• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FX 4100 failing on 4.4+ mhz OC

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Infinite66

Member
Joined
Jun 27, 2012
I managed to overclock my fx 4100 to 4.3, and am looking to push to 5 as the final goal. The max voltage on my cpu is 1.550 and at 4.3, i am barely pushing 1.3. I adjust the voltage to compensate for the higher overclock (anything 4.4 and up), but computer still fails to boot. Any suggestions? :bang head
 
Welcome to OC Forums!

First of all, it is very unlikely you will get to 5 ghz on air cooling. We have had a rash of new members recently that have some pretty wild expectations about overclocking and I simply don't know where they are getting these numbers. With that cooler and motherboard you will be doing good to get 4.5 ghz.

Let me be clear that when we talk about overclocking around here we mean a stable overclock that will pass at least a two hour Prime95 stress test. We aren't just talking about stable as being able to boot into Windows.

So let me suggest backing up to 4.3 and running the Prime95 blend test and see if you can pass two hours without blue screen, lock up, spontaneous restart or having one of the Prime core workers stop while the others keep going. While you're at it, download and install HWMonitor and have it open before and during the Prime95 stress test to monitor temps and voltages. When the stress test is done, post back to us with an attached pic of the HWMonitor interface. To attach a pic, crop and save it with Snipping Tool in Windows Accessories or something equivalent. Then click on Go Advanced at the bottom of any new post screen. When the advanced window appears, click on the little paperclip icon at the top. The rest will be obvious.

While you're at it, download and install CPU-z. It's as great system reporting tool that gives lot's of information about bios settings.

When running the Prime95 stress test, you want to keep the "core" temps from exceeding 60c and the CPU (socket) temp from exceeding 70c. Stop the test if either occurs.
 
Last edited:
Thanks for the help. I already have core temp and prime 95 for testing, and the cpu ran stable at 21-24 degrees under full load of prime 95's blend. I only ran for about half an hour, but have stress tested for 8 hours straight on this speed with no hiccup in the past. So 5 is not attainable, stable at least huh? Hmmm might just stick with what you said 4.5 ish.
 
Can you please post a screenshot like this with idle (Min.) temperature showing in HWMonitor before starting Prime95, then take the screenshot after running prime for at least 15 minutes?

Open HWMonitor at idle, then CPU-Z windows, then prime and run prime for 15 minutes then take the screenshot.

primeexample.jpg
 
It is usually a smart idea to disable all the green, power-saving, down-throttling technology that can cause instability in the higher overclock ranges.

In bios, we advise disabling: Cool N Quiet, Turbo, C1E, C6 and APM.

In Windows Control Panel Power Options we recommend configuring it to High Performance.

Sounds like your core temps are nice n' cool at this point but keep in mind that the FX CPUs core temp readings are usually skewed on on the cool side, often by about 15c.

Keep raising your multiplier by .5x increments until you cannot pass a 20 minute Prime95 blend test and then add a .025 increment of vcore. Of course, the relationship of overclock increase to voltage increase is far from linear. The last .1 or .2 ghz increments of speed gain usually take a lot more to achieve than the first .7 ghz increase. That's when you begin to "hit the wall". After finding where that wall is, then run longer Prime stress tests and make adjustments to top her off.
 
It is usually a smart idea to disable all the green, power-saving, down-throttling technology that can cause instability in the higher overclock ranges.

In bios, we advise disabling: Cool N Quiet, Turbo, C1E, C6 and APM.

In Windows Control Panel Power Options we recommend configuring it to High Performance.

Sounds like your core temps are nice n' cool at this point but keep in mind that the FX CPUs core temp readings are usually skewed on on the cool side, often by about 15c.

Keep raising your multiplier by .5x increments until you cannot pass a 20 minute Prime95 blend test and then add a .025 increment of vcore. Of course, the relationship of overclock increase to voltage increase is far from linear. The last .1 or .2 ghz increments of speed gain usually take a lot more to achieve than the first .7 ghz increase. That's when you begin to "hit the wall". After finding where that wall is, then run longer Prime stress tests and make adjustments to top her off.
Well he already has 4.3 stable...4.4 isn't working so lets find out what voltages, temperatures, etc he is running first. 20c is not even close to real.

The reason for disabling power saving technologies isn't because of instability, but rather...causing stability.

C&Q disables automatically over a certain multiplier, I have not tested on FX but Phenom II this is 18.5, which is why AMD had to stop at 3.7 GHz X4s (980) without new stepping. Enable/disable C1E + C6 along with C&Q.
Turbo should be disabled because of voltage hike over already high voltages in OC situations, and frequency hike in single thread or lightly threaded workloads, where if pushing limits already will cause instability / damage
APM is new with FX CPUs, it will throttle the CPU when a predefined TDP limit within the CPU is reached. Prime95, heavily multithreaded programs tend to set this off. This is not the same as VRM/PWM Over-Current Protection which is a feature of motherboards, independent from whatever CPU is installed.

High Performance, I recommend too but usually the default Balanced works fine. Windows 7 works in tandem with universally defined features like C1E, C6, C&Q so if you forget one place, it should kick in in the other.

Thanks for the help. I already have core temp and prime 95 for testing, and the cpu ran stable at 21-24 degrees under full load of prime 95's blend. I only ran for about half an hour, but have stress tested for 8 hours straight on this speed with no hiccup in the past. So 5 is not attainable, stable at least huh? Hmmm might just stick with what you said 4.5 ish.
Your room is probably at least as warm as 21c so this number is just absolutely wrong. :shrug:

I'd ask you to just list stuff out like "what idle and load temp, what idle and load CPU voltage, what CPU-NB clock, what CPU-NB voltage etc etc etc"

but most of that should be supplied with the screenshot. :thup:
 
Last edited:
Hello everyone. Sorry for my lack of response. I am not at my computer right now so I cannot post images. I wasn't using hwmonitor. I had it confused with core temp program. Each of my cores didn't reach past 25 degrees celcius during prime 95's stress test. This is with a fresh appliance of IC diamond thermal paste. I can list stuff though as far as voltage etc. Is core temp not accurate?
 
No, core temp is not accurate. We need to look also at the "CPU" (socket) temp in this case to get a more realistic picture of what your temps are.
 
No, core temp is not accurate. We need to look also at the "CPU" (socket) temp in this case to get a more realistic picture of what your temps are.

Ok I will make sure to uninstall core temp then. Was wondering why my temps were WAY below my first overclock. Used HWmonitor for my first time, did a fresh install of windows and forgot to get it again.When I get the chance to do everything, Ill make sure to post all the pics/ info
 
Hello everyone. Sorry for my lack of response. I am not at my computer right now so I cannot post images. I wasn't using hwmonitor. I had it confused with core temp program. Each of my cores didn't reach past 25 degrees celcius during prime 95's stress test. This is with a fresh appliance of IC diamond thermal paste. I can list stuff though as far as voltage etc. Is core temp not accurate?
You were around 45 degrees then.

Rule of thumb is to add 20c. Some guys don't believe me or get rather conservative, or believe the motherboard sensor is always accurate which doesn't make sense since it's just a different sensor, and only ASUS/Gigabyte boards have been tested for accuracy, ASUS being relatively close and Gigabyte reading low a good 5-10c.

Who knows about your Extreme3.
 
Back