Sign in with
Sign up | Sign in
Your question

Failing prime 95, with no overclock on my 8350

Tags:
  • Overclocking
  • Components
  • Graphics
Last response: in CPUs
Share
18 August 2014 00:58:28

I have several workers stop on my 8350 whenever I run the blend test. My voltage for my 8350 in my gigabyte 990fxa ud3 is at 1.4 and my 8gb 1600MHz vengeance ram voltage is a correct too (I also ran memtest and that went fine). I am no overheating, my CPU is at the stock voltage, and I am using the latest version of prime 95. I have an HX PSU from corsair if you guys wanted to know that.

More about : failing prime overclock 8350

a b K Overclocking
18 August 2014 01:07:24

the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx
m
0
l
18 August 2014 01:09:28

ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?
m
0
l
Related resources
a b K Overclocking
18 August 2014 13:53:51

It could be a bad memory module, or less commonly, a bad motherboard slot.

Run Blend while testing memory modules one at a time. Shut down your rig and switch the power supply off. If no switch, then unplug it. Remove all but one memory module, then power up and run Blend. Repeat the procedure for each module in a given slot. 4 modules and 4 slots = 16 individual tests. Be sure to keep track of which modules have been tested in which slots.

An alternative is to run Memtest. Google the latest version, download and run. Use the same procedure as described above. You can also try Windows Memory Diagnostic, which is built into Windows.

If your memory runs OK on these tests, then at least you know what the problem isn't.
m
0
l
a b K Overclocking
18 August 2014 14:04:30

^+1

happen to me in the past with my RAM going bad
m
0
l
a b K Overclocking
18 August 2014 16:55:28

bbq108 said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?


Intel Burn Test - AVX (you might need to become a forum member to download this)

And IBT-AVX is accepted as an alternate stressing program for FX cpus by overclockers far and wide. If prime can run on your system that's fine, but for many piledriver owners it cannot.
m
0
l
a b K Overclocking
18 August 2014 17:57:35

P95 v26.6 is pre-AVX / FMA code, so if 26.6 produces the same problem as 28.5, then perhaps you should consider looking elsewhere to troubleshoot the problem, such as memory.
m
0
l
a b K Overclocking
18 August 2014 20:35:20

anyway i think mod should move this thread to CPU section
m
0
l
a b K Overclocking
18 August 2014 21:22:43

Done!
m
0
l
24 August 2014 19:02:07

CompuTronix said:
It could be a bad memory module, or less commonly, a bad motherboard slot.

Run Blend while testing memory modules one at a time. Shut down your rig and switch the power supply off. If no switch, then unplug it. Remove all but one memory module, then power up and run Blend. Repeat the procedure for each module in a given slot. 4 modules and 4 slots = 16 individual tests. Be sure to keep track of which modules have been tested in which slots.

An alternative is to run Memtest. Google the latest version, download and run. Use the same procedure as described above. You can also try Windows Memory Diagnostic, which is built into Windows.

If your memory runs OK on these tests, then at least you know what the problem isn't.


I ran memtest before and nothing was wrong with my RAM.
m
0
l
25 August 2014 02:02:11

By default settings I've seen evidence to suggest that memtest may not produce enough stress on the memory system while testing to reveal many problems. Make sure to run it in multi-threaded mode to flood the memory controller with tests at a very high saturation rate for both the memory controller and memory bus/chips etc.
m
0
l
30 August 2014 04:53:33

ingtar33 said:
bbq108 said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?


Intel Burn Test - AVX (you might need to become a forum member to download this)

And IBT-AVX is accepted as an alternate stressing program for FX cpus by overclockers far and wide. If prime can run on your system that's fine, but for many piledriver owners it cannot.


I ran that for a while and my system is unstable.
m
0
l
a b K Overclocking
30 August 2014 07:02:48

bbq108 said:
ingtar33 said:
bbq108 said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?


Intel Burn Test - AVX (you might need to become a forum member to download this)

And IBT-AVX is accepted as an alternate stressing program for FX cpus by overclockers far and wide. If prime can run on your system that's fine, but for many piledriver owners it cannot.


I ran that for a while and my system is unstable.


ok. so lets work this one out.

correct me if i'm wrong at any point.

1) fx8350
2) stock settings
3) gigabyte ga990fxa-ud3 motherboard
4) failing Prime95 & IBT-AVX
5) your stock vcore is 1.4v?

m
0
l
a c 425 K Overclocking
30 August 2014 07:14:50

Try raising the DRAM voltage slightly + 0.05 or so over spec
m
0
l
a b K Overclocking
30 August 2014 07:26:12

Tradesman1 said:
Try raising the DRAM voltage slightly + 0.05 or so over spec


That was going to be my first suggestion too... though i wanted to confirm first what his system was set up like.
m
0
l
31 August 2014 06:15:10

ingtar33 said:
bbq108 said:
ingtar33 said:
bbq108 said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?


Intel Burn Test - AVX (you might need to become a forum member to download this)

And IBT-AVX is accepted as an alternate stressing program for FX cpus by overclockers far and wide. If prime can run on your system that's fine, but for many piledriver owners it cannot.


I ran that for a while and my system is unstable.


ok. so lets work this one out.

correct me if i'm wrong at any point.

1) fx8350
2) stock settings
3) gigabyte ga990fxa-ud3 motherboard
4) failing Prime95 & IBT-AVX
5) your stock vcore is 1.4v?



The stock vcore is 1.404v (I just checked)

For some reason, I looked at it before, and I could have sworn it said 1.45
m
0
l
a b K Overclocking
31 August 2014 18:42:22

bbq108 said:
ingtar33 said:
bbq108 said:
ingtar33 said:
bbq108 said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


So it is normal to fail prime 95 in less than 10 seconds?

And could you post a link to where to download the intel burn test?


Intel Burn Test - AVX (you might need to become a forum member to download this)

And IBT-AVX is accepted as an alternate stressing program for FX cpus by overclockers far and wide. If prime can run on your system that's fine, but for many piledriver owners it cannot.


I ran that for a while and my system is unstable.


ok. so lets work this one out.

correct me if i'm wrong at any point.

1) fx8350
2) stock settings
3) gigabyte ga990fxa-ud3 motherboard
4) failing Prime95 & IBT-AVX
5) your stock vcore is 1.4v?



The stock vcore is 1.404v (I just checked)

For some reason, I looked at it before, and I could have sworn it said 1.45


thats a high vcore... but i'll roll with it. try what tradesman1 suggested... bump your dram voltage in your bios by +0.05V
m
0
l
31 August 2014 20:29:48

Before I do that, is 1.5v normal for stock DRAM voltage?
m
0
l
a b K Overclocking
31 August 2014 20:36:54

bbq108 said:
Before I do that, is 1.5v normal for stock DRAM voltage?


that's pretty standard.
m
0
l
a b K Overclocking
31 August 2014 23:00:16

I'm going to place bets on it being a heat/throttling issue.
1.4v is remarkably high for stock, not sure why it would be set to that other than it being a really really bad bin from the factory, which doesn't make a whole lot of sense either considering they would probably throw that into the 8320 section or scrap it completely.

From what I remember of stressing my CPU at stock before my initial overclock, mine did actually bounce up to 1.4v when it was using the Turbo frequency of 4.2GHz across some cores. A great over estimation from AMD considering I'm now chugging 4.5GHz across ALL cores at a load voltage of 1.38v. (Load Line Calibration setting).

My recommendation is to hop in to the BIOS and disable TurboCore temporarily and re-run the tests.
Also watch your temps, if it's the stock cooler it's remarkably bad at cooling - While running a stress test, open HWMonitor and watch the FX-8350 temp value (Not the CPU value, that measures socket temp).
Don't let it go over 62c.

Finally, what revision number of the UD3 is your board? I seem to recall Gigabyte cheaped out on some power delivery bits on the earlier revisions, shouldn't hurt at stock, but it may be doing so.
m
0
l
1 September 2014 03:14:51

Distello said:
I'm going to place bets on it being a heat/throttling issue.
1.4v is remarkably high for stock, not sure why it would be set to that other than it being a really really bad bin from the factory, which doesn't make a whole lot of sense either considering they would probably throw that into the 8320 section or scrap it completely.

From what I remember of stressing my CPU at stock before my initial overclock, mine did actually bounce up to 1.4v when it was using the Turbo frequency of 4.2GHz across some cores. A great over estimation from AMD considering I'm now chugging 4.5GHz across ALL cores at a load voltage of 1.38v. (Load Line Calibration setting).

My recommendation is to hop in to the BIOS and disable TurboCore temporarily and re-run the tests.
Also watch your temps, if it's the stock cooler it's remarkably bad at cooling - While running a stress test, open HWMonitor and watch the FX-8350 temp value (Not the CPU value, that measures socket temp).
Don't let it go over 62c.

Finally, what revision number of the UD3 is your board? I seem to recall Gigabyte cheaped out on some power delivery bits on the earlier revisions, shouldn't hurt at stock, but it may be doing so.


My temperatures are perfectly fine though. I have an H105 with Noctua NF-F-12 fans and Arctic Silver 5 thermal paste.

I have a rev 3.0 board.
m
0
l
a b K Overclocking
1 September 2014 15:08:13

bbq108 said:
Distello said:
I'm going to place bets on it being a heat/throttling issue.
1.4v is remarkably high for stock, not sure why it would be set to that other than it being a really really bad bin from the factory, which doesn't make a whole lot of sense either considering they would probably throw that into the 8320 section or scrap it completely.

From what I remember of stressing my CPU at stock before my initial overclock, mine did actually bounce up to 1.4v when it was using the Turbo frequency of 4.2GHz across some cores. A great over estimation from AMD considering I'm now chugging 4.5GHz across ALL cores at a load voltage of 1.38v. (Load Line Calibration setting).

My recommendation is to hop in to the BIOS and disable TurboCore temporarily and re-run the tests.
Also watch your temps, if it's the stock cooler it's remarkably bad at cooling - While running a stress test, open HWMonitor and watch the FX-8350 temp value (Not the CPU value, that measures socket temp).
Don't let it go over 62c.

Finally, what revision number of the UD3 is your board? I seem to recall Gigabyte cheaped out on some power delivery bits on the earlier revisions, shouldn't hurt at stock, but it may be doing so.


My temperatures are perfectly fine though. I have an H105 with Noctua NF-F-12 fans and Arctic Silver 5 thermal paste.

I have a rev 3.0 board.


There's more to throttling than just CPU temps, if the VRMs get too hot then they're going to adjust the power being delivered to compensate. Dropped power = dropped frequency = dropped performance.

Disable TurboCore and try it again.
m
0
l
a b K Overclocking
1 September 2014 15:21:13

ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


This.

m
0
l
2 September 2014 01:56:35

RobCrezz said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


This.



I ran the burn test and still failed.
m
0
l
a b K Overclocking
2 September 2014 09:07:40

bbq108 said:
RobCrezz said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


This.



I ran the burn test and still failed.


System crashes? Or just gets the unstability warning?
m
0
l
15 September 2014 20:59:42

RobCrezz said:
bbq108 said:
RobCrezz said:
ingtar33 said:
the whole bulldozer architecture doesn't like prime95... some chips just will never pass a prime burn on it. my own 8320 won't pass one over 4.5ghz no matter what i do. That's when i uncovered this issue with a little research. If you want to stress test that cpu, try intel burn test -avx


This.



I ran the burn test and still failed.


System crashes? Or just gets the unstability warning?


sorry for the late reply, I have been really busy lately. I think that windows movie maker and rust crashing and locking up may have something to do with this, but generally it is just instability warnings.
m
0
l
15 September 2014 22:04:05

Distello said:
bbq108 said:
Distello said:
I'm going to place bets on it being a heat/throttling issue.
1.4v is remarkably high for stock, not sure why it would be set to that other than it being a really really bad bin from the factory, which doesn't make a whole lot of sense either considering they would probably throw that into the 8320 section or scrap it completely.

From what I remember of stressing my CPU at stock before my initial overclock, mine did actually bounce up to 1.4v when it was using the Turbo frequency of 4.2GHz across some cores. A great over estimation from AMD considering I'm now chugging 4.5GHz across ALL cores at a load voltage of 1.38v. (Load Line Calibration setting).

My recommendation is to hop in to the BIOS and disable TurboCore temporarily and re-run the tests.
Also watch your temps, if it's the stock cooler it's remarkably bad at cooling - While running a stress test, open HWMonitor and watch the FX-8350 temp value (Not the CPU value, that measures socket temp).
Don't let it go over 62c.

Finally, what revision number of the UD3 is your board? I seem to recall Gigabyte cheaped out on some power delivery bits on the earlier revisions, shouldn't hurt at stock, but it may be doing so.


My temperatures are perfectly fine though. I have an H105 with Noctua NF-F-12 fans and Arctic Silver 5 thermal paste.

I have a rev 3.0 board.


There's more to throttling than just CPU temps, if the VRMs get too hot then they're going to adjust the power being delivered to compensate. Dropped power = dropped frequency = dropped performance.

Disable TurboCore and try it again.


I havent gotten to disabling the turbo core yet, but I believe the VRM's are overheating as when running intel burn test, my CPU will bounce from 3.4-3.6GHz and randomly jump to 4GHz for a second . It keeps going 3.4 to 4 to 3.4 to 4. I read on amazon about someone having the same issue but on a revision 4 board (has better VRM cooling).
m
0
l