Question Ryzen 1700 Instability?

May 18, 2001
171
0
81
#1
Hi there,

I've had a build for about 2 years now, and have been dealing with system instability for the last 4 months or so. I've followed the stickies and THINK it's my CPU and wanted some advice. I noticed 2 main sorts of bugs so far:
  1. System crashes when mining Bitcoin (NiceHash). I'm using both GPU and CPU mining here. It's weird, the PC doesn't shut off or throw an error, but I notice it goes "quiet" (i.e. fans on low, GPU no longer cranking) and the screen is black and unresponsive. Have to do a hard reboot and then it's fine.
  2. I get the occasional screen tear, I think mainly when gaming. Always on the right hand size of my screen, its a half a second vertical column of pink/purple/green. Can't be more than 200 pixels wide (<5% of my screen) and goes away right after.
Given these errors, at first I thought it was my GPU. I did an RMA with gigabyte but am getting the same issues even with the new one, so it cannot be that.

Since then, I've done the following:

  1. Ran MEMTest with no issues (got to "100%" or beyond with no errors, had it running for a few hours)
  2. Ran Prime95 and am getting the same "black screen" crash as with mining. I don't have an error log or anything to go to there.
My Ryzen was slightly overclocked (3.8ghz turbo), so I brought it down to 3.6 (so below spec I think?) but still getting issues. At a 1.37 voltage.

What are my next steps? How can I further test the CPU?

My system is as follows:

  1. Ryzen 1700
  2. Nvidia GTX 1080 Ti
  3. 16gb RAM G.Skill FlareX
  4. Asus X370 Taichi
  5. Seasonic M12II 620wt PSU
  6. Crucial M4 128gb SSD
  7. Samsung 840 and 960 EVO SSD
 
Aug 25, 2001
42,942
373
126
#2
I was in much the same boat, not too long ago.

I have a Ryzen R5 1600,
was using an ASRock AB350M Pro4 micro-ATX (that I kept updated), now I'm using a Gigabyte AX370-Gaming ATX
was using 2x8GB Team Group Vulcan DDR4-3000 @ 2933, but when I doubled-up the kit to 4x8GB, I had to drop the speed down to 2400. Now running a GSkill Ripjaws V DDR4-3000 16GB (2x8GB) kit
using various AMD RX-series GPUs, RX 460, 470, 560, 570.

Mining in Win10 64-bit with Nicehash.

Was initially overclocked to 3.8Ghz on 120mm AIO water, but kept having the occasional (maybe every 2-3 weeks) "glitch". I also saw those graphical glitches while mining, after a few days, just a split-second sort of vertical rectangular black appearing. I still wonder if those were command-prompt windows, and the Nicehash client was trojaned, and I got hacked. The download seems to get flagged.

I finally decided to remove the overclock, and run it at stock. The incidents of "black screen / monitor sleep/wake hang", have greatly decreased.

Still, my primary slot in the AX370-Gaming seemingly died. I have yet to swap CPUs to verify if it was the slot or the CPU PCI-E lanes.

I would remove the overclock, it matters little to the CPU mining. (For me, literally a couple of pennies a day.)
 
Aug 25, 2001
42,942
373
126
#3
Ran Prime95 and am getting the same "black screen" crash as with mining. I don't have an error log or anything to go to there.
That might be related to your PSU.

I also lost an RX 460 card early on, like maybe a year or so in. Was using an Antec Neo Eco 620(C?), which is supposedly a "good" SeaSonic-made model. (Or was that Delta?)

I am currently using an EVGA G1+ 80Plus Gold 650W.

Before I killed the primary PCI-E x16 slot, I had two RX 570 8GB MSI cards cranking, mining. I pulled 460W at the wall according to my KAW. By comparison, Folding was like 390W.
 

coercitiv

Diamond Member
Jan 24, 2014
3,046
332
136
#4
OP, I doubt it's the CPU - the first two possible culprits I would consider are the mainboard and then the PSU. Given this happens with no load on the GPU as well, I am inclined to think it's the mainboard.

A fault within the mainboard CPU power stage could alter the voltage the CPU receives and make the system unstable. (If true) this type of degradation could be gradual, with one or two isolated incidents at first, then more and more problems as time passes by, culminating with failure to boot into Windows towards the end. Increasing voltage settings and/or lowering frequency might help a while, but only as temporary solution.

If you can, test with other components - starting with another PSU. If you have a friend with a Ryzen system willing to do some testing, remove the CPU and try it in another system. The PSU and CPU are arguably easier to remove, that's why they usually come first. You can also do the same thing with the MB as you did with the dGPU, just RMA and get another one (or get it back tested at least) - however I would only do that in conjunction with making sure the PSU isn't faulty either.
 
May 18, 2001
171
0
81
#5
Ah no way, glad I asked here! @coercitiv that narrative would definitely fit what happened with me. The bugs were rare at first, now is consistent.

@Larry, I switched my mobo settings back to default and am running Prime95 again. Hopefully it will be stable. The OC was more for fun and gaming than mining speed anyway.

Given what you guys are saying, I think I agree the PSU is less likely to be the issue. "Pure" GPU mining works fine, it was just when CPU was being used. Interestingly it DOES crash on benchmark mode, but I am assuming that is because it wants to test the CPU as well.

It's unlikely, but possible I could get access to an alternative PC this weekend to test. Otherwise sounds like the best route is to just RMA the mobo and see, right?
 

coercitiv

Diamond Member
Jan 24, 2014
3,046
332
136
#6
Otherwise sounds like the best route is to just RMA the mobo and see, right?
If alternate components for testing are not available then yes, you ought to consider what to RMA next. My bet would be the mainboard - but you have to remember - I'm just expressing opinions in blind mode here, with no time spent on that computer and little info on your warranty conditions.
 
Aug 25, 2001
42,942
373
126
#7
"Pure" GPU mining works fine, it was just when CPU was being used. Interestingly it DOES crash on benchmark mode, but I am assuming that is because it wants to test the CPU as well.
Is the CPU checked-off as an active device for mining, before doing the benchmarks? If not, then it won't benchmark the CPU. It can still crash, if your GPUs are drawing too much power, etc. I had to lower my Power Limit in Wattman, in order to be able to pass the bench,mark without crashing, for the GPUs. I had to do that both with RX 570 cards, as well as R7 260X cards and 270X cards.

Also, if you want to "summon" me, use @VirtualLarry . @Larry is a different member here. (Hey Larry!)
 

cfenton

Senior member
Jul 27, 2015
257
23
71
#8
H
My Ryzen was slightly overclocked (3.8ghz turbo), so I brought it down to 3.6 (so below spec I think?) but still getting issues. At a 1.37 voltage.
The 1700 is 3.0ghz base, 3.7ghz single-core turbo, and 3.2ghz all-core turbo, I think. So if you're setting the multiplier to 36, you're still above base spec. I think the other posters suggesting it's the motherboard or PSU are probably right, but it's worth testing the CPU at stock settings just to be sure.
 

Shivansps

Platinum Member
Sep 11, 2013
2,394
157
126
#9
Im having a very SIMILAR problem with my 1700... from November or so, the pc just "hangs" but only on idle, never with load.

For example i could be gaming for 4 hours, everything OK. Then i close the game and go to the bathroom, when i come back the pc not longer responds, it is just "stuck" on the Windows desktop.

Same thing happens when i turning on the pc next day, some times it gets stuck on a black screen just before showing up the Windows login.

I had this happen when i was on the pc, the mouse starts to get slow and then the whole thing hangs.

I really had no time to look for a solution yet. It dosent seem like a common hardware failure to me, there is no bsod what is really STRANGE...
 

killster1

Diamond Member
Mar 15, 2007
4,235
85
91
#10
Im having a very SIMILAR problem with my 1700... from November or so, the pc just "hangs" but only on idle, never with load.

For example i could be gaming for 4 hours, everything OK. Then i close the game and go to the bathroom, when i come back the pc not longer responds, it is just "stuck" on the Windows desktop.

Same thing happens when i turning on the pc next day, some times it gets stuck on a black screen just before showing up the Windows login.

I had this happen when i was on the pc, the mouse starts to get slow and then the whole thing hangs.

I really had no time to look for a solution yet. It dosent seem like a common hardware failure to me, there is no bsod what is really STRANGE...
your problem sounds gpu related. i would update my gpu, check the cables dust it out / software update
 
May 18, 2001
171
0
81
#11
@VirtualLarry I switched all my mobo OC settings to default. Prime95 was able to run over night. Trying NiceHash (did have only GPU checked) again now.

It may be incidental, but oddly when gaming now my rig is far more unstable. Getting 3x more crashes than I used to in one game (two other games not affected).
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
17,232
632
136
#12
@VirtualLarry I switched all my mobo OC settings to default. Prime95 was able to run over night. Trying NiceHash (did have only GPU checked) again now.

It may be incidental, but oddly when gaming now my rig is far more unstable. Getting 3x more crashes than I used to in one game (two other games not affected).
Sounds like a video card problem ??

Use software to UNDER clock that 1080ti, and that should tell you for sure.
 
May 18, 2001
171
0
81
#13
OK I'll try that. Would be ridiculously bad luck, this is a new 1080 Ti that I got back from Gigabyte on an RMA 2 weeks ago.
 

JDG1980

Golden Member
Jul 18, 2013
1,532
20
136
#14
Since the 1700 is a first-generation Ryzen CPU, it might be worth considering the possibility of the segfault bug being an issue. Even though the steps for replication involve Linux, there's no reason to believe it can never affect Windows. Instructions to reproduce the bug can be found here - if it exists, AMD will replace the CPU.
 

oynaz

Platinum Member
May 14, 2003
2,448
1
81
#15
In my entire career, I have seen a faulty CPU causing intermittent crashes exactly once. They tend to either work as intended, or not at all.

By your symptoms, I would rate the possible culprits as mobo 75%, RAM 20%, PSU 4%, CPU 1%.

With the caveat that I am Europe-based. I understand that power can be quite flaky in the US, so if you have that issue, swap the PSU and RAM possibilities.
 
Oct 9, 1999
11,266
47
126
#16
In this order i would check/swap the following with known good parts:

1. PSU
2. Mobo
3. RAM
 
May 18, 2001
171
0
81
#18
@Smoove910 I used the UEFI auto-updater tool and apparently I have the latest BIOS

As an update: setting my mobo/OC settings back to default have fixed most of my system stability. I can mine, pass Memtest and Prime 95, and game mostly fine.

Oddly, I have on game that has become incredibly unstable ever since this new RMA and system change. In-game performance is OK, possibly a bit worse, and loading performance (e.g. loading a new map) is significantly worse (I used to load first, I now load last amongst my friend group).

Given it's just this game I am thinking it could be software related, but so odd that I am getting far buggier now than I did 3 months ago with the exact same set up.
 
Last edited:

AnitaPeterson

Diamond Member
Apr 24, 2001
5,564
3
81
#19
I would add my voice to those who say the PSU might be at fault. Try a new one, with more wattage. Return it if the troubles persist. But dirty power would do very nasty things to a system.

And even though I like AMD, I believe their chips are more sensible to this phenomenon than Intel ones.
 
Last edited:
May 18, 2001
171
0
81
#20
@Markfw, I will try underclocking on my next gaming session. I've also tried a clean install of the game in question.

So an odd thing happened this morning - I had a weird screen bug ( ). The screen and system were working fine. Acer screen menus popped up normally, and when I plugged and unplugged my iPhone I could hear the "ding" sound of a USB device being plugged in.

This screen error is similar to the "screen tears" I was experiencing before. Only difference is this did not end, and was across the entire screen instead of a momentary <100 pixel wide set of lines.

Consequently, I dived right back into system testing. I was able to do the FurMark for over 1.5 hours (I was told going longer is risky for system safety due to temps?). I am currently doing the OCCT 3D GPU test.

OCCT also has a PSU test. I am thinking of running that over night.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
17,232
632
136
#21
That reeks of a bad GPU. underclocking it, and under volting it would prove that. That screen print almost assures that its a GPU problem.
 
May 18, 2001
171
0
81
#22
Is there a specific stability test you'd recommend? So far been doing a medley of gaming/mining with Fur Mark or OCCT, but happy to try something else.

The bright side is I'm actually using a Gigabyte 1080 Ti Gaming OC 11G, so the clocks being to heavy have more merit.

On the flip, wow I am never buying Gigabyte again. This is my 3rd RMA with them. First RMA they shipped me had a bent metal IO plate, and now this one was clearly DOA.
 
May 18, 2001
171
0
81
#23
OCCT stability test has given me some interesting results:

1) I failed ("caught error") when doing one PSU test. It doesnt say what the error is, but has provided me detailed voltage graphs for all my hardware. Would anyone know how to interpret them?
2) Generally for the GPU and PSU test, while it doesn't "fail", it does stop after a few hours (randomly) saying "user stopped". I am not stopping it, so it's either a software bug or there is a minor crash?

Has anyone used this test?

In either case, I assume this is more evidence it is the PSU? I'm filling out RMAs for everything at this point so might be getting new hardware all around :p
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
17,232
632
136
#24

Rudy Toody

Diamond Member
Sep 30, 2006
4,006
65
106
#25
Since the 1700 is a first-generation Ryzen CPU, it might be worth considering the possibility of the segfault bug being an issue. Even though the steps for replication involve Linux, there's no reason to believe it can never affect Windows. Instructions to reproduce the bug can be found here - if it exists, AMD will replace the CPU.
My 1700X had that bug. It would show up intermittently on BOINC, viewing HD video, and while using Mathematica. AMD replaced it and everything is swell!
 


ASK THE COMMUNITY

TRENDING THREADS