PC randomly loses power and restarts.

[Rahnak]

@tabascosauz Alright, just finished running the HCI memtest and it didn't catch any errors. Is it safe to rule out memory at this point?

 

[xtreemchaos]

i had the same problem about 4 months ago with a b450 aorus elite it turned out it was nothing to do the the mobo it was the psu, i got another and everythings hunky dory now.

 

[Chomiq]

@tabascosauz Alright, just finished running the HCI memtest and it didn't catch any errors. Is it safe to rule out memory at this point?
View attachment 165941

I just noticed, both you and me have Focus Plus Gold 750W. That being said mine is rock solid for months now and I don't see how my 1060 would trigger OCP on it. Serial number starts with R19 so I guess it's been manufactured in 2019 (purchased in mid August last year) so it should have the fixes for OCP applied.

 

[Bill_Bright]

Is it safe to rule out memory at this point?

No software based memory tester is 100% conclusive. RAM can pass all tests and still fail during normal use. If these testers report errors, the RAM is bad. But the only conclusive way to test RAM is to use sophisticated and very expensive test equipment, like this $2,495 Memory Tester (and that's for the cheap model)! So it is usually easier (and cheaper!) to swap in known good RAM and see what happens.

 

[Rahnak]

i had the same problem about 4 months ago with a b450 aorus elite it turned out it was nothing to do the the mobo it was the psu, i got another and everythings hunky dory now.

Did it occur randomly as well?

I just noticed, both you and me have Focus Plus Gold 750W. That being said mine is rock solid for months now and I don't see how my 1060 would trigger OCP on it. Serial number starts with R19 so I guess it's been manufactured in 2019 (purchased in mid August last year) so it should have the fixes for OCP applied.

Mine's R17 and I purchased it November 2017. It was rock solid until the board swap as well. If it turns out to be the PSU, it is an odd coincidence. Also, what is OCP?

No software based memory tester is 100% conclusive. RAM can pass all tests and still fail during normal use. If these testers report errors, the RAM is bad. But the only conclusive way to test RAM is to use sophisticated and very expensive test equipment, like this $2,495 Memory Tester (and that's for the cheap model)! So it is usually easier (and cheaper!) to swap in known good RAM and see what happens.

That's a big oof. At the moment I don't have another DDR4 kit I could swap it for. I have been eyeing Patriot's Viper Steel 3733 CL17 kit. A few people here on the forum had good luck with it and it has great price/perf. But I don't want to spend the money just to have slightly better numbers that I won't really notice in day to day usage unless I have some indicator that my memory is defective in some way.

 

[xtreemchaos]

yes it could go a week without doing it then do it 5 times in one day, it didnt matter if i was playing a game or it was on idle. it was like id turned the power off. the bad psu was a corsair cx 750w and i replaced it with a seasonic gx850w.

 

[Bill_Bright]

That's a big oof.

I know. Not many have access to spare, current DDR4 that they are will to risk. It can be a problem - even for pros.

 

[Ja.KooLit]

Ive had this last week and that was because I was OCing my memory. But at some point, even on all stock, it will still shutdown during gaming. Funny enough, it is not all the time.

Anyway, I have read about temperature of DIMMs can cause shutdowns and or instability. I placed a 120mm fan (smallest size that I have lying around) directly to the dimms. and seems to have fixed the problem. I have installed Aida64 extreme to monitor my DIMMS temp and I test both with or without fan. I noticed that when DIMM temps reaches 50 deg, I experienced some instability. I was thinking to buy a dimm fans (G.skill turbulence) and 1 with corsair.

Anyway, it has been 3 or 4 days already and so far did not experienced any shutdowns. So might wanna try to kinda ventilate your dimms.

 

[tabascosauz]

@tabascosauz Alright, just finished running the HCI memtest and it didn't catch any errors. Is it safe to rule out memory at this point?
View attachment 165941

Probably run it overnight once to be sure. Or was that already overnight? Around 7-8 hours gets me to about 1400% coverage on most of the instances.

Honestly, doesn't look a lot like a memory issue if it passes HCI. You mentioned that you used DRAM Calc once, do you already know what ICs you have?

On a side note, did you update your BIOS recently or purchase this board new? I had similar symptoms with random restarts without warning on BIOSes with early AGESA versions (1.0.0.0/1.0.0.1/early 1.0.0.2) on two different B550 boards. The crashing didn't always happen at load. The only real fix there was to wait for a more mature firmware.

Ive had this last week and that was because I was OCing my memory. But at some point, even on all stock, it will still shutdown during gaming. Funny enough, it is not all the time.

Anyway, I have read about temperature of DIMMs can cause shutdowns and or instability. I placed a 120mm fan (smallest size that I have lying around) directly to the dimms. and seems to have fixed the problem. I have installed Aida64 extreme to monitor my DIMMS temp and I test both with or without fan. I noticed that when DIMM temps reaches 50 deg, I experienced some instability. I was thinking to buy a dimm fans (G.skill turbulence) and 1 with corsair.

Anyway, it has been 3 or 4 days already and so far did not experienced any shutdowns. So might wanna try to kinda ventilate your dimms.

Are you...the same guy commenting on buildzoid's channel?

You have B-die. Of course it's temperature sensitive. If your B-die is reaching 50 degrees, it's obvious you need more airflow because B-die needs to be kept below 50 at all times. OP is having the random crashes at idle/desktop, not only gaming sessions.

OP is also on 3200/16 @ 1.35V, not pushing his RAM at 3733/3800.

 

[Rahnak]

Probably run it overnight once to be sure. Or was that already overnight? Around 7-8 hours gets me to about 1400% coverage on most of the instances.

Honestly, doesn't look a lot like a memory issue if it passes HCI. You mentioned that you used DRAM Calc once, do you already know what ICs you have?

On a side note, did you update your BIOS recently or purchase this board new? I had similar symptoms with random restarts without warning on BIOSes with early AGESA versions (1.0.0.0/1.0.0.1/early 1.0.0.2) on two different B550 boards. The crashing didn't always happen at load. The only real fix there was to wait for a more mature firmware.

That was during the day, but it was for nearly 8h as well. I didn't have anything else running, how is yours so much faster? Can't be the time of day.

I did purchase the board new, back in December. I updated to the latest available BIOS at the time and then I pretty much forgot BIOSes existed. Since then 3 more versions came out, as I posted earlier. If the PSU doesn't turn out to be the issue, my next step was probably updating the BIOS to the F20 version.

So about the memory.. I just went ahead and downloaded Thaiphoon again to get a print and just realized I made an embarrasing mistake. I could've sworn the first time I ran the program it said the chips were Samsung B-Die, which I thought was odd, but ok. So that's what I put in the DRAM Calculator. Turns out they're obviously C-Die so that explains why the timings didn't work. I also have some old screenshots of AIDA64 tests, one with this board at XMP speeds and two with the B450, the first at XMP speeds and the 2nd running at 3600 (totally forgot that happened). MSI BIOS has a dropdown were you can select several memory timings to try out and I picked one of the 3600 with the highest timings and it worked.

 

[Ja.KooLit]

Probably run it overnight once to be sure. Or was that already overnight? Around 7-8 hours gets me to about 1400% coverage on most of the instances.

Honestly, doesn't look a lot like a memory issue if it passes HCI. You mentioned that you used DRAM Calc once, do you already know what ICs you have?

On a side note, did you update your BIOS recently or purchase this board new? I had similar symptoms with random restarts without warning on BIOSes with early AGESA versions (1.0.0.0/1.0.0.1/early 1.0.0.2) on two different B550 boards. The crashing didn't always happen at load. The only real fix there was to wait for a more mature firmware.



Are you...the same guy commenting on buildzoid's channel?

You have B-die. Of course it's temperature sensitive. If your B-die is reaching 50 degrees, it's obvious you need more airflow because B-die needs to be kept below 50 at all times. OP is having the random crashes at idle/desktop, not only gaming sessions.

OP is also on 3200/16 @ 1.35V, not pushing his RAM at 3733/3800.

nope. this is my first time to comment about memory wahaha. Who is bullzoid?

Oh and thanks. I didnt know about B-die.

 

[John Naylor]

1. Your image shows the list of errors, but to diagnose the problem, you need to actually select the warning or error and examine what it says. The message you posted in Message # 11 occurred during the reboot process. It's basically saying "Hey, I'm rebooting cause something bad happened .... Go back in the logs to the point where system crashed and see what caused the problem".

2. Loading the XMP profile loads various changes to the SPD profile which increase voltages and other setting as necessary to allow the RAM too run at that (overclocked) speed. You don't want to try running w/o this profile loaded unless you have a known stable condition, and want to see if you can lower voltages. Saw this note from Intel when DDR4 1st arrived on the scene; can't speak to AMDs policy.

"1.2V or lower = Best for DDR4 .... 1.35V = okay voltage for overclocking kits .... 1.5V =absolute max voltage allowed for Intel XMP 2.0 profiles and max suggested voltage"

3. We always run memtest86+, before leaving the office ... check results next morning ...at least 12 hours.

4. Since it is a new board, have you gone thru the process of troubleshooting the board w/ MoBo Tech Support via phone ? This step is recommended and will be required anyway before doing an RMA. Note Gigabyte (and Asus) TS won't talk to you unless you have latest BIOS installed. At least that has been my experience in recent years.

5. You have an excellent PSU ... a PSU swap is still worth a shot ... might also try swapping the cable ... or temporarily installing w/o the big bend/ Finally make sure it's a UPS cable and not in 8 pim PCIE cable. Some MoBos use same connector socket.

6. What are CPU voltages and core temps ? Do RAM voltages match the XMP Profile ? ... might have to tweak a bit more.

7. You changes boards ... did you format the boot drive and reinstall windows ? Did you remove all the drivers supplied with the old B series board before taking it out ? Did you install all the drivers that came with the new X series board ? Sometimes the "install new over old" doesn't remove all of the old stuff and Windows is trying to load drivers for hardware it can't find.

8. A GFX card issue doesn't seem likely due the trigger points being all over the place.

 

[tabascosauz]

That was during the day, but it was for nearly 8h as well. I didn't have anything else running, how is yours so much faster? Can't be the time of day.

I did purchase the board new, back in December. I updated to the latest available BIOS at the time and then I pretty much forgot BIOSes existed. Since then 3 more versions came out, as I posted earlier. If the PSU doesn't turn out to be the issue, my next step was probably updating the BIOS to the F20 version.

So about the memory.. I just went ahead and downloaded Thaiphoon again to get a print and just realized I made an embarrasing mistake. I could've sworn the first time I ran the program it said the chips were Samsung B-Die, which I thought was odd, but ok. So that's what I put in the DRAM Calculator. Turns out they're obviously C-Die so that explains why the timings didn't work. I also have some old screenshots of AIDA64 tests, one with this board at XMP speeds and two with the B450, the first at XMP speeds and the 2nd running at 3600 (totally forgot that happened). MSI BIOS has a dropdown were you can select several memory timings to try out and I picked one of the 3600 with the highest timings and it worked.

View attachment 165952View attachment 165953View attachment 165954View attachment 165955

If that 16-19-19 profile is actually stable then that's actually a pretty stellar achievement for C-die. Everywhere I look, Samsung 8Gb C is one of the worst DDR4 ICs you can get. People are reporting that they can't get up past 3400/18, and that DRAM voltage hurts C-die overclockability above even just 1.35V.

Flare is a bit of an older kit; my older Trident Z kit was actually mislabelled by G.skill as 4Gb D-die in Thaiphoon, it was E-die. If the sticker on your sticks has a 042 code (right above the barcode) that ends in ...8810C, then Thaiphoon's probably right, C-die.

There's not a whole lot you can do with C-die beyond what you're already at. Also, those AIDA results look okay on latency but low on bandwidth. Whether that's just a single-rank performance deficit, I dunno.

You should probably try a newer BIOS if the crashing continues. What's the current AGESA for your current BIOS revision per CPU-Z (mainboard tab)?

C-die is pretty uncommon in DRAM discussions in the desktop space. I keep hearing the same claim tossed around that C-die is designed for laptop SO-DIMM use, which would explain its negative scaling beyond 1.35V and more or less topping out at 3200.

nope. this is my first time to comment about memory wahaha. Who is bullzoid?

Oh and thanks. I didnt know about B-die.

I see, weird coincidence lol. Well, it's B-die doing 3800/15 so you know what they say - with great power comes great responsibility only B-die is extremely sensitive to temperature

 

[Rahnak]

@John Naylor Thanks for the input. To answer some of your questions.

1. I just realized I cut some important information from the 2nd event viewer screen. I meant to show that the Error entry said the system restarted at 1:59 pm but the last log I have is from 10 minutes earlier, so nothing really that can tell me what went wrong. Here it is.


2. I've read XMP can sometimes be unstable on AMD, but I haven't been able to get near XMP speeds by setting the timings manually. It goes without saying I don't have a lot of experience with overclocking.

3. I ran it for about 8h once. No issues. I could certainly give it another go.

4. No, I haven't reached out to Gigabyte tech support. Decided to try my luck with TPU members first.

5. That sounds like a mistake I could've accidentaly made. Would the 8pin cable even work if it was connected to the PCIe socket?

6. CPU temps are between mid 40s-50s (summer without AC). SoC voltage is 1.063V. DRAM voltage is between 1.356 and 1.368. A little higher than 1.35 but close enough? Reading were taken from HWiNFO64.

7. I did a clean Windows install when I switched boards, yes.

@tabascosauz It probably wasn't 100% stable. I don't remember having the system at 3600, so I most likely went back to the XMP profile. I bought this kit because I read some "horror" stories about memory compatibility with Ryzen and that was one of the advertised features of the FlareX line. And also because it was cheap. In hindsight it was not a good choice.
Took a picture of the sticker, c die confirmed.


Current AGESA version is 1.0.0.4. F20 is AGESA ComboV2 1.0.0.2

 

[Deleted member 193596]

can you run prime95 (stock cpu)?
can you run Memtest for 10 minutes without an error?

yes? replace your PSU.

 

[Waldorf]

@John Naylor
lots of ppl with the better Aorus boards (x570) have sudden reboots, not one has any log showing why,
as its a crash-reboot and win cant right anything.

Technically amd dont have to run with XMP enabled, as its intel, amds is called AMP/A-XMP,
and only kits with that label are officially "certified" to do clocks/timings listed.
That said, most x570 should be able to do xmp if on newer bios.

Gb support is useless, as they officially still claim there are no issues (stemming from bios), unless from someone here and there stating
they know about it (being a problem) and rmaing a board that has issue from bios chip will not fix anything, as you get the same thing again.
outside the faact that lots of ppl have issues using the latest bios, so that as a requirement to get help with +200$ hw is a joke.
another indicator is that most of their x570 boards already have multiple revisions, mine had 2 within the last 4 month, why if there is no issue?


@WarTherapy1195
stability in prime95 or memetest doesnt mean its the psu (or not).
mine would pass both and still had issues.