• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

The system reboots itself automatically when Prime95 is run

How long have you had your EVO212 on the cpu? If it's over 2 years maybe it's time for some new thermal paste. The other thing as other people have said is memory clocks and timings. Time to put everything to stock and see if it's stable.
 
my guess is the PSU. Do you have the specifications for it? (usually in table format where you can see the different ratings: Voltages, Amperes, Watts, etc)

I also have some questions, Are you gaming with it for years already and just tried prime95 now to find it crash? if yes, then, I guess it's fine for gaming but not for stress tests. If it's not causing you problems at day to day stuff, even gaming, and have lasted this long already, I guess it's kind of fine. or it could just be coincidence that your rig is failing as well. also, any newly installed components?
 
77 is still warm, so the amb during day is higher
 
I'd guess the PSU, find another one to test it with.

If not that it's likely the memory or worst case the CPU is bad.

But more than likely I'd guess a better PSU would fix the issue, assuming you're running everything stock I'd also ignore all the people telling you to tweak things mobo related. If you're running everything stock than tweaking mobo/memory/cpu settings will at best just mask the issue.
 
How long have you had your EVO212 on the cpu? If it's over 2 years maybe it's time for some new thermal paste. The other thing as other people have said is memory clocks and timings. Time to put everything to stock and see if it's stable.
I upgraded my PC around 7~8 months ago. I bought the 4790K and EVO212 in January and the PSU a year earlier.
my guess is the PSU. Do you have the specifications for it? (usually in table format where you can see the different ratings: Voltages, Amperes, Watts, etc)
I also have some questions, Are you gaming with it for years already and just tried prime95 now to find it crash? if yes, then, I guess it's fine for gaming but not for stress tests. If it's not causing you problems at day to day stuff, even gaming, and have lasted this long already, I guess it's kind of fine. or it could just be coincidence that your rig is failing as well. also, any newly installed components?
I had several restarts while gaming as well, although I don't play games, I member I had reboots when my cousin was playing Battlefield 1 and also GTAV on my PC around the same time I bought the parts
 
Are you using 2T command rate for that 20GB of ram? (2 x (2+8)?) Try running with just 2x8
 
Last edited:
to check the cpu for stability nip into the bios and lower the multiply from its stock 40 down to 36.. if it works at 36 it will tell you your cpu is crapping out..

if that dosnt work its down to swapping out parts.. it could be the mobo.. the ram or the psu.. swapping out parts is sometimes the only way to find out the faulty part..

my 4790K died on me.. the mobo was also dead.. i assume one part died and took out the other part.. which part died first i havnt a clue.. thats the only reason i am now running a 7700K.. :)

trog
 
77 is still warm, so the amb during day is higher
I just rechecked and it seems I made a mistake, its actually 19 C now. When I ran the first test at 2 PM it was 25 C.
I'd guess the PSU, find another one to test it with.
If not that it's likely the memory or worst case the CPU is bad.
But more than likely I'd guess a better PSU would fix the issue, assuming you're running everything stock I'd also ignore all the people telling you to tweak things mobo related. If you're running everything stock than tweaking mobo/memory/cpu settings will at best just mask the issue.
I had previously run memtest and it found no issues. I checked all ram sticks, first I tried one 8Gig stick, then I added another 8 Gig stick. then I added another 2gig and finally the last 2 Gig stick and all went just fine without any issues.
After 12 hours of running prime95, nothing happened!
These are the last temps.
Code:
  DATE      TIME      MHz    CPU_0 CPU_1 CPU_2 CPU_3  LOAD%  GPU
09/28/17  09:01:50  3990.76    52    55    58    55   100.0   40
09/28/17  09:01:51  3990.76    51    55    56    53   100.0   40
09/28/17  09:01:52  3990.76    50    56    59    55   100.0   40
09/28/17  09:01:53  3990.76    54    55    58    53   100.0   40
09/28/17  09:01:54  3990.76    56    58    57    53   100.0   40
09/28/17  09:01:55  3990.76    54    57    57    55   100.0   40
09/28/17  09:01:56  3990.76    54    55    60    53   100.0   40
09/28/17  09:01:57  3990.76    55    58    61    51   100.0   40
09/28/17  09:01:58  3990.76    53    56    61    57   100.0   40
09/28/17  09:01:59  3990.76    56    56    59    56   100.0   40
09/28/17  09:02:00  3990.76    55    58    57    54   100.0   40

I'm going to run small fft now as suggested before,
I, however, noticed the whole time, CPU frequency was locked at 3990.6, I remember when a CPU-heavy process was to run, it would overclock itself so that it could go up to at least 4.20. This did not happen and I wonder why!
 

Attachments

- Doubt this is a PSU problem (P95 is by far not the highest power draw your system can have, and it is MUCH lower than 700w. Even 300w 12V would be sufficient, even with a serious OC on the CPU). Its also not old. On the other hand, the PSU may have been DOA, you just didn't realize it yet, which would explain the reboots you had earlier while gaming. Its worth swapping this part and run with it for a time, even if just for your own confidence in the system.

When you next upgrade something, sell that PSU anyway, brand screams unreliable.

- Motherboard is suspect. Shitty LLC setting, too low vCore to let the CPU boost reliably on all cores. Note that Turbo clocks are reduced when all cores are loaded, but not back to 4.0 I believe (4.2 is your all-core turbo? Can check in BIOS). Its also an ASUS board and they are not problem free in the software department. Like others said, remove everything that isn't the Chipset drivers or otherwise required. If it has an ASUS sticker on it and can be removed do so.

- Memory: unlikely. I'd expect BSODs.

Worth doing: run OCCT and dive into the power graphs you get after running a stress test. Especially the LLC one.
 
Last edited:
- Doubt this is a PSU problem (P95 is by far not the highest power draw your system can have, and it is MUCH lower than 700w. Even 300w 12V would be sufficient, even with a serious OC on the CPU). Its also not old. On the other hand, the PSU may have been DOA, you just didn't realize it yet, which would explain the reboots you had earlier while gaming. Its worth swapping this part and run with it for a time, even if just for your own confidence in the system.

When you next upgrade something, sell that PSU anyway, brand screams unreliable.

- Motherboard is suspect. Shitty LLC setting, too low vCore to let the CPU boost reliably on all cores. Note that Turbo clocks are reduced when all cores are loaded, but not back to 4.0 I believe (4.2 is your all-core turbo? Can check in BIOS). Its also an ASUS board and they are not problem free in the software department. Like others said, remove everything that isn't the Chipset drivers or otherwise required. If it has an ASUS sticker on it and can be removed do so.

- Memory: unlikely. I'd expect BSODs.

Worth doing: run OCCT and dive into the power graphs you get after running a stress test. Especially the LLC one.
The problem with PSU is that my choices here are very limited, either I can buy this brand or Cooler Master MasterWatt Lite 700W which has been made available the only couple of months ago.
and unfortunately, I don't have any spare part so that I can swap them. That's why I'm asking for help so that I can narrow it down to a specific part and then either buy a new one or send it back to for warranty/guarantee whichever applies.
By the way after 4 hours of prime 95 with small fft, nothing happened, but I had several interesting observations :
First of all here are the starting and ending entries for CPU temperature when I started the small fft :
Code:
  DATE      TIME      MHz    CPU_0 CPU_1 CPU_2 CPU_3  LOAD%  GPU
09/28/17  09:10:15  4260.13    36    33    37    33     4.8   39
09/28/17  09:10:16  4250.15    36    34    37    33     4.6   39
09/28/17  09:10:17  4270.11    36    33    36    34     3.8   39
09/28/17  09:10:18  4000.73    61    65    67    63    73.3   39
09/28/17  09:10:19  3990.75    66    68    70    66   100.0   39
09/28/17  09:10:20  3990.75    65    69    72    67   100.0   39
09/28/17  09:10:21  3990.75    65    68    71    66   100.0   39
09/28/17  09:10:22  3990.75    67    70    74    68   100.0   39
09/28/17  09:10:23  3990.75    67    69    74    67   100.0   39
09/28/17  09:10:24  3990.75    68    69    72    67   100.0   39
09/28/17  09:10:25  3990.75    68    70    71    68   100.0   39
09/28/17  09:10:26  3990.75    68    72    73    68   100.0   39
09/28/17  09:10:27  3990.75    61    67    68    62   100.0   39
09/28/17  09:10:28  3990.75    69    72    75    69   100.0   39
09/28/17  09:10:29  3990.75    68    71    74    68   100.0   39

...

09/28/17  13:01:45  3990.75    71    73    76    71   100.0   38
09/28/17  13:01:46  3990.75    71    72    74    71   100.0   38
09/28/17  13:01:47  3990.75    72    72    75    70   100.0   38
09/28/17  13:01:48  3990.75    72    73    75    70   100.0   38
09/28/17  13:01:49  3990.75    66    66    74    70   100.0   38
09/28/17  13:01:50  3990.75    71    72    75    71   100.0   38
09/28/17  13:01:51  3990.75    71    72    75    70   100.0   38
09/28/17  13:01:52  3990.76    72    72    75    71   100.0   38
09/28/17  13:01:53  3990.75    71    73    75    71   100.0   38
09/28/17  13:01:54  3990.75    71    73    75    71   100.0   38
09/28/17  13:01:55  3990.76    71    73    75    72   100.0   38
09/28/17  13:01:56  3990.76    69    73    76    71   100.0   38
09/28/17  13:01:57  3990.75    72    73    74    70   100.0   38
09/28/17  13:01:58  3990.76    72    73    75    71   100.0   38
09/28/17  13:01:59  3990.75    71    73    72    71   100.0   38
09/28/17  13:02:00  3990.75    71    72    75    71   100.0   38
The ambient temperature is the same, it seems small fft makes the cpu hotter!
I also noticed the vcore voltage being around 1.09v while at the moment where nothing is running, it fluctuates between these 3 numbers : 1.280 to 1.182 to 1.231
cpu-zvcore.JPG
cpu-zvcore2.JPG
cpu-zvcore3.JPG


Could this be the cause?
I downloaded IntelBurnInTest, do you think it would be better to run some tests with it instead of prime95 and see the results (before starting OCCT tests)
 
The problem with PSU is that my choices here are very limited, either I can buy this brand or Cooler Master MasterWatt Lite 700W which has been made available the only couple of months ago.
and unfortunately, I don't have any spare part so that I can swap them. That's why I'm asking for help so that I can narrow it down to a specific part and then either buy a new one or send it back to for warranty/guarantee whichever applies.
By the way after 4 hours of prime 95 with small fft, nothing happened, but I had several interesting observations :
First of all here are the starting and ending entries for CPU temperature when I started the small fft :
Code:
  DATE      TIME      MHz    CPU_0 CPU_1 CPU_2 CPU_3  LOAD%  GPU
09/28/17  09:10:15  4260.13    36    33    37    33     4.8   39
09/28/17  09:10:16  4250.15    36    34    37    33     4.6   39
09/28/17  09:10:17  4270.11    36    33    36    34     3.8   39
09/28/17  09:10:18  4000.73    61    65    67    63    73.3   39
09/28/17  09:10:19  3990.75    66    68    70    66   100.0   39
09/28/17  09:10:20  3990.75    65    69    72    67   100.0   39
09/28/17  09:10:21  3990.75    65    68    71    66   100.0   39
09/28/17  09:10:22  3990.75    67    70    74    68   100.0   39
09/28/17  09:10:23  3990.75    67    69    74    67   100.0   39
09/28/17  09:10:24  3990.75    68    69    72    67   100.0   39
09/28/17  09:10:25  3990.75    68    70    71    68   100.0   39
09/28/17  09:10:26  3990.75    68    72    73    68   100.0   39
09/28/17  09:10:27  3990.75    61    67    68    62   100.0   39
09/28/17  09:10:28  3990.75    69    72    75    69   100.0   39
09/28/17  09:10:29  3990.75    68    71    74    68   100.0   39

...

09/28/17  13:01:45  3990.75    71    73    76    71   100.0   38
09/28/17  13:01:46  3990.75    71    72    74    71   100.0   38
09/28/17  13:01:47  3990.75    72    72    75    70   100.0   38
09/28/17  13:01:48  3990.75    72    73    75    70   100.0   38
09/28/17  13:01:49  3990.75    66    66    74    70   100.0   38
09/28/17  13:01:50  3990.75    71    72    75    71   100.0   38
09/28/17  13:01:51  3990.75    71    72    75    70   100.0   38
09/28/17  13:01:52  3990.76    72    72    75    71   100.0   38
09/28/17  13:01:53  3990.75    71    73    75    71   100.0   38
09/28/17  13:01:54  3990.75    71    73    75    71   100.0   38
09/28/17  13:01:55  3990.76    71    73    75    72   100.0   38
09/28/17  13:01:56  3990.76    69    73    76    71   100.0   38
09/28/17  13:01:57  3990.75    72    73    74    70   100.0   38
09/28/17  13:01:58  3990.76    72    73    75    71   100.0   38
09/28/17  13:01:59  3990.75    71    73    72    71   100.0   38
09/28/17  13:02:00  3990.75    71    72    75    71   100.0   38
The ambient temperature is the same, it seems small fft makes the cpu hotter!
I also noticed the vcore voltage being around 1.09v while at the moment where nothing is running, it fluctuates between these 3 numbers : 1.280 to 1.182 to 1.231
View attachment 92539 View attachment 92540 View attachment 92541

Could this be the cause?
I downloaded IntelBurnInTest, do you think it would be better to run some tests with it instead of prime95 and see the results (before starting OCCT tests)

The vCore values look normal, and yes, they can change under load but realistically, are supposed to go up under load and drop down to something around 0.9v at idle - this is part of what Load Line Calibration manages as well (LLC I was referring to earlier). And if the motherboard does that too aggressively, you may or may not briefly drop under the required voltage at some points, this is a behavior called vDroop. This can be fixed by going into motherboard settings, find LLC, and play with the different settings; more often than not, it needs to be a bit tighter (ie a more aggressive LLC). This usually is something only needed for overclocking, so it being required at stock can point at either some motherboard weirdness, or PSU stability. Either way its a setting you can tweak that may mitigate the problems in hardware.

You can also see the P95 run: starts with core at 4.3 Ghz and drops down to 4.0 as the load increases. This is normal behaviour though.

1.28v is rather on the high end by the way, which I'm guessing is the voltage you see when you run the heaviest (AVX) P95 runs. It is however within spec and not dangerous to have, given sufficient cooling. A basic tower like the 212 you have should keep that at or around 80C, which is fine (not a temp that will cause crashing per sé).

In the end, beyond software/BIOS tweaks, if you don't have spare parts, you'll find yourself grasping at straws to really find the problem. Swapping parts and excluding them is really the only way to be sure what the issue is here. In your case you just need more time: use different tweaks, and see if the problem happens again. Be smart and write down somewhere what you have done and in what combination/situation so you don't waste time doing the same stuff.

Bottom line though, spontaneous reboots point at brief moments of power interruption or voltage drops, if they are hardware related (because what really happens is: your system loses power, and the automatic reboot kicks in as power is restored). I really doubt this is a memory problem.
 
small fft will make the cpu hotter and raise the cpu core voltage.. that is normal..

its cooks my overclocked intel chips (100 C) i dont use it.. its not a good stress test.. stop using it..

the intel burn in test is better.. its made by intel to test intel chips..

just for the record my 4790K chip was stable at 4600mz on a 1.28 vcore.. i ran it for couple of years like this.. at default it was way lower.. similar to your 1.82.. but not running p95..

i use the furmark cpu burner as a stabilty test.. i just leave it running in the background while i browse.. if it dosnt crash after a few hours i consider the system stable enough for me..

i still dont think its your psu the system is pulling about 150 watts at the most when the 1080 gpu isnt running..

trog
 
small fft will make the cpu hotter and raise the cpu core voltage.. that is normal..

its cooks my overclocked intel chips (100 C) i dont use it.. its not a good stress test.. stop using it..

the intel burn in test is better.. its made by intel to test intel chips..

just for the record my 4790K chip was stable at 4600mz on a 1.28 vcore.. i ran it for couple of years like this.. at default it was way lower.. similar to your 1.82.. but not running p95..

i use the furmark cpu burner as a stabilty test.. i just leave it running in the background while i browse.. if it dosnt crash after a few hours i consider the system stable enough for me..

i still dont think its your psu the system is pulling about 150 watts at the most when the 1080 gpu isnt running..

trog

Ehh hate to break it, but if you cannot reliably run any P95 test without throttling, you need to dial down the clocks or the voltage. Tiptoeing around the thermal limits does not a stable OC make...

Its fine for your purposes perhaps but don't bring this as gospel...
 
While running Inter Burn In Test at Maximum preset, I faced a BSOD , Driver_IRQL_NOT_LESS_OR_EQUAL .
the standard, and high were completed successfully, however, on the Maximum preset, after 5 or maybe 6 tries, it gave me a BSOD (I have never gotten a BSOD before! this the first time!)
Here are the logs :
Code:
----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
Clock Speed: 3.99 GHz
Active Physical Cores: 8
Total System Memory: 20417 MB

Stress Level: Standard (1024 MB)
Testing started on 9/28/2017 1:23:04 PM
Time (s)        Speed (GFlops)        Result
[13:23:19] 9.003    99.2893            3.156799e-002
[13:23:34] 8.984    99.4968            3.156799e-002
[13:23:48] 8.937    100.0260            3.156799e-002
[13:24:04] 9.056    98.7101            3.156799e-002
[13:24:19] 9.475    94.3417            3.156799e-002
[13:24:33] 8.970    99.6527            3.156799e-002
[13:24:48] 9.157    97.6175            3.156799e-002
[13:25:03] 8.918    100.2347            3.156799e-002
[13:25:17] 8.928    100.1196            3.156799e-002
[13:25:32] 8.988    99.4585            3.156799e-002
Testing ended on 9/28/2017 1:25:32 PM
Test Result: Success.
----------------------------
Code:
----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
Clock Speed: 3.99 GHz
Active Physical Cores: 8
Total System Memory: 20417 MB

Stress Level: High (2048 MB)
Testing started on 9/28/2017 1:26:21 PM
Time (s)        Speed (GFlops)        Result
[13:26:58] 25.427    103.9251            4.753050e-002
[13:27:35] 25.670    102.9446            4.753050e-002
[13:28:12] 25.334    104.3080            4.753050e-002
[13:28:49] 25.431    103.9086            4.753050e-002
[13:29:27] 25.737    102.6767            4.753050e-002
[13:30:04] 25.330    104.3234            4.753050e-002
[13:30:40] 25.263    104.6005            4.753050e-002
[13:31:18] 25.914    101.9724            4.753050e-002
[13:31:55] 25.373    104.1474            4.753050e-002
[13:32:32] 25.312    104.3999            4.753050e-002
Testing ended on 9/28/2017 1:32:32 PM
Test Result: Success.
----------------------------
And this is the Maximum preset which failed :
Code:
----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
Clock Speed: 3.99 GHz
Active Physical Cores: 8
Total System Memory: 20417 MB

Stress Level: Maximum (10064 MB)
Testing started on 9/28/2017 1:32:43 PM
Time (s)        Speed (GFlops)        Result
[13:38:26] 281.138    108.4470            3.376867e-002
[13:44:01] 275.762    110.5612            3.376867e-002
[13:49:41] 280.290    108.7753            3.376867e-002
[13:55:18] 276.878    110.1156            3.376867e-002
[14:00:54] 276.799    110.1470            3.376867e-002
[14:06:30] 277.204    109.9861            3.376867e-002
[14:12:05] 276.073    110.4369            3.376867e-002

I went to the BIOS and loaded the default optimized settings (default settings) and booted up into Windows.
While in the BIOS, I noticed there were several options for GDI+ VRM?! such as CPU CUrrent Capability, CPU Power Phase Control and CPU LLC which all were set as Auto. What does Auto mean in this context?
does it mean it automatically changes ? or its a simple term which actually denotes the default value?
I mean for LLC there are 4 values:
Regular or 0%,
Medium or 25%
High or 50%
Ultra hight 75%
Extreme 100%
I mean the Turbo mode and SpeedStep are both enabled, don't I need to set LLC to at least medium or high when Turbo mode is enabled?
update :
second try was successful :
Code:
----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
Clock Speed: 4.03 GHz
Active Physical Cores: 8
Total System Memory: 20417 MB

Stress Level: Maximum (9846 MB)
Testing started on 9/28/2017 2:53:14 PM
Time (s)        Speed (GFlops)        Result
[14:58:39] 264.965    111.2956            3.412462e-002
[15:04:01] 263.188    112.0472            3.412462e-002
[15:09:28] 269.898    109.2613            3.412462e-002
[15:14:48] 262.241    112.4516            3.412462e-002
[15:20:06] 261.015    112.9797            3.412462e-002
[15:25:25] 261.716    112.6772            3.412462e-002
[15:30:44] 262.068    112.5261            3.412462e-002
[15:36:05] 264.215    111.6117            3.412462e-002
[15:41:24] 261.871    112.6107            3.412462e-002
[15:46:42] 260.689    113.1212            3.412462e-002
Testing ended on 9/28/2017 3:46:43 PM
Test Result: Success.
----------------------------
I'll try to run the Maximum preset once again and if succeeded I'll try OCCT and post the results back here so hopefully, I can get something out of it.
 
Last edited:
what i am getting out of it.. is even at normal default clocks your 4790K chip is no longer stable..

vayra.. tip toeing around thermal limits is par for the course with later generation intel cpus.. :)

trog
 
Last edited:
Fair enough.

Make me a favor, stop using Prime95 and switch to AIDA64 run a stress CPU test and let's see if it crash, if it doesn't crash run one again AIDA64 stress test but this time Stress FPU.

View attachment 92542
OK, I'm running the test now, when they are done, I'll report back.
by the way how many hours should I let the test continue? is 1 hour enough for each test?
Thank you by the way
 
Last edited:
OK, I'm running the test now, when they are done, I'll report back.
by the way how many hours should I let the test continue? is 1 hour enough for each test?
Thank you by the way

Anytime.

After how many minutes/hours Prime95 crashed?
 
Anytime.

After how many minutes/hours Prime95 crashed?
in my first attempt, it caused the system to reboot just after couple of minutes. However subsequent tests didn't create any issues even after 12 hours of constant running of P95.
So far there has not been any issues with the CPU stress
 
Your cooler ambient temps may have something to do with that... it was like 6C difference?
 
Your cooler ambient temps may have something to do with that... it was like 6C difference?
Is it not normal for that to be different? after all there are other parts which emit heat as well.

By the way I stopped the CPU stress after 2 hours and now start the FPU test. I'll report back in 2 hours.
 
Last edited:
Back
Top