Athlon II X4 OC problem

[LoneNinja]

I've had an Athlon II X4 630 overclocked to 3.5Ghz since January and am now facing stability problems. For nearly half a year it has worked flawlessly and proven to be Prime95 stable for around 5hrs, and hasn't caused a single crash. Originally it was at 1.45V, and I recently began having games crash so I bumped it up to 1.475V and the problem went away. I just installed Windows 7 Pro 64bit( replacing Vista Business 64) a few days ago, and it has booted into windows just fine at 3.5Ghz and installed the OS without a problem. Now if I try to run prime95 at 1.45V it crashes instantly, and at 1.5V it consistently crashes within 5 minutes. I'm looking for possible reasons that my overclock is no longer stable as I'm clueless. It might be worth mentioning that I defaulted by BIOS settings and now Win 7 won't load, it blue screened trying to start and is now in a start up repair because it refuses to start due to hard ware changes.

System Info
Windows 7 Pro 64bit
Athlon II X4 630
Gigabyte 780G AM2+ (I can get the exact model if needed, gonna need to look through receipts if windows doesn't load)
Gskill DDR2 1066 2x2Gb
Seagate 500Gb
Sapphire Radeon 4670
Antec Earthwatts 380W
Some random DVD drive

My bios settings were.
CPU Clock Ratio 14X
CPU Northbridge Freq 8X
CPU Frequency 250
Memory Clock X4 800mhz(1000mhz after overclock)
DDR2 Voltage Control 1.9V(default was 1.8V)
CPU NB Vid Control +.05(not sure what that even is)
CPU Voltage +.05

Thanks in advance to any suggestions I receive, right now I'm just trying to get windows to boot at default again.

 

[DrMrLordX]

Honestly, it sounds like one of two things happened:

either your chip degraded, or your ambient temperatures rose. Possibly both.

Have you been able to get the machine running at stock settings with some extra CPU voltage?

edit: also, what are you using to cool the x4 630?

 

[busydude]

Try running your memory at default voltage and frequency. The other thing I can think of is regarding ambient temps. Did you monitor temps of your CPU?

 

[LoneNinja]

I got my computer up and running stock settings again, some how my sata settings were preventing windows from loading. My motherboard is the GA-MA78GM-S2HP if anyone needs to know. I've got a Cooler Master Hyper TX3 on it, and I use hardware monitor on a regular basis actually so I know the core temp has never exceeded 55C and was usually 45-49C. I also moved recently, my ambient temps right now because I'm in a basement is only 68F, which is actually lower than the apartment I was living in by a good 10F.

Something that might be worth noting is my BIOS shows 1.4V as the default Vcore, and if I leave it at that and load into windows CPUZ is reporting 1.488V. However when I add additional voltage it reads properly within windows.

I doubt the cpu degraded, although it's been on almost 24/7 I'm a user of K10stat and the majority of the time it was on it was actually underclocked undervolted to 2.0Ghz 1.2V, if anything degraded it's got to be my board, I've had an X2 5000BE and X2 7750BE both on here overclocked prior to this Athlon II 630.

 

[cusideabelincoln]

Like already mentioned, maybe the summer heat is getting to your chip or your board.


Seems like your option is to just re-overclock the thing, or just leave it at stock.

 

[DrMrLordX]

Given the cooling then, I'm guessing it's your ambients. You need better cooling or a less ambitious overclock. Which is . . . just what cusideabelincoln said. Well, mostly.

 

[ShawnD1]

Check if the mobo is degraded. Put all the settings to what you were running before but set the muliplier for the cpu a lot lower.

Run prime for 4096kb min and max size, run in place. If the board is broken, the computer will crash immediately

 

[heyheybooboo]

I've got a Gigabyte 785g-u2sh and IIRC with the vcore on Auto it tends to over-volt.

Did you pull the battery when you reset?

Have you tried to manually set the voltage to 1.325v?





--

 

[ShawnD1]

I've got a Gigabyte 785g-u2sh and IIRC with the vcore on Auto it tends to over-volt.

Did you pull the battery when you reset?

Have you tried to manually set the voltage to 1.325v?

From his first post, it sounds like he has been manually setting the voltage this whole time.

One thing to consider is your CPU-northbridge voltage. I've seen a few overclocking articles say that this one can really put a limit on your overclocking. My own experience is that my gigabyte motherboards wouldn't overclock for shit if I left the CPU-NB voltage at "normal" (1.175 stock?). When set to 1.25V, I could overclock my Athlon II X4 620 by 20% and at the same time the CPU is undervolted by 0.05V and it's actually stable like that.

I think it's only gigabyte motherboards that have that problem because of the way the voltages are controlled. On my gigabyte boards, the voltatages are all-or-nothing auto control. If I want to manually set the CPU voltage, I need to manually set every other voltage, including the northbridge. My Asus boards don't have this problem because each voltage has its own option for being "auto". As a result, my Phenom II X2 and X6 systems (Asus boards) don't need the CPU-NB screwed with because "auto" automatically scales them with the bus speed, and I can play around with just the CPU voltage.

edit
I'm not sure if those CPU-NB voltages I posted are correct. Do not base your voltages on anyone else's voltages!

 

[Smoove910]

Does your ram require 1.65v? What is it set at?

 

[LoneNinja]

Does your ram require 1.65v? What is it set at?

2.0V actually, it's a ddr2 kit.

I've got a Gigabyte 785g-u2sh and IIRC with the vcore on Auto it tends to over-volt.

Did you pull the battery when you reset?

Have you tried to manually set the voltage to 1.325v?

My board actually just has normal(1.4V) or I can do +/- by increments of .025, can't actually set it to a select voltage.

Right now I've got all my original settings in place except I dropped the cpu multiplier down to 13x so the cpu is running at 3.25Ghz, it's been stable prime95 for about 15min so far, obviously I'm going to let it run for a few hours. At least it didn't crash in the first few minutes, maybe my cpu just won't do 3.5 stable any more for some reason. Makes me tempted to get a Phenom II X6, lol

 

[LoneNinja]

Failed stability testing 3.25Ghz @ 1.475Vcore, that is just plain terrible. I think I'm gonna throw it in my other computer and see how it fairs there.

 

[busydude]

are you sure that your memory is not a cause for your probs? Just trying to help you isolate the issue.

 

[ShawnD1]

Failed stability testing 3.25Ghz @ 1.475Vcore, that is just plain terrible. I think I'm gonna throw it in my other computer and see how it fairs there.

3.25ghz / 14 = 232 bus?

Drop the CPU multiplier to 5x. While the CPU is underclocked, run Prime95 using 4096k in-place. If it still fails, you'll know that it's the motherboard that is failing, which it probably is.

One thing to try is to google for overclocking your specific motherboard. There are lots of little gimicky things that might be specific to your board. For my motherboard, someone figured out that going past 270mhz bus requires a 0.1V boost to memory voltage. That's something not many of us would figure out on our own.

 

[LoneNinja]

3.25ghz / 14 = 232 bus?

Drop the CPU multiplier to 5x. While the CPU is underclocked, run Prime95 using 4096k in-place. If it still fails, you'll know that it's the motherboard that is failing, which it probably is.

One thing to try is to google for overclocking your specific motherboard. There are lots of little gimicky things that might be specific to your board. For my motherboard, someone figured out that going past 270mhz bus requires a 0.1V boost to memory voltage. That's something not many of us would figure out on our own.

Actually did 13 x 250 = 3.25Ghz. It was only 1 core that failed prime, about 1 1/2 hours in to it.

Would an individual core crash because of memory? Sounds to me like a cpu problem, but I'm overly accustomed to overclocking with BE chips, this is my first non BE overclock since a socket 754 Sempron.

 

[evolucion8]

Actually did 13 x 250 = 3.25Ghz. It was only 1 core that failed prime, about 1 1/2 hours in to it.

Would an individual core crash because of memory? Sounds to me like a cpu problem, but I'm overly accustomed to overclocking with BE chips, this is my first non BE overclock since a socket 754 Sempron.

Well, it could be a possibility, may be the sofware pointed certain core to a faulty memory address and that may cause the issue, but I think that in the small FFT tests the RAM is barely used, use MemTest x86 just in case, but I think that if its your CPU, you will probably need better coolling. My GF has an Athlon II X4 635 and even though it idles at 30C, when running at full power, you can feel the heat coming out of the Antec 900 case and that's with a stock cooler and a HD 3870 with stock cooler which dumps air outside of the case.

 

[DrMrLordX]

Actually did 13 x 250 = 3.25Ghz. It was only 1 core that failed prime, about 1 1/2 hours in to it.

Would an individual core crash because of memory? Sounds to me like a cpu problem, but I'm overly accustomed to overclocking with BE chips, this is my first non BE overclock since a socket 754 Sempron.

Quick question: when Prime95 fails, does it take the entire system down with it (lock up/reboot)? In my experience with my x4 635, if Prime95 fails gracefully, it's the NB or memory that are unstable.

If it BSODs or just resets, it's unstable cores.

Because on my quad, when it was NB/memory, Thread 1/Core 1 would always crap out on me. The others would just keep on truckin.

 

[LoneNinja]

Quick question: when Prime95 fails, does it take the entire system down with it (lock up/reboot)? In my experience with my x4 635, if Prime95 fails gracefully, it's the NB or memory that are unstable.

If it BSODs or just resets, it's unstable cores.

Because on my quad, when it was NB/memory, Thread 1/Core 1 would always crap out on me. The others would just keep on truckin.

At 3.5Ghz whole screen just artifacted and computer rebooted, I downed the cpu multiplier to 10X and left everything else the same and Prime95 began to fail individual cores without crashing windows.

I'm actually at nearly 2hrs of stability with 3.5Ghz @ 1.45V right now, my memory is still at 500mhz too. I increased Northbridge Voltage by .1 and something called CPU NB by .1, and it appears to have made it stable. I'm not sure why though since I have my northbridge multiplier lowered so it's still running at a flat 2.0Ghz.

 

[DrMrLordX]

Interesting. I've never seen Prime95 cause screen artifacting. That almost implies problems with the video card or PCI-e controller. Almost.

After reading your post, I think we can safely say that you had NB instability, at the very least. CPU-NB voltage is the voltage to the on-die northbridge (NB) on a K10.5 chip. Long story short, it is my experience that increased heat in the cores causes the NB to want more voltage to stay stable on a propus chip (my Sargas behaves the same way). It sort of tells me that your chip wants more cooling, though I'm not 100% convinced that NB stability was your only problem.

It is a bit dismaying that you needed extra CPU-NB for stability when your NB speed is only at 2 ghz.

Did you have any problems applying the TIM? Or maybe it's just time to get a better HSF.

 

[richierich1212]

Are you using the IGP by any chance?

 

[ShawnD1]

Interesting. I've never seen Prime95 cause screen artifacting.

I've seen this. I was testing how low I could put the voltage on my Athlon X4 620 (undervolting because it's summer and I don't want the heat). Prime95 still ran great with no errors, but the screen got all fucked up with a bunch of horizontal lines. All I was lowering was the CPU voltage. I did not touch any other setting; stock speed, stock memory volts, stock northbridge, stock everything. Yes it has integrated video.

 

[LoneNinja]

Turns out it's still not fully stable. It won't crash during game play or anything else I use this PC for anymore, it even passes AMD overdrive stability testing, but Prime95 still eventually crashes it.

For those who keep commenting it's a possible heat issue, I don't see how that's possible, it's 53C right now full load.

I'm stressing again, I bumped cpu vcore up to 1.475, Northbridge to 1.3V, and CPU NB is +.125. I'm positive it's a cpu problem and not a board/memory as it runs fine if I drop the cpu multiplier to 10X giving me a 2.5Ghz clock speed.

 

[richierich1212]

I'm still doubting it's your CPU, since you've only had it overclocked like that for half a year.

 

[LoneNinja]

Strange, but I've got it stable now for close to 5 hours in Prime, still going. The combination of dropping my ram multiplier down to 3.33 and having this jacked up Northbridge voltage seems to have done it. Really is a shame since it's stable at lower clock speeds with the faster ram speed. I really wish AMD would just release an Athlon II X4 BE, lol

 

[richierich1212]

still priming?