FX-8150: Defective/Weak Core?

[Tradermole]

When I overclock my CPU and prime95 fail. It is ALWAYS core #6 that fail, all the other cores run fine for the whole 24hrs prime95 test run. This force me to use fairly high voltage to achieve a stable overclock.

Do I have a lemon with a defective core? Or this is common for the FX series?

 

[dtrunk]

depends. what does your OC look like. Please post hardware information, temps while underload, and post screenies of CPUZ (cpu/memory/spd tab)

 

[Tradermole]

hardware/overclock info
CPU: FX 8150 (3.6ghz stk) @ 4.6Ghz 1.50v
MOBO: Asus Crosshair V Formula (bios 1301)
MEMORY: Corsair Vengeance 2x4GB DDR3 @ 1600 7-8-8-24-1t
GPU: 2xGigabyte Radeon 7970 OC (@1100mhz, stk volt)
CPU COOLER: Corsair H100 Push/pull (outflow) 64c MAX under load
CPU LLC: Medium, CPU over Current 140%
CPU/NB LLC Regular, CPU/NB Over Current: 100%
CPU Voltage Frq: 550, CPU PWN Phase: Extrem, CPU PWN Mode: Extrem
CPU Freq: 200, CPU multiplier: 23x, CPU/NB Voltage: 1.3
Ram Set @1600 in bios
Main HD: Intel 160GB SSD Sata II
HDD: 2 x 2TB WB Green SATAII 32mb cache
HDD: 1 x 1TB Seagte
CASE: Corsair 800d
PSU: Evolution 85+ 1050W
OS: Windows 7 Home x64 SP1


i need 1.5v to stable @4.6.(24hrs prime95). but i can go as low a 1.4250v and all cores will hold execpt #6 (in prime95). it is always the same core that fail

here are the screen cap of CPUZ

 

[dtrunk]

i don't know the FX processors as well, so i may not be the best to comment. Is the CPU NB running at 1.3v with 100% current (no over current?) your issue sounds like a voltage issue, imho; but i know nothing.

your ram is also running a little faster than its 800xmp settings; but i think that would cause other issues (like lock up.) when I had issues with stability in prime95, it was voltage.

 

[trents]

Usually, it is the same core that fails. Pretty typical. My 960T would always drop core 4 as a quad and when I unlocked it to a six core it would always be core 6. A little more core voltage took care of both. What are your temps like on 1.5 vcore?

Compare you current memory timings in "Memory" tab of CPU-z with the "SPD" tab which are the manufacturer's recommendations. You will see that you need to relax the tRC a little and most importantly, relax the CR from 2T to 1T. It might also help if you give your memory voltage a bump from 1.5 to 1.55 and your CPUNB a couple of bumps over stock. Even though this is not a "cores" issue per se, don't forget the memory and the CPU work together as a system.

 

[RGone]

Compare you current memory timings in "Memory" tab of CPU-z with the "SPD" tab which are the manufacturer's recommendations. You will see that you need to relax the tRC a little and most importantly, relax the CR from 2T to 1T. It might also help if you give your memory voltage a bump from 1.5 to 1.55 and your CPUNB a couple of bumps over stock. Even though this is not a "cores" issue per se, don't forget the memory and the CPU work together as a system.

^ +1 ^ FTW.

 

[joshhua5]

Quick question, how can you tell which core is holding you back?

 

[trents]

When you run Prime95 you will have a "worker pane" displayed for each core. The calculations in a pane will cease and you get an error message in that pane when a core flakes out. When only one core drops out like that (as opposed to getting a BSOD or spontaneous restart of the computer) it generally means your are pretty close to being stable. If the same worker pane drops out repeatedly, that's your weak core.

 

[Tradermole]

Trents: it is always the same core that fail, in prime95, no BSOD. my temp @1.5 is arround 57c to 63c. Individual core are in the high 40's low 50's a full load. Memory setting is as PER the BOX SPECS.

JOshhua5: even when a core faild, prime95 continue running and it complete is 24hrs run whitout error on the other cores and it's always core #6 that fail.

 

[keny]

There's always going to be a weak core that fails prime, it is just the way prime is, it finds the weakest link and that happens to be your core 6,if you're memory was stressed to its limit it would prob bsod, say you managed to get it stable and everything else was stable as you progress I would guarantee that it would fail again on the next increase, it's the norm and nothing is out if the ordinary.

 

[RGone]

There's always going to be a weak core that fails prime, it is just the way prime is,

I don't think I can agree with that statement as written. If it were true then this CPU forum section would be FULL of such posts saying that one core failed and is that my 'weak' core and the forum is n0t full of such posts.

The OP starts a thread that says one core fails in prime 95 and wants to know if that is normal of FX cpus. Well n0 it is not normal. Why is it not normal? Well he is clocked to 4.6Ghz and that is the problem. Overclocked there is one core that is n0t stable or n0t able to work reliably at 4.6Ghz. Not 4.6Ghz capable at 1.5Vcore. 1.5Vcore is a great deal to use day in and day out just to reach 4.6Ghz.

The truth is that his cpu is not fully stable on all cores at 4.6Ghz with 1.5Vcore. At 4.4Ghz it is likely all the cores would be stable. Probably even a little less Vcore would be needed. I would not have asked about 'defective' or weak core since there is no guarantee that any of the cores are able to do 4.6Ghz.

The truth is that the OP's cpu is not 4.6Ghz stable on all cores. There are even some boards with a bios that allows setting the speed of 'each' core and if his allowed such then he could clock that core 'downward' and then with that core at its' stable speed then P95 would not fail.

But the truth is he does not have a fully stable working cpu at 4.6Ghz overclock. I bet that core would never fail at 4.2Ghz which is the 'fastest' that TurboBoost would use and what AMD would have tested the cpu at in order to sell it.

PS: IF I had been writing the first post, my post would have asked the question >> My cpu is passing P95 at 4.4Ghz or so but one core fails at 4.6Ghz and is there anything I can do to help that core OR do I need to drop my max cpu speed in order that all cores are stable? There is no guarantee that any of the FX 8150s are stable at 4.6Ghz. So the cpu is neither defective nor really weak. Just not able to overclock to an arbitrary speed of 4.6Ghz.

 

[Tradermole]

to rgone: first my CPU is Prime95 stable at 4.6 1.5v. i runned numerous 24 hrs prime95 run with no problem. But the CPU is also stable @ 4.6 1.4250v WHITOUT CORE #6. and after a lot of tweak. i got a CPU temp of 57c to 63c at full load (core hover in the high 40c, lower 50c). wich is fine. my question was (and you anwser it): i got a weak core. that require to put a lot mroe voltage to get 4.6 stable (1.5) then it whould nornamly need.

And according to you, it is uncommon. so i got unlucky when i got my CPU.

 

[RGone]

I guess it comes down to what a person considers a stable overclock. Most of us consider an 8 core cpu to be stable at a particular speed if all 8 cores pass at that speed. If my 8 core has a failing core at some speed, then my cpu is not stable at that speed.

 

[Bobnova]

I've had CPUs where the whole thing falls apart under load, I've also had CPUs that had one distinctly weak core.
It's fairly common in my experience, though I wouldn't say "always".
I try not to say always anyway really, as being able to genuinely say always is almost never happens. Never being another one of those words

Outside of medium-obscure benching practices it doesn't really matter though, unless your BIOS is willing to apply per-core multipliers (I've yet to see one that does, AOD will but that is windows OCing) you're stuck with whatever the maximum speed for the worst part is.

 

[RGone]

I guess when I said individually adjustable multipliers, I was thinking in terms of some Asus boards with UEFI bioses that have TurboEVO that works with the UEFI bios to adjust individual core mutipliers.

 

[Tradermole]

Mobo is Crosshair formula V. and using the bios i can only disable TWO CORES at a time. and the Asus windows overclock utility does not allow idividual multiplier for clock speed.

Unless i am blind and did not see it.

AMD Overdrive let you lower multiplier for a pair of cores.... but does not work on the crosshair v

 

[keny]

I don't think I can agree with that statement as written. If it were true then this CPU forum section would be FULL of such posts saying that one core failed and is that my 'weak' core and the forum is n0t full of such posts.

The OP starts a thread that says one core fails in prime 95 and wants to know if that is normal of FX cpus. Well n0 it is not normal. Why is it not normal? Well he is clocked to 4.6Ghz and that is the problem. Overclocked there is one core that is n0t stable or n0t able to work reliably at 4.6Ghz. Not 4.6Ghz capable at 1.5Vcore. 1.5Vcore is a great deal to use day in and day out just to reach 4.6Ghz.

The truth is that his cpu is not fully stable on all cores at 4.6Ghz with 1.5Vcore. At 4.4Ghz it is likely all the cores would be stable. Probably even a little less Vcore would be needed. I would not have asked about 'defective' or weak core since there is no guarantee that any of the cores are able to do 4.6Ghz.

The truth is that the OP's cpu is not 4.6Ghz stable on all cores. There are even some boards with a bios that allows setting the speed of 'each' core and if his allowed such then he could clock that core 'downward' and then with that core at its' stable speed then P95 would not fail.

But the truth is he does not have a fully stable working cpu at 4.6Ghz overclock. I bet that core would never fail at 4.2Ghz which is the 'fastest' that TurboBoost would use and what AMD would have tested the cpu at in order to sell it.

PS: IF I had been writing the first post, my post would have asked the question >> My cpu is passing P95 at 4.4Ghz or so but one core fails at 4.6Ghz and is there anything I can do to help that core OR do I need to drop my max cpu speed in order that all cores are stable? There is no guarantee that any of the FX 8150s are stable at 4.6Ghz. So the cpu is neither defective nor really weak. Just not able to overclock to an arbitrary speed of 4.6Ghz.

Maybe worded wrongly but i would still say that 9 times out of ten prime would fail on the exact same core on every increase once you start to reach the limits of the chip.If you overcome the "first worker stopped" the next time you increase the clock it will be the same worker that stops again, there always seems to be one particular core that lets the chip down.As you say, the chip is not weak or defective, its just the limit of the chips ability with the op's particular hardware, maybe it would do more on different hardware but who knows.
@ OP can you not limit the percentage usage of each core in the bios, im pretty sure the crosshair would have that option and try knocking core six down a notch and see if that helps

 

[trents]

We all have two arms and tw0 legs. One will be weaker than the other, either left or right if we could measure the difference with enough precision. Same with CPUs. Just part of the inexactness of the manufacturing process or the materials that are used.

 

[Qatar]

here is my final oc on the 8150 for 24/7 use its really hard with this cpu using an h100 it just generates to much heat,,hope it helps...

 

[keny]

65c on the core would scare me to death, those load temps are way too high!