Some overclocking questions

C

Corbin

Limp Gawd
Joined
Apr 5, 2008
Messages
509
last night i oced my e6750 to 3.0 ghz and it was stable for 10 and a half hours in prime95 small FFT testing. So today I bumped the chip up to 3.2ghz (8x400) and I ran prime95 for about 2 hours and then it failed reading something along the lines of HARDWARE FAILURE v5 or something. I have my vcore at 1.375v (reads 1.328 in cpu-z) and my ram is at ddr400 4-4-4-12 with a DRAM at 1.91v. What should I do to make it stable in prime95? Also when running prime should I use small FFTs, large FFTs, or blend?

also while testing the 3.2ghz oc it was 50-52C the whole time.
-specs in sig
 
bump the voltage up.. thats what it usually means.. also make sure your memory is running 1:1 timing with the FSB which im guessing it is since your doing 8x400..
 
ok thanks. ill try that. one other thing tho I was playing CoD4 with my current settings (3.2ghz) and about an hour in i got a BSOD. booted back up and i looked at my temps and my 2 cores were fine but the temp that reads "temp 2" was at 50C and the icon showed a little fire (meaning hot I'm guessing) after a bit it went down to 49 and below and showed a green check mark next to it. what reading is this or do you need more information? I was thinking it was along the lines of the NBCore (MCH) or SBCore (ICH)getting too hot. or am I wrong?

was using speedfan to measure these temps but also used core temp for just the 2 cores.
 
the fire symble is just a preset warning thing in speed fan.. you can change it to anything you want..

easiest way to find out which sensors are which.. you can either load up everest and see which ones the temps are closest to.. or load up the processor and see which ones change and how drastic.. cpu will usually climb the highest.. the NB temp will go up maybe 5-6C depending on the board.. and the system temp really wont change..

there should be 1 cpu temp sensor(motherboard sensor) 1 northbridge 1 system/southbridge.. then the 2 core readings..
 
sirmonkey1985 said:
the fire symble is just a preset warning thing in speed fan.. you can change it to anything you want..

easiest way to find out which sensors are which.. you can either load up everest and see which ones the temps are closest to.. or load up the processor and see which ones change and how drastic.. cpu will usually climb the highest.. the NB temp will go up maybe 5-6C depending on the board.. and the system temp really wont change..

there should be 1 cpu temp sensor(motherboard sensor) 1 northbridge 1 system/southbridge.. then the 2 core readings..
Click to expand...

ok cool I'll try that later tonight. sadly I have to go for a bit now :(
 
so i dled everest and hw monitor and along with speedfan I can put them together to get the names for each thing. on speedfan temp 1/2/3 equals TMPIM 1/2/3 on hw monitor but I don't know exactly which is the cpu/north/south bridge. Also the ACPI is at a steady 40C. can anyone help me here?

and along the lines of this I have my SB adn NB set to auto in the bios and I know i shouldn't do taht so where's a good place to start it at?
 
Corbin said:
so i dled everest and hw monitor and along with speedfan I can put them together to get the names for each thing. on speedfan temp 1/2/3 equals TMPIM 1/2/3 on hw monitor but I don't know exactly which is the cpu/north/south bridge. Also the ACPI is at a steady 40C. can anyone help me here?

and along the lines of this I have my SB adn NB set to auto in the bios and I know i shouldn't do taht so where's a good place to start it at?
Click to expand...

http://en.wikipedia.org/wiki/Advanced_Configuration_and_Power_Interface
ACPI cannot be at 40C, its not something that has a temp.

Only use Realtemp to measure CPU temp (if you want any form of comparison metric) and make sure TjMax is configured correctly for your CPU.
Stay below 70C and find your stable overclock.

This beta version sets TjMax correctly for the latest 45nM CPU's and previous CPU's.
http://i4memory.com/f79/realtemp-2-75-beta-8996/
 
speedfan automatically uses 40C as the ACPI.. means absolutely nothing.. shows up at 40C no matter what, every system ive used shows 40C for the ACPI in speedfan.. i usually just go in the config and uncheck it..

NB and SB your fine leaving them at auto.. motherboards rarely adjust the voltages on either of those.. so auto = default voltage

to do like i said to find the temps with speedfan is turn on prime95 and let it put the system at full load.. just watch the temps rise in speedfan and from there you can pretty much take an educated guess on what is what by watching the temps..
 
so i found out which ones were the NB and SB voltage and I left them both on auto in the bios. Now I bumped up my voltage to 1.4v to see if it wuld run stable 10+ hours in prime95.
Now I have one question. In the bios there is a setting between NB and SB and the Vcore/Dram that states "ht voltage". what is that? Should I leave it at auto? also is 1.4v too high for a 3.2GHZ oc on a e6750? I will try to lower it if it's stable but I jsut wanted to know.
 
Try to find the max FSB and max CPU clock separately, so you know which is holding you back.

I wouldn't worry about temps for now, they're not a good indicator of whether you're overclocked too much.
 
Varmint said:
Try to find the max FSB and max CPU clock separately, so you know which is holding you back.

I wouldn't worry about temps for now, they're not a good indicator of whether you're overclocked too much.
Click to expand...

Well the thing is 3.2GHZ is ALL I want to push this chip as I want it to last a while and it's already been running at stock for a year. I just wanted to know if all the voltages were alright so it would maximize my cpu's "life".
 
Corbin said:
Well the thing is 3.2GHZ is ALL I want to push this chip as I want it to last a while and it's already been running at stock for a year. I just wanted to know if all the voltages were alright so it would maximize my cpu's "life".
Click to expand...

Your CPU is spec'd up to 1.5V by Intel, so you're well within the safe range. However it could be that you need to adjust NB voltage. Auto doesn't necessarily mean stock, it could be higher than stock.

http://processorfinder.intel.com/details.aspx?sSpec=SLA9V

Try 1.4V for vcore, or try setting your NB voltage to whatever is normal, or slightly above, instead of auto.
 
ok thank you I'll try that. Now one last thing that is mostly directed to what sirmonkey1985 said earlier about the nb raising 5-6C when having the stress test to see what "Temp" it is in speedfan. In my speedfan I have temps that read Temp1, Temp2, Temp3, Core0, Core1.
now I think Temp2 is the NB but it only raises from 35C to 37/38C while prime95 is running. the Temp1 goes from 19C to 46C and Temp 3 stays at 22C. So is it safe to say that Temp2 is my NB, Temp3 is my SB, and Temp1 is my cpu?
 
There's no indirect way to read NB temps, you can use a temp probe, or there are some complicated tricks on some motherboards, but certainly not through Speedfan or any other software.
 
ok so can i basically ignore those temps then and just stay focused on the core temps? Or do they have some value that I should be concerned with?
 
Corbin said:
ok so can i basically ignore those temps then and just stay focused on the core temps? Or do they have some value that I should be concerned with?
Click to expand...

Don't overclock your MCH/NB much, say +0.1V is ok. It's very robust generally so don't worry about it. Core temps keep under 70C load, but even with low temps if your voltage is too high it'll shorten the lifespan. I'd keep it under 1.5V, since that is Intel's spec for the e6750.
 
ok cool. thx for all the help everyone who posted here. You've allowed me to get to 3.2ghz! :D
 
another quick question... When testing in prime 95 which of the 3 settings should I use (small FTT, Large FTT, Blend)? or should I test it on all 3? And how long would you test to see if it is stable? I see people testing for 24 hours but is this really necessary? Will 12 or maybe even 10 work? or should I do 24 just to make sure?
 
e6750 should do 3.4 easy. I say OC to 3.6 given that you have a decent cooler. It wouldn't be considered [H], otherwise. :D

But, if you're happy with 3.2ghz, that's cool.

Oh, to answer your question, use small fft to test CPU stability. No need to run 24 hours, unless you really want to, and 10 hours should be more than enough.
 
oh ok cool. It's been about 10 hours now..

and the reason I'm only doing 3.2 is that everytime i upgrade my computer I'm saving the old parts so I can build my mom a nice pc for free. right now all i need is an os/mobo/cpu for her and she gets a decent system for no cost (her current pc is 6 yrs old) and I would like to give her this cpu seeing that it is still alive and only have to buy a mobo/os for her.
 
One last thing that I'm not clear about. In an above post Varmint said "try setting your NB voltage to whatever is normal, or slightly above, instead of auto". the lowest my NB goes is 1.20v so should I set it to that and or put it a bit above? or is it absolutely fine on auto?
 
Corbin said:
One last thing that I'm not clear about. In an above post Varmint said "try setting your NB voltage to whatever is normal, or slightly above, instead of auto". the lowest my NB goes is 1.20v so should I set it to that and or put it a bit above? or is it absolutely fine on auto?
Click to expand...

If you are completely stable with 1.20v, there's no need to bump this voltage up.
 
ok cool. I guess I'll just update you guys on what's going on. I got it to 3.2GHZ stable for 10 hours in prime with a 1.4 vcore. Then i bumped the vcore down to 1.385 (or 1.387 im not sure which) and it was perfectly stable for 13 hours in prime. I then went to play CoD4 and about an hour in (this was single player) my comp froze. So i rebooted and turned on the temp displays for my cpu and gpu. turns out my gtx 260 was getting in the 70-72C range so i just bumped the fan up from stock (40) to 65 and it's been fine (in the 50-60 range now). Thing is I tried playing CoD two more times and it froze within 15 minutes of each. Could anyone know why this is?

ps i just reverted the cpu back to stock seeing this is my spring break and I want to get a little gaming in, but I remember all the things to change in the bios to get it back to 3.2.
 
Reduce clocks one at a time and see what allows it to run stable.
You already know how instability is caused.
Its up to you to find the cause.
 
Corbin said:
Then i bumped the vcore down to 1.385 (or 1.387 im not sure which) and it was perfectly stable for 13 hours in prime. I then went to play CoD4 and about an hour in (this was single player) my comp froze. .
Click to expand...

This has happened to me countless times. Basically the reason why I personally don't base system stability on stress tests alone.

I think you will need to up the cpu and/or NB voltages. Try upping the vcore a notch or two first and play CoD4 again. You are well under the voltage limit for e6xxx cpu's so you shouldn't have any problems if you go above 1.4v.

If you are unstable with vcore above 1.4v, try upping the NB voltage to see if that helps also.
 
hdnut said:
This has happened to me countless times. Basically the reason why I personally don't base system stability on stress tests alone.

I think you will need to up the cpu and/or NB voltages. Try upping the vcore a notch or two first and play CoD4 again. You are well under the voltage limit for e6xxx cpu's so you shouldn't have any problems if you go above 1.4v.

If you are unstable with vcore above 1.4v, try upping the NB voltage to see if that helps also.
Click to expand...

ok thx. I'll be sure to do this once I start testing again (school work has caught up with me on this spring break, so I may nto be able to test til monday).

overall thx for the help.
 
You must log in or register to reply here.
Back
Top