PC Processors

[buhster]

Hello. For a year now, since the moment of purchase, I suffering from the problem of randomly dead freeze whis black screen. I have never used overclocking CPU and already replaced the power supply, RAM, video card, CPU cooling and improved the cooling of the case. The problem remains.

I can not understand the pattern. The temperature of the processor does not rise above 60 °C as the system temperature (in fact, I have never seen more than 57 °C). The hang is not related to the load, because may appear at any time (under full load in the game and without load at all). Operating time without hanging up varies from 1 minute to 16 hours. And this is not a BSOD - just a black screen with no signs of life, requiring a reboot. I tried to lower the voltage vsoc. I thought that the motherboard tore them too high. I tried to use stock values. I tried to use RAM from 2400 to 3466 recommended by the manufacturer of the motherboard as fully compatible. Nothing helps. In the end, I had two suspects left: the motherboard and the processor. Help me to understand.

My system at the moment:

• CPU: AMD Ryzen 7 2700
• MB: MSI x470 Gaming Plus (BIOS ver.: 7B79v19 as latest stable)
• RAM: Kingston HX434C19FB2K2/16 (DDR4 16Gb (2x8Gb) 3466 MHz pc-27600 HyperX FURY Black, was previously installed RAM 2x8Gb DDR4 Samsung M378A1K43CB2-CRC)
• GPU1: Radeon RX Vega 10 (MSI Radeon RX Vega 64 Air Boost 8G OC, was previously installed Gigabyte RX 480 8G Gaming 1)
• GPU2: Quadro P400 2GB (HP 1ME43AA - removing the card does not solve the problem)
• Sound card: Asus Xonar DX (removing the sound card does not solve the problem, sound card integrated into the motherboard is disabled)
• NVMe SSD: Samsung Evo 960 (MZ-V6E250BW)
• SSD: SanDisk SDSSDHII240G
• PS: Seasonic Focus+ Gold 750FX
• CASE: fractal design define R4 (installed x3 120mm cooler, two on the front panel, one on the back. Speed control is set manually and the fans operate at 80% at 50 °C and at 100% at 55 °C)
• CPU cooller: AMD Wraith Max (was previously installed AMD Wraith Spire)

[misterj]

buhster, I am with you.  I would suggest you see if you can get MSI to issue an RMA.  It is really to hard call between Processor and MB.  Maybe MSI can suggest something.  Please post a screenshot of Ryzen Master (RM) - simply drag-n-drop the image into your reply.  RM is the only reliable reporter of temperatures and other parameters.  Enjoy, John.

[buhster]

1. Almost immediately after the start of the OS. The browser is running.

2.launched cinebanch

3. FurMark CPU burner 10 minutes

[misterj]

buhster, looks OK to me.  Did you have any crashes running these?  The RED and YELLOW TDC and EDC, are restricting your performance while you have plenty of room for CPU temperature to rise (95C).  See here:

If you are willing to OC, then you can increase the limits on TDC and EDC.  You can do this in RM by selecting Precision Boost Overdrive and changing the limits for both TDC and EDC.  Beware of the warning about voiding your warranty.  Have you talked to MSI about an RMA?  The memory voltages are shown as zero.  This usually means that some changes have been made in the BIOS.  Are yous running any OC software other than RM or making manual changes in the BIOS?  Thanks and enjoy, John.

[buhster]

Hello again, John. I did not plan to overclock the processor (especially since I already have some incomprehensible problem). I contacted MSI and they asked me to conduct a series of tests. In particular, I ran Memtest86 (it took a little more than 3 hours and did not reveal any problems), and I also ran various tests in Windows (FurMark, Cinebanch, CPU-Z CPU stress test, for a total of about 2.5 hours, and this again did not reveal the problem).
Regarding the memory - yes, the voltage is set to 1.2 because the memory has an xmp profile at 3466 MHz 19-23-23-42-80 1.2v. But when choosing this profile, the motherboard automatically sets the memory voltage to 1.35V and nb soc to 1.16V - after that the computer does not boot. I manually set the voltage to 1.2 for memory and 1.005 for nb soc. The system starts whis 1.16 for nb soc, but it seems to me that in this case the problem manifests itself more often. I tried the value less and the minimum when the system starts up is 9.75v, however if you set this value then frequent freezing appears in the games. With any values of the voltage for nb soc, the problem is present, even when the memory is set as 2400 MHz and the motherboard sets the voltage for nb soc to 8.75V

[misterj]

buhster, where are you getting the nb soc voltage?  In RM there is a VDDCR SOC (0 also for you).  Did you change this?  The SOC (System On a Chip) does clock management and memory controller.  Right now mine is running at 0.9 Volts.  I really do not know what you are talking about.  Please post a screenshot.  I would expect memory at 3466 MHz to be set to about 1.35 Volts.  Mine runs at 1.35 Volts at 3200 MHz.  Are you running MSI's OC software?  Thanks and enjoy, John.

[buhster]

I did not use any overclocking software. As I said, I have never done any overclocking. I set the parameters for lowering the voltage in BIOS (a little later I will try to take a picture of the settings). And I set the memory voltage to 1.2V because this memory should work as per the memory manufacturer's specifications. The fact that this memory should work at 3466 MHz on 1.2 V is also written in the compatibility list for the motherboard here: https://www.msi.com/Motherboard/support/X470-GAMING-PRO#support-mem-14

The manufacturer of RAM does not guarantee its operation at 1.35v.
I also read on the MSI forum that for the NB SoC, the voltage should be up to 1.2V, so most manufacturers set it to 1.16V at 3466 MHz, but users write that it is better to set the voltage in this range from 1.0125V to 1.05V and many recommend 1.025V for memory at that frequency.

I can try to set the value that other users recommend and test it, but I need at least 8 hours to do it. As I already wrote, the problem rarely manifests itself quickly. And now I think that the problem does not appear more than 8 hours if the computer was not loaded at that time. That is, if 8 hours - only the desktop works (only those programs that the system launched from autoload work), then nothing will happen, but if, for example, I launch some heavy application (game or test) within 1-2 hours, then from 6 to 8 hours I will go to the black screen. I would be satisfied if the system could work at least 16 hours. In this case, I would not have noticed, because I never use a computer for more than 16 hours. And again, more chance to get off in a black screen under load. In Windows log (what I forgot to write), this is displayed as "The system rebooted, shutting down with errors. Possible causes of error: the system stopped responding, a critical failure occurred, or the power was suddenly turned off." There are no other errors except this one.

[misterj]

buhster, HWInfo is one of the least reliable SWs you can use and I do not know what CPU NB/SoC is.  You said nb soc was 9.75 Volts and now I assume that was a typo.  There is a SoC and sometime people report that a very little increase in Voltage will aide the memory.  We cannot really tell because all the memory and  VDDRC SoC voltages are zero in RM.  You or some SW has altered things in BIOS.  You need to remove these changes and go from there.  What is your MB vendor?  Thanks and enjoy, John.

[buhster]

All settings are set to auto. A-XMP profile is disabled. As I wrote my motherboard is MSI X470 Gaming Pro. I meant 0.975V, sorry for that.

[vaevictis77]

Buster - I tend to agree with misterj's thinking on this.

The symptoms you describe sound like one of the issues I 've encountered when my memory settings aren't right. I'm running G.Skill Trident-Z 3600mhz CL17 - and uses Samsung B-Die. I can get it to boot fine at 3400 and 3466. Over that gets unstable fast. 

While it boots and runs fine 99% of the time (2700X, water-cooled, etc) There are certain apps and scenarios that will trip it up and boom - black screen. Sometimes at boot up black screen....and changing settings down didn't fix it - to the point where I had to flash my BIOS and start again. I'm sure I could spent many, many, many more hours to get it 100% stable at 3400 but it's not worth it to me. Runnin at 3200 CL14 uses less power and I'm never going to tell the difference anyway. By the way - I've settled on 1.365V to run 32gb at 3200Mhz with tight timings. Granted, 16gb of the same quality memory would place less stress on the mem controller, and more easily run a bit 'faster' from a 32gb set.

Going by the RAM you mentioned: 346Mhz @CL19, as well as the current Amazon price, I am not surprised if its having issues at the rates speed. Keep in mind - being rated for a speed doesn't mean you should.  Ryzen gen 2 supports rated speeds of 2933mhz. 

I'd suggest lowering the speed to stock levels. You can throw more voltage at it as well - but do it slowly and carefully and know you're taking a risk which goes up the further you go out of spec  

[buhster]

Hello, vaevictis77. Thank you for your participation. The memory I am using is really cheap and cannot boast timings, however, the motherboard manufacturer claims that this memory is fully compatible with the motherboard I use. That is why I chose this memory. And yes, I know that the documented memory frequency for my processor is 2933 MHz. But the motherboard manufacturer claims that my memory should work in my configuration at a frequency of 3466 MHz.
I’ve already run 2x8GB Samsung C-Die in my system (if I’m not confusing anything - 10 characters in the memory marking indicate the type of chips, and I used M378A1K43CB2-CRC) at frequencies from 2400 MHz 17-17-17-39 (1.2V) up to 3466 MHz 18-19-19-39 (1.25V), while the problems were similar, both at the nominal frequency and during memory overclocking. I decided to use the memory modules that the motherboard manufacturer recommends. Perhaps the problem is related to the incorrect astromatic setting of parameters such as tRFC (on the old BIOS versions from MSI, this parameter did not increase automatically), I do not know. In any case, I like you and John tend to think that this is probably a problem with incorrect memory parameters. But maybe I managed to solve my problems by manually setting the NB SoC voltage in the BIOS to the recommended by users for memory at a frequency of 3466 MHz.

[taskforce131]

It is AMD software issue

https://community.amd.com/t5/drivers-software/rx-5700-possible-fix/m-p/439507 

[-sum1-]

Check if this answers your problem.

https://community.amd.com/message/2935926 

[buhster]

Finally I solved my problem. Replacing the motherboard with Asus PRIME X470-PRO. I had to reinstall the OS because the hard drive started to fail (after logging in, the head constantly parked with a characteristic sound and reinstalling all the drivers did not help). After that, everything works fine. I noticed that the voltages are much lower compared to MSI. In general, as far as I know, all modern MSI motherboards have this problem, regardless of the chipset. Probably this can be corrected manually, but I don't want to dance with a tambourine trying to find the correct value for stable operation.

Thanks to everyone who tried to help.

[gmergulhao]

Sorry to disappoint you but that is not all... I had a PRIME X470-PRO and replaced it with a MSI x570-A PRO and both have same behavior. Black screen with cursor before POST beep. High BIOS time on Windows Task manager and for me at least, worst of all: Deep sleep bug (even with latest agesa-v2 1.8...
Our main similarity: Ryzen 2700 Series.

so far I've tried:

CMOS Reset, Mainboard replacement. Windows clean install (also without any peripherals). BIOS Downgrade. single channel ram. setup outside the case to exclude short-circuit.

Asus support insists it has to do with my PSU being older than 7 years old... Never failed on my previous setup i5-5500k
A friend of mine also owns a 2700x. Deepsleep doesnt occur on his setup. Slow Boot timing is pretty much the same as mine (20sec +).

I might try Ryzen 3, otherwise, idk.... might send AMD my power bill for not being able to send the pc into sleep

My setup:

Ryzen 2700X

Asus x470-prime pro / MSI x570-A pro
Gskill 16gb (2x8GB) DDR4-2132 (1066) default/xamp1/2
Asus RTX 2060 Super
Corsair 650w 80 PLUS Bronze

[ArbitorA]

Not meaning to necro this thread, but for sure has something to do with the specific hardware.

CPU: Ryzen 7 2700 with Scythe Big Shuriken 3 (goes max ~65 C on load)

Motherboard: Asus Strix B450-I STRIX

Memory: G.Skill 2x8GB Trident Z RGB, DDR4 3200MHz, CL16, 1.35V (F4-3200C16D-16GTZR)

Graphics card: GTX 1070 Dual

PSU: Seasonic 650W FOCUS SGX 650

Been occurring for me several months multiple times. No OC, PBO disabled. However like pointed in several threads, I have Corsair mouse (iCue software) and sometimes peek my temperatures with hwinfo64, but do not generally leave them on or use icue's temperature features.

Will try however to see if these help at all.