhammer 40K at all. It performs correctly. When you Alt+Tab out of a game for example. When you Alt+Tab out of a game and back to GPU-Z or some other monitoring program, the temps go down, everything returns to normal immediately. That was not the case with the first one.

Now, talking about Kombustor. Feel free to try it, but I see I was using an older version. They work backward, the way they number their version numbers work. Most software, the higher the number, the older it is. The link I put was for the newest one. Mine was v 3.5.2 from 2015, so if you want to take the test I did, you'll need to get v. 3.5.2.