• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Prime95 Error: E8500 Overclocking

S

SpYdeR187

Limp Gawd
Joined
Jan 8, 2007
Messages
236
Hey guys,
I'm fairly new to overclocking so bear with me here. I couldn't find the answer in the OCing guide that was posted in this forum so I decided to post. I'm currently running my E8500 at 347x9.5 stable. Every time I go over 347 FSB frequency I get a prime95 rounding error after about 3 minutes. What can I do to resolve this problem?

My current numbers are:
core voltage- 1.3(according to bios but drops to ~1.25 according to CPU-z)
Ram voltage - 2.0
NB - auto
SB - auto
RAM clocked in at 5-5-5-18

My tech knowlege is average so if responses could be kept simple that would be great. If you guys need any other numbers to help me out let me know and I will provide them. Also, can someone tell me about ram dividers in further detail? I notice that every time I change my FSB the DDR2-XXXX changes. What should I choose when it changes? The computer is in my signature.
 
First, up the ram to 2.1 volts and up the CPU to 1.32 in the bios and see if that helps.
 
Try upping the voltage by a 1-2 notches and continue testing, As Exe said, make sure your ram is running at its rated voltage unless you have it running over 1066.

Your motherboard should be able to handle those clocks without a voltage bumb, so im sure its more cpu voltage bound.

Also, stop using the primitive Prime95, Download Intelburntest and get results if your clocks are stable faster.
 
Thanks for the quick info. Is intelburntest really better than prime95? And about the ram rating...if 1066 is not an option do I just pick the one closest to 1066 that is lower? Thanks.
 
SpYdeR187 said:
Thanks for the quick info. Is intelburntest really better than prime95? And about the ram rating...if 1066 is not an option do I just pick the one closest to 1066 that is lower? Thanks.
Click to expand...

precisly. unless you KNOW for sure that the RAM can overclock to that speed using a certain setting for the voltage. since you said you were new to this, drop it to the next lower number before 1066.
 
cool thanks! According to the guide that was posted by Graysky on this forum, FTT test should be used when using Prime95. I've also been reading around and I see that some people use the blend test. What is the difference and would it be better if I used blend?(currently using FTT)
 
SpYdeR187 said:
cool thanks! According to the guide that was posted by Graysky on this forum, FTT test should be used when using Prime95. I've also been reading around and I see that some people use the blend test. What is the difference and would it be better if I used blend?(currently using FTT)
Click to expand...

use blend if youre OCing your RAM, as blend loads everything into your RAM. Small FFT is the best for CPU stress testing. I've yet to figure out what large FFT is for.
 
SpYdeR187 said:
Thanks for the quick info. Is intelburntest really better than prime95? And about the ram rating...if 1066 is not an option do I just pick the one closest to 1066 that is lower? Thanks.
Click to expand...

Its faster, its stresses more, and a few runs = 4+ hours prime stable.

As for your ram question, Yes, just up/lower the fsb to match it, you might be able to lower the cpu multi to keep your ram at its rated speed and where you want the cpu to be.

What i use to make sure my overclock is stable is just ibt and memtest.
 
I would not leave NB set to Auto. Try setting it around 1.37v or so, maybe even 1.35v. Auto has a tendency to set it too high or too low.

I also like to sue OCCT to stress stability.
 
I do 10 runs of IntelBurnTest using about 3.25GB of ram, which takes about an hour, and then run Prime95 in blend mode for a few hours to make sure the memory is stable.

One note about IntelBurnTest... depending on how much ram it's using and several other factors, the amount of stress applied to your cpu may vary quite a bit between tests. The GFlops number is a good indicator of how stressful the runs are. I suggest running a single run of IntelBurnTest 5 or 6 times to get an idea of how much your GFlops vary. I figured out for example that sometimes IntelBurnTest will only run at 16 GFlops and my cpu doesn't get very hot. Other times it will be pushing close to 20 GFlops, in which case my cpu is burning up and much more likely to fail a run. So whenever I test, I keep restarting IntelBurnTest until I get a GFlop reading of near 20, so I know IntelBurnTest is stressing my processor as much as it can.

Did anyone understand that lol? Basically IntelBurnTest doesn't apply the same amount of stress each time, so you have to watch the GFlops to make sure you're getting a good test. It's very random.
 
space_ghost said:
I would not leave NB set to Auto. Try setting it around 1.37v or so, maybe even 1.35v. Auto has a tendency to set it too high or too low.

I also like to sue OCCT to stress stability.
Click to expand...

Sorry about the dumb question but what does the NB do and how does the voltage changes on the NB affect the cpu?
 
Your memory controller is on the NB, so if you having stability issues, given it a little more voltage can help with some extra stability. The NB also controls the pci-express lanes. Try not to ever go beyond 1.5 volts, would create a lot of heat and would be hard to keep cool. I have mine running at 1.35 volts with my E8500 at 4ghz with a 1.298 CPU vcore. I am also running 4 sticks of ram, which stresses the NB even more.
 
Alright here's what happend. I set my voltages to what was suggested here in this thread:
Vcore - 1.32
Vram - 2.1
NB - 1.3(couldn't do 1.35 or 1.37 because the bios kept rounding)

So I ran prime small FFT over night and I'm going to assume that sometime during the 6 hour period the comp got a bsod and restarted because when I checked this morning I was at a new Windows load. What should be my next step? Should I just keep increasing my Vcore until the system is stable? If so, what voltage should I not pass to be safe(in bios). Also, how reliable is OCCT in checking stability? I'm thinking of using it because I heard you can just test for 1 hour.
 
You can increase the vcore, I would not go beyond 1.36. If that doesn't help, keep in mind you may have just hit the limit of chip. What FSB are trying to run?
 
im trying to get up to 3.5 ghz so i think its 368 FSB frequency? When you say 1.36 vcore do you mean in bios or what cpu-z reads? Would it still be safe if I go higher if I keep the chip cool? (maybe around ~55 load). Also, is it really possible that i hit the limit this low? I hear a lot of people have OC'd this way higher than me =(
 
1.36 in the bios, it is technically safe to go higher than that as long as your can keep it cool, but Intel's 45nm chips are more sensitive to voltage the AMD's. Too much voltage can seriously degrade the life of the chip. Keep in mind that overclocking is not guaranteed, I had a E8400 that would not stabilize past 425 fsb, and it took 1.36 volts to get it stable. Most other people would hit 4.2+ ghz with those kind of voltages. I would also try running Memtest 86+ using your current configuration to make sure your not getting ram errors that could be what is causing Prime95 to fail.
 
Ok so I think I made some progress last night. I increased my cpu volts to 1.3625 and only 1 of the cores failed overnight(after 3 hours). The other one was still running. Does this mean that I still need to bump up the voltage? Also I started using speedfan to monitor temps and I noticed 2 things. 1) the core readings are different from my core temp readings. Speedfan reads it at about 49-50c where as coretemp shows it around 43-45c. Which one should I trust? 2) what is the difference between the cpu temp and the core temp? When I woke up this morning the core temp of the 1 running core was still at ~49c but the cpu temp read 123c! Is this an error of some sort?
 
Get Realtemp, it is the most accurate for Intel processors, what stepping is this chip?
 
I had issues overclocking with my E8400, which was a C0, my E8500 is a E0.
 
you dont. it chooses you. unless you are buying from frys or microcenter where you can actually look at the box. its on the side, s-spec. or from an online store like tankguys where they will spec the stepping, but charge extra.
 
Alright guys heres an update...after failing a prime blend test in 10 minutes, i decided to let the cpu run a small FFT test. To my surprise the cpu is stable after 11 hours of the test. Does this mean that my memory is bad? I'm about to run memtest but what can I do if it turns out that the memory is the culprit?
 
SpYdeR187 said:
Alright guys heres an update...after failing a prime blend test in 10 minutes, i decided to let the cpu run a small FFT test. To my surprise the cpu is stable after 11 hours of the test. Does this mean that my memory is bad? I'm about to run memtest but what can I do if it turns out that the memory is the culprit?
Click to expand...

most likely. check your memory voltage settings, and run memtest overnight. if it spits out errors, your RAM is toast.
 
You can loosen the timings. Not sure what they are set at in bios.
 
You must log in or register to reply here.
Back
Top