BSOD BCCode 124 constant restarts!

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710
Been getting these for a while now, really annoying because if I'm recording then I lose the entire recording. If I'm gaming, I lose everything after the last save.

I did a little reading and found out this is hardware related so...

Haven't done memtest yet.
Am overclocking, FX6300 at 4.2 GHz with Gigabyte GA78lmt-usb3 mobo. H100i GTX cooler, sits at 30 degrees max when torture testing in Prime95. Only for 5-10 minutes though.

I didn't change the RAM frequency, just the CPU ratio.

Any help?
Problem signature:
Problem Event Name: BlueScreen
OS Version: 6.1.7601.2.1.0.256.1
Locale ID: 4105

Additional information about the problem:
BCCode: 124
BCP1: 0000000000000000
BCP2: FFFFFA80075868F8
BCP3: 0000000000000000
BCP4: 0000000000000000
OS Version: 6_1_7601
Service Pack: 1_0
Product: 256_1

Files that help describe the problem:
C:\Windows\Minidump\072215-9438-01.dmp
C:\Users\Vanalk\AppData\Local\Temp\WER-14102-0.sysdata.xml
 
Solution
Ok, so increasing CPU ratio is overclocking if you raise it, which you have. There are many overclocking guides out there that can explain in great detail how it's done. Since you have an FX series CPU, it's multiplier is unlocked in both directions which makes it fairly simple to overclock.

In the most basic form overclocking is done by raising the multiplier by a small amount (1 or .5 if you have that fine grained option in your BIOS) and test for stability. Of course you also need to monitor your CPU temperature while stress testing. If you are stable and still have thermal room (ie it's not too hot while stress testing), then you can raise the multiplier and repeat. Eventually you will hit one of two barriers. Either you...

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


I don't know much about overclocking. It's what I'm trying to learn right now.

How would I up the vcore? and what is the vcore?

Here's what CPU-Z shows for what i think is related or might be important.

FX6300
Core Voltage: 1.200V - 1.212V as of typing this.
Core Speed: 4199.77 MHz
Multiplier: x21.0 (7-21)
Bus speed: 199.99 MHz
HT Link: 2199.88 MHz


Also this isn't really related to this, but if you do know the answer then perfect. Using a CHEAP motherboard, whenever I OC to 4.3GHz, I get a restart and windows says it was a Blue Screen (like this one). Happens right after I start torture testing using prime95.
 
Ok, so increasing CPU ratio is overclocking if you raise it, which you have. There are many overclocking guides out there that can explain in great detail how it's done. Since you have an FX series CPU, it's multiplier is unlocked in both directions which makes it fairly simple to overclock.

In the most basic form overclocking is done by raising the multiplier by a small amount (1 or .5 if you have that fine grained option in your BIOS) and test for stability. Of course you also need to monitor your CPU temperature while stress testing. If you are stable and still have thermal room (ie it's not too hot while stress testing), then you can raise the multiplier and repeat. Eventually you will hit one of two barriers. Either you will become unstable (what you are experiencing) or you'll hit your thermal ceiling. If you become unstable, then you need to raise your CPU voltage (Vcore) if you wish to go further. If you reach a thermal limit you shouldn't go further on the cooling solution you have. Now Vcore and thermal limits work against each other. Typically as long as you can raise the multiplier without using voltage, your temperatures won't increase by much. Once you start adding voltage though, temperatures can climb quite quickly. There is a third limit I haven't mentioned yet, because you'll usually hit one or both of the other two first. This is a Vcore limit. It's not so much a limit as a suggested maximum that overclockers have found to be the absolute maximum that you should go without damaging your CPU. Again though, most likely you will hit either a thermal limit or a stability barrier first.

Now if you intend to overclock, the most important thing I can tell you is to not attempt it on stock cooling. Get an aftermarket cooler. Overclocking on stock cooling is pointless as you will hit a thermal limit almost immediately. Then next thing is look for an overclocking guide, preferably one catering to your motherboard, or at least one very similar. If you can't find that, find one that is specific to the FX series CPU's. You may have to translate the guide to accommodate the naming convention of your BIOS settings.

BSOD's when overclocking, are indicators that you are not stable. If a overclock is borderline, you may find the system is relatively stable when it's idling or doing non-intensive tasks. However when you start to stress test it, it crashes / BSOD's / freezes. Now depending on how quick this happens after starting the stress test will give you some idea how close you are to being stable. If it crashes almost immediately or within a few minutes, then you are far from stable. However if you find it will stress test for a couple hours before it BSOD's, then you are getting pretty close to the correct settings.

I can't stress this enough, so I'm going to say it again. Your temperatures are very important to the longevity of your CPU and stability of your system. Next to too much voltage, excessive heat can cause permanent damage to your CPU. So manage your cooling. As I've said already, DO NOT OVERCLOCK on stock cooling.
 
Solution

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


I've got a H100i GTX water cooler that I bought specifically for this reason. Luckily I got a dual rad and not single rad.

Stock cooler used to be 45 degrees idle. Now it's 9 degrees idle. I'm using prime95 to stress test right now and I'm only getting a max of 49 degrees. It just sits at 40 degrees, spikes to 48 degrees, drops down to 38 degrees and repeat.
I'm guessing TjMax is the highest temp it should be at, which it shows 80 degrees. Though I read that AMD said 65 degrees max.

So I upped the vCore and it's been good so far. 4.3 GHz is stable so far, everything looks good. Temps are fine, torture testing with temps as idle with stock cooler. I'm going to try and hopefully without damaging anything, overclock a tad more.
Oh, my BIOS said that over voltage can cause permanent damage. And right now the 1.35 V is light-red. 1.45 and above are red, so is this something I should worry about? Red doesn't look like a happy colour.
 
Ok, well I just did some quick checking on Bulldozer / Piledriver and most of the guides are saying that 1.40 - 1.45V is safe on air. I came across one that said 1.55V, I assume this means the absolute maximum. I'd say from that, you should be safe up to 1.45V as long as your temperatures are OK.

There is something off about your temperatures. You idle temp is 9C which I assume is below your ambient temperature unless your computer is in a walk in cooler. With your cooling, you should be idling a few degrees above ambient. So I'd say anywhere from 25 to 30 degrees if your ambient temperature is 22C. What are you using to monitor your temperatures?

If you are using Prime95, Small FFT's are going to heat and stress your CPU the most. If you are trying to quickly dial in on a target frequency (assuming you have one in mind), you can use short stability tests of a couple hours. However once you've hit your target frequency, you'll want to leave it at least 8-12 hours. Even longer if you want to be sure, some say 24 hours or more.

So before you get too far, I think it's advisable to figure out what your actual temperatures are.
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


I use CoreTemps to look at temperatures. But when I looked in the BIOS this morning, it showed 30-35 degrees.
It was a little chilly that day, I just quickly checked the weather app and say I think 22 degrees. Read online that moisture starts to form when temps go below ambient? Got the window and door open for airflow.

Also people say that voltage and heat will kill your CPU faster. Since temperatures aren't a problem for me, would raising voltage to 1.45 make my CPU literally die and not work faster? And also CPU-Z shows the voltage anywhere between I think 1.22 to 1.464 V. Switches between those 2 when torture testing. Also I use the top one, Small FTT for maximum heat.
How do I find out the voltage? Cause it's past 1.45 according to CPU-Z even though in the BIOS it shows 1.375 V. So confused here.

What would you recommend to check temperatures?
 
I use HWInfo to monitor all my fan speeds, voltages, temperatures and a few other things. Here is a screenshot of my desktop. I use a Gadget that ties into HWInfo to display all of the information I am interested:

Desktop%20Screenshot.jpg


It takes a little time to configure, but I find for me, it's worth it. You can configure pretty much however you want. If you like graphs over time, you can do that. If you only like numbered values, you can do that. You can configure the graphs to draw the traces in different colors to indicate different things.

As for your temperatures, I'm pretty certain it won't idle at 9C. That would mean your room temperature would have to be like 5C or less.

Yes voltage will kill a CPU faster than heat. That is unless you run it without a heatsink for extended periods. However running high temperatures over long periods of time will degrade a CPU as well, it just takes longer. Going too high with voltage can cause insta-death though.

If you are concerned about your Vcore, I would try another monitoring program. If you like the idea that I've used, then HWInfo. If you want something simpler, then HWMonitor (made by the same company), or Speccy. This will either confirm what you are seeing in CPU-Z or given you reason to question it. However I wouldn't be to worried about it going to 1.464V periodically. Though a jump of almost 90mV from what you have it set to is weird. It is normal for Vdroop to occur when stress testing. In an idle condition the voltage will be higher (if Cool n' Quiet are disable and using manual Vcore), and then droop under load. Motherboards that cater to overclocking will have a Line Load Calibration that will eliminate or minimize the amount of Vdroop.

Can you take a photo of your BIOS setup on the voltage control page? Post it here.
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


IMG_1110.jpg
[/url][/img]

Two things I can't seem to control at the bottom, cut off. Both have the word Vcore at the end of it.

Anything higher on the voltage and it goes red. Right now it's pink.

Edit: Dang that is small, here's the link: http://postimg.org/image/6k0nlyab1/
 
The last two items are for display only, don't worry about these.

The red indication is probably to make it obvious that you do so at your own risk, which anytime you overclock you are doing so.

Under Advanced BIOS Features you have the Load Line Control (Line Load Calibration on ASUS motherboards) this should help with Vdroop. What are the options under this setting? The manual doesn't show what they are.

I've also included a link to a forum thread at overclocker.net which directly relates to what you are doing:

Overclocker.net

 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


Okay, I'll take a picture of the advanced features.
Are my settings good? Hopefully nothing is wrong. Are there any suggestions on what I should change?
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


Okay so the LLC options are Auto, Regular and Extreme.
When I set it to Regular, I also overclocked to 4.8 GHz but wouldn't boot so I dropped back down to 4.5GHz but it would crash whenever I ran P95. So I set LLC back to Auto and overclocked it to 4.6GHz and it seems to be fine.

Also my vCore is still 1.355V in the BIOS so why does HWinfo show voltage as 1.467 or 1.464? I've also noticed that the frequency drops to 2.5GHz here and there so maybe that's why the voltage is going crazy.
 
I would see set LLC to either Extreme. This should keep Vcore stable. Then have a look at Vcore in HWInfo. If you are crashing in Prime 95, then you need more Vcore. Likely the reason it stabilized after putting it back to Auto is because it's allowing Vcore to go higher than your set point.

As for your frequency dropping, that is a function of Cool 'n Quiet (SpeedStep for Intel). To lower power usage during idle, the OS will drop the multiplier and Vcore to save power and lower temps. However this feature can make your overclock unstable. Lot's of overclocking guides tell you to turn these features off in the BIOS.
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


Anytime I set LLC to extreme, I always crash after using P95. Yes I set the voltage higher. I think I did something stupid, I did 4.8GHz and it couldn't boot so the computer just reset the ratio back to 3.5GHz. Now my computer will always crash when starting P95. *Instantly set it back to 4.6GHz. Nothing else changed. All my settings are correct, but now even if I back down to 4.5GHz it will crash. I can't do anything if my computer will crash anytime it does something intense.

I looked at HWiNFO64 and the motherboard section shows Vcore as 1.322V. But if I scroll up to CPU, it shows "Core #0 VID" as 1.175V. Is my voltage too low or is that normal? I'm so confused, I think I broke my computer.
 
If you've come to this point, I'd say reset the BIOS to default via the jumper. The process is detailed in your motherboard's manual.

Once everything is back to default, go back to Prime 95 and make sure you are stable at default. Hopefully your instability is some residual settings or something getting stuck in the BIOS. Once you have verified your stable in default, then slow go back to your last "stable" settings. Overclocking is supposed to be done in slow deliberate steps. Jumping too quick can cause this (best case scenario), or worse, cause damage to your CPU, motherboard or both.

As for VID, that will be static if you are using manual to control your voltage and will very closely match what you've set in the BIOS. If the voltage is set to Auto, it's what the motherboard is setting the voltage at. Vcore is the actual measured value of what is being supplied to the CPU. Depending on the accuracy of the circuit measuring it, the accuracy of the software monitoring it, the accuracy of the VRM setting the voltage, and the present load placed on the CPU, these values will vary slightly from VID. For instance I've set my CPU voltage to manual and 1.115V in the BIOS, HWInfo shows VID @ 1.116V (very close to what I've set in the BIOS) and my Vcore varies depending on the load slightly, right now at 100% load (WCG) is 1.136V on three cores, the last is 1.120V. Your offset between the two seems quite high, but as I mentioned, there are a few factors that can affect the measured value of Vcore.
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710


I just clicked on "Restore Optimized Defaults" and it booted up fine without crashing when torture testing.

So I'm back at 4.6 GHz stable. I read somewhere about the CPU Frequency set to 210MHz so I set it to 210MHz (default 200) and it showed 4.83GHz. I also remembered I could tinker with the (I think it's called a FSB) so I just did 10MHz up.
No luck, it either couldn't boot or it just crashed when torture testing.

All the settings are the same as in the pic I posted, except for the 200MHz changing to 210MHz, making the CPU ratio show 4830 MHz. I mean I'm not sure why, but I just feel like hitting 4.8 or 5GHz. Yes for 24/7 even though I know it's not going to happen and I'll probably break my computer.

But basically do you think there's anyway I can go higher than 4.6GHz for 24/7 or is this the max I should go? 4.7GHz didn't seem to work (only changing ratio).

Also how long do you think my computer will last? (I think that sometimes people mean how long before my hardware will be too old) Like before the CPU or motherboard will literally die and not work at all. Which one do you think will die first?
 

Hrlyrdr22

Reputable
Apr 18, 2015
188
0
4,710
Also I remembered you said 1.45V is good on air, so does that mean with a water cooler I can up the voltage even more for a safe 24/7? Using a H100i GTX Right now got AF120 quiet edition fans on there but the fans that came with are like double the RPM.
 
As far as going higher, it's possible as long as you don't have to raise Vcore much more. I'm not sure how accurate your Vcore readings are, but if they are to be believed then you are at the limit of where I'd go. You might be able to go to 1.5V as long as your load temps are OK. Remember though, voltage is going to degrade the processor much faster than heat will.

As to how long your CPU will last under those conditions 24/7, that is really hard to say. Some CPU's will handle voltage better than others. Usually you know you have a good one if your temperatures are good with higher voltages. That said, we are talking about billions of transistors built on a process that is measured in billionths of a meter (nm). Electromigration is the factor here and the more voltage, the more current flows through the circuit paths and transistors. The more the electrons flowing through these circuit paths, the more it erodes them. Much like how high waters in rivers and streams wear out the banks particularly at bends in the stream. Eventually these paths will wear through, breaking completely or becoming intermittent. So now that the semiconductor lesson is over, it's really hard to say, it'll depend on you CPU and how much continuous load you put on it while it's overclocked. I'd venture a guess that if you don't get too crazy it'll last until at least the point that you'll want to upgrade.

I'll also mention with respect to voltage, these are loose guidelines given in overclocking forums. It's based on trial and error. The error being broken CPU's. Therefore there is no hard science involved with respect to coming up with these limits.

As to raising your reference clock, you might not want to play with this too much. Overclocking it will overclock your HyperTransport bus. It's rated to 2600MT/s (HT version 3), there is some tolerance for overclocking it, but not much. If you continue to push the reference clock, then you may have to lower the HyperTransport multiplier to keep it down. There has been no evidence that overclocking the HyperTransport gains any performance benefits. So if you end up pushing your base clock, lowering the multiplier for the HyperTransport so that it's less than 2600MT/s isn't going to hurt performance either. Beyond that, overclocking the reference clock is also going to overclock your RAM, which may lead to instability. So to combat this, you may have to lower your RAM speed to keep it stable. If I were in your position, I'd just stick to raising the core clock multiplier and leave the base clock alone. That is really the whole reason to have an FX processor, easy overclocking via the core clock multiplier.