issues with a 1600x

[Coldkil]

Ok, i have had this random BSODs on a client's pc - it was a pretty standard build for his son who does a photo/video editing school and he wanted a "starter" machine that wasn't immensely underpowered but also not costing him way too much.

Basic specs are:
- Ryzen 5 1600x
- Asus Prime B350
- 16GB 3000MHz Corsair LPX ram
- GTX 1060 6GB from ASUS
- EVGA G3 650w PSU

The cases were mainly two:
- a VIDEO_SCHEDULER_INTERNAL_ERROR that popped basically randomly, either while gaming or using the Adobe suite, but also happened while idle; i searched everywhere for a solution but the only thing i found is that a number of people with a Ryzen+NVidia combo were having the issue. Only real workaround i found was a guy that set the Windows Power Options to "high performance", and disabling a couple of NVidia services.

- an INTERNAL_POWER_ERROR that started appearing after the previous one; a fast check and i discovered that CPU voltage was all over the place ranging from 1.3v to 1.5v even at idle. I should have solved that by setting manually the CPU voltage in the BIOS to 1.3 (via offset mode) AND returning to "balanced" power plan in Windows; now the CPU is at 0.8 while idle and reaches up to 1.3v on load (i didn't touch C states in BIOS).


Question is: i don't think the two things are actually connected and i am pretty sure that the CPU voltage was already all over the place even with the balanced plan but voltage set on "auto" in the BIOS. There's no overclock applied anywhere and i even reverted the RAM speed to the base clock (thus not running at 3000 for the moment, just for the sake of stability).

So, is in your experience something that should be solved with my intervention or is it a signal of a faulty hardware that needs to be replaced? My bet is on the motherboard if this is the problem.

[potis]

So, is in your experience something that should be solved with my intervention or is it a signal of a faulty hardware that needs to be replaced? My bet is on the motherboard if this is the problem.

And you are probably correct, didnt have this issue, but had the issue with "Wait 2 months until they BIOS patch so you can have stable RAM settings" on 2 builds with the 1000 series.

I can assume you flashed/checked the BIOS version on the mobo but it can just simply be that, try flashing the newer ones, quick check reveals a new version 1 month ago.

[Coldkil]

And you are probably correct, didnt have this issue, but had the issue with "Wait 2 months until they BIOS patch so you can have stable RAM settings" on 2 builds with the 1000 series.

I can assume you flashed/checked the BIOS version on the mobo but it can just simply be that, try flashing the newer ones, quick check reveals a new version 1 month ago.

Ok, i updated the bios like last week, i can check the version and see iof there's something newer on the ASUS website. If nothing solves, then i'll try to make them upgrade to a B450 that seems to have less issues (even with 1st gen Ryzen).

[Thunderball]

It's probably the power sensing circuit. Happens sometimes when you crack or pull away some SMDs while mounting a cooler. You should check what's the actual voltage on the CPU and RMA the board if the voltage actually behaves like that.

[Vash The Stampede]

3000MHz Corsair LPX ram with a Ryzen CPU? No kidding. I have a similar issue with my Ryzen 7 1700 with 3000MHz Corsair LPX ram but running an AMD RX 480. I had an ASRock AB350 Pro4 and that thing would BSOD once a week randomly. I have since switched over to MSI B350 Tomahawk and that doesn't BSOD anymore. It was a known issue with all ASRock B350's. I would think Asus has a reputation for better quality, but a lot of the B350's are kinda junk.

You may want to check that ram to see if it's Hynix memory, as Ryzen is known to run unstable with Hynix. I have 4 sticks of Corsair LPX memory but each stick is 4GB and one pair is Micron and another pair is Hynix. I've tested this memory to hell and back with memtest86 and some other tests all day with no errors, but when I removed the Hynix sticks my AMD drivers have stopped randomly crashing. It may not even be a Ryzen issue as I've heard Intel owners experiencing issues with Hynix. Run RAMMon to see if that memory is Hynix.

[Coldkil]

3000MHz Corsair LPX ram with a Ryzen CPU? No kidding. I have a similar issue with my Ryzen 7 1700 with 3000MHz Corsair LPX ram but running an AMD RX 480. I had an ASRock AB350 Pro4 and that thing would BSOD once a week randomly. I have since switched over to MSI B350 Tomahawk and that doesn't BSOD anymore. It was a known issue with all ASRock B350's. I would think Asus has a reputation for better quality, but a lot of the B350's are kinda junk.

You may want to check that ram to see if it's Hynix memory, as Ryzen is known to run unstable with Hynix. I have 4 sticks of Corsair LPX memory but each stick is 4GB and one pair is Micron and another pair is Hynix. I've tested this memory to hell and back with memtest86 and some other tests all day with no errors, but when I removed the Hynix sticks my AMD drivers have stopped randomly crashing. It may not even be a Ryzen issue as I've heard Intel owners experiencing issues with Hynix. Run RAMMon to see if that memory is Hynix.

Nice to know. I can check - voltages are now on check since i managed to fix them from bios, but i'm still waiting for feedback over the week. I can definitely check that.

EDIT: new day, new blue screen. Now it's an IRQL_NOT_LESS_OR_EQUAL, and at this point i'm kinda out of options if it's the hardware. Any suggestion to check something before just telling them that we need to replace some pieces?

EDIT2: if im going towards this route, i'm gonna hunt for a b450 and completely new ram like the gskill flare (probably a 2400 since prices are really high for that).

[Hotchocolate]

dont u have some hardware of ur own at hand to switch things like ram and gpu and see if it changes anything?

[Coldkil]

dont u have some hardware of ur own at hand to switch things like ram and gpu and see if it changes anything?

I have to check, probably i have the ram (it's a couple kingston sticks) but still isn't listed as compatible (though probably will work) and won't exclude the issues completely; i don't have a mobo, and GPU has been tested already and works perfectly.
@Vash The Stampede it is indeed Hynix and everything seems to point out the ram as the culprit.

[Thunderball]

EDIT: new day, new blue screen. Now it's an IRQL_NOT_LESS_OR_EQUAL, and at this point i'm kinda out of options if it's the hardware. Any suggestion to check something before just telling them that we need to replace some pieces?

All those blue screens sounds like overcurrent protection kicking in.

[Coldkil]

All those blue screens sounds like overcurrent protection kicking in.

May be. Changing the mobo is the next step - we'll try with a ram change (if it's possible, the guy decided it was better to buy it from somewhere else than amazon, his hassle to RMA it while i lend him my spare one) and if it's the same thing all over i'll make him go with a b450 and possibly something more "premium".

This until the CPU itself is faulty. Voltages are in check now since i fixed them in bios.

[Vash The Stampede]

I have to check, probably i have the ram (it's a couple kingston sticks) but still isn't listed as compatible (though probably will work) and won't exclude the issues completely; i don't have a mobo, and GPU has been tested already and works perfectly.
@Vash The Stampede it is indeed Hynix and everything seems to point out the ram as the culprit.

One other thing to look into is try raising the voltage on the SOC. The SOC clocks at the same speed as the memory and that tends to get unstable if the memory clocks too high. I think the default voltage is 0.95v or something, so try raising it to 1v or 1.05v. The Kingston should work, despite what they may say. You should also trying lowering the memory clock to something like 2666Mhz or about that area. If you do buy new memory then look for Micron or Samsung as those aren't problematic. Samsung's are especially good. Good luck finding a manufacturer who lists what memory manufacturer is used for their ram.

[Fascinate]

Just run it at stock and if he wants to overclock use ryzen master, i found that to work better than bios options when i had my x370 asrock. Its annoying that you have to do it after a reboot but it only takes a couple seconds to apply saved settings.

[Coldkil]

One other thing to look into is try raising the voltage on the SOC. The SOC clocks at the same speed as the memory and that tends to get unstable if the memory clocks too high. I think the default voltage is 0.95v or something, so try raising it to 1v or 1.05v. The Kingston should work, despite what they may say. You should also trying lowering the memory clock to something like 2666Mhz or about that area. If you do buy new memory then look for Micron or Samsung as those aren't problematic. Samsung's are especially good. Good luck finding a manufacturer who lists what memory manufacturer is used for their ram.

Just run it at stock and if he wants to overclock use ryzen master, i found that to work better than bios options when i had my x370 asrock. Its annoying that you have to do it after a reboot but it only takes a couple seconds to apply saved settings.

Point is everything is at stock already. I have even undervolted the CPU to keep it in check. I have deactivated XMP profiles on the ram. Changing RAM is kinda easy. Changing the motherboard a little more a hassle (and i want to go by steps).

[Wermys]

Motherboard or Power supply. Do you have a different power supply you could use temporarily to eliminate that as a factor? The only time I had problems with a Ryzen setup was with a x370 itx board supplied by biostar that I bought on the cheap. Which I shouldn't have bought I might add because its junk. It was really finicky about using Hynix ram at its rated speed and had to throttle it back to 2400 to get it to work properly even though the ram worked fine in a asus b350 prime and an x470 asrock itx board I had also. Anyways run another power supply in, fortunately you can just unplug everything and plug in an alternate in to test if you have one lying around. Otherwise as others have mentioned either ram or board. And I lean towards the board myself.

[Coldkil]

i'll have to do that too.

[Vash The Stampede]

Point is everything is at stock already. I have even undervolted the CPU to keep it in check. I have deactivated XMP profiles on the ram. Changing RAM is kinda easy. Changing the motherboard a little more a hassle (and i want to go by steps).

Trust me, if you're running that Hynix memory at 3Ghz then increase the SOC voltage. It may make a difference.

[Coldkil]

Trust me, if you're running that Hynix memory at 3Ghz then increase the SOC voltage. It may make a difference.

I'll do that, point is that i'm not running it at 3000. It's going at its base clock, and i'm still having the issue.

[Vash The Stampede]

I'll do that, point is that i'm not running it at 3000. It's going at its base clock, and i'm still having the issue.

So you're running the memory at 2133Mhz? If that's the case then a SOC voltage increase don't do anything. Almost everyone who runs Ryzen with memory speeds at 2666Mhz or higher needs to increase the SOC voltage to something like 1.1v. The highest I believe is 1.15v. But yea, Hynix memory for a lot of Ryzen owners has been an unstable mess for them.

[Coldkil]

So you're running the memory at 2133Mhz? If that's the case then a SOC voltage increase don't do anything. Almost everyone who runs Ryzen with memory speeds at 2666Mhz or higher needs to increase the SOC voltage to something like 1.1v. The highest I believe is 1.15v. But yea, Hynix memory for a lot of Ryzen owners has been an unstable mess for them.

Yes, 2133 stock - was the first things i touched coupled with CPU voltage to be sure that everything was in line and wasn't screwing things up. I think that's the main reason honestly, at worst i think something on the mobo was faulty. Once they get the ram replaced, we'll see what happens.

First blue screen seemed mostly related to GPU - never reappeared after i stopped the NVidia services. Same for the power error, fixed by setting a flat CPU voltage that's in line with the specifics. Now this IRQL error that i have usually seen with faulty memory sticks - and you give me a very likely solution. I'm pretty confident honestly.

[Ahovv]

Do you guys not have XMP profiles or something?