E8400 OC Question, Could It Work? [Dell Optiplex 755]

Mr.Scott said:

My E8400.

Click to expand...

holy...



ps.. but why not 5ghz? (just kidding )

Tech Tweaker said:

My mistake, that is 800MHz memory

Click to expand...

cool ok. i dont know why i didnt post the memory stick picture earlier (especially since i have gone quite overboard with the pics )

kitt said:

holy... View attachment 160438

View attachment 160439

ps.. but why not 5ghz? (just kidding )

Click to expand...

Couldn't make it on just the chiller.
Not a particularly good chip. Just average.

Mr.Scott said:

Couldn't make it on just the chiller.
Not a particularly good chip. Just average.

Click to expand...

Was that max speed or benchable Scotty?

Johan45 said:

Was that max speed or benchable Scotty?

Click to expand...

Benchable at 4860 on most benches.
Heater benches like CineBench at just under 4.8 Gig

hey again i was wondering if anyone can tell me, regarding what has been said about whether a PC PSU can cope with the extra voltages when you are overclocking.. is there a way to know if your psu is coping ok (eg. a monitoring program etc) or is it not easy to know if your psu is under strain/becoming faulty/damaged etc

kitt said:

hey again i was wondering if anyone can tell me, regarding what has been said about whether a PC PSU can cope with the extra voltages when you are overclocking.. is there a way to know if your psu is coping ok (eg. a monitoring program etc) or is it not easy to know if your psu is under strain/becoming faulty/damaged etc

Click to expand...

You'll need a digital multimeter if you want to see if your power supply is delivering the proper voltages on the rails.

http://forums.extremeoverclocking.com/showthread.php?t=137886

Black wire = Ground
Yellow wire = 12 volts
Red wire = 5 volts
Orange wire = 3.3 volts

Other wires you don't really need to worry about for the purposes of testing a power supply, so long as it is connected to a motherboard in a working PC/computer system.

For testing, touch the black (Ground) tester probe to a black wire on the power supply (preferably on a 4-pin molex connector), and touch the other probe (which is typically red) to the wire of the rail you want to test. For the 3.3v/Orange wire you'll usually have to look on the 20/24-pin motherboard connector, as that rail isn't used much in other connectors these days. And you'll want to have the multimeter set to the Volts test setting.

great thanks for that i do have and use a multimeter so will not have any trouble performing some tests.
hey i also stumbled across a nice (i think) fall back plan incase the LGA775 pad modding does not work out..
i noticed this cpu below:

Core 2 Duo E7600 SLGTD (R0, with VT) 3.07 GHz 3 MB 1066 MT/s 11.5× 0.85–1.3625 V 65 W LGA 775 May 2009

this E7600 cpu does not appear to be too far off from the E8400 in performance:
http://cpuboss.com/cpus/Intel-Core2-Duo-E8400-vs-Intel-Core2-Duo-E7600
...but i notice the E7600 is 1066 MT/s unlike the E8400 @ 1333 MT/s
so that seems like good news i think? ie. i would have thought a '1066->1333 pad mod' on the E7600 should work if the '1333->1600 pad mod' on the E8400 fails (or am i way off here, ie. is it a case of, they will most likely either both work, or none of them will )

i ended up getting impatient waiting for the silver conductive paint to arrive so decided to do a test using tin foil instead. as far as i could tell i did a good job of connecting together the 2 necessary cpu pads together.
the result, no success. the pc would not post or go to the bios screen, the vga didnt send any signal to the monitor. the pc did turn on, but just hung with the fans blowing at full speed.
i can only assume that it didnt work because, i modded the E8600 cpu to become a 1600mhz fsb cpu, but it looks like the intel G35 Q35 chipset does not support that (see pic below).
anyway so the next step, i will try to get an E7600 cpu instead (3.06ghz, 1066mhz fsb) and attempt to pad mod that, to make it 1333mhz fsb. i will post the results here so it might help other dell optiplex users.

That's a bummer, thanks for the follow up though.

i would suggest looking at what cpus that board supports. as their is an extreme cpu intel release using the 400mhz fsb, if the board supports that cpu then you could use the 400mhz fsb. though the no boot could also be because there is not enough voltage to the cpu, so you want to Vid-Pad the cpu. this isnt something that can be used for all cpus, what i mean by that there is not set use this or that for all. you first need you Vid of the cpu from coretemp/realtemp, then look at intels vid table for the cpu in their spec sheet PDF's. from there you can figure out the signaling of the pads to what you need to get to a higher voltage. it really does require a bit of digging and reading. Using alu foil does work but not as stable when installing a cpu as using the liquid defroster repair kit. might sound backwards but bet a 266/1066mhz fsb cpu, with the right cpu multi ark.intel.com and pad mod it for 333/1333mhz. that way you can get in the ball part of 3.6ghz but on stock voltage might be luck of the draw. im not sure how many cpus if any of the core 2's hit that high a cpu speed without some kind of increase to cpu voltage. the good news is that 266/1066mhz fsb cpus are some what cheap (i guess depending on where you live) so looking for a 266/1066mhz part might make more sense.

i will admit it looks like i missed a few of your other posts. in regards to the E7600 and going for 333mhz that would be higher then 3.6ghz and hitting that with no voltage increase is likely to not happen. i would suggest going with the E7400 with its 11x multi, so 11x333=3.66ghz. as the E7600 has a 11.5x multi, it may not sound like much but apply that to the 333mhz fsb and you wind up with 3.8ghz. now i have never seen a core 2 cpu hit 3.8ghz without a voltage increase. in most cases i have seen core 2 cpus hit 3.2-3.4ghz range with no increase in voltage. with out futher modding as in using the Vid-Pad mod the E7400 is the best shot though you are better off looking for the highest clocked E7x00 or E8x000 cpu and putting it in the board. as far as performance your not going to notice it with cpu speed being the same, while a higher fsb with the same speed is better. L2 size once past 2mb total doesnt offer any improvement that you would be able to notice in gaming or day to day use. Unless your doing heavy calcs like running the SuperPi 1m game, thats about the only area the more L2 will make a difference in performance.

Hello Evilsizer, thanks for the information, i won an E7600 on ebay a few days ago, here:
http://www.ebay.co.uk/itm/141585705043?nma=true&orig_cvip=true
so i will see how i get on with it when it arrives in a day or two.
im thinking of disabling several parts of the PC, as it is only used for emulating retro videogames, it is never used on a network or online etc. so i could disable the onboard modem, onboard network, serial port, parallel port, i dont need to ever use the CD ROM drive so that power/cables could be disconnected, as well as the front USB and audio ports. is doing all of that going to save much voltage so to speak, or is that irrelevant?
thanks!

Evilsizer said:

might sound backwards but bet a 266/1066mhz fsb cpu, with the right cpu multi ark.intel.com and pad mod it for 333/1333mhz. that way you can get in the ball part of 3.6ghz but on stock voltage might be luck of the draw. im not sure how many cpus if any of the core 2's hit that high a cpu speed without some kind of increase to cpu voltage. the good news is that 266/1066mhz fsb cpus are some what cheap (i guess depending on where you live) so looking for a 266/1066mhz part might make more sense.

i will admit it looks like i missed a few of your other posts. in regards to the E7600 and going for 333mhz that would be higher then 3.6ghz and hitting that with no voltage increase is likely to not happen. i would suggest going with the E7400 with its 11x multi, so 11x333=3.66ghz. as the E7600 has a 11.5x multi, it may not sound like much but apply that to the 333mhz fsb and you wind up with 3.8ghz. now i have never seen a core 2 cpu hit 3.8ghz without a voltage increase. in most cases i have seen core 2 cpus hit 3.2-3.4ghz range with no increase in voltage. with out futher modding as in using the Vid-Pad mod the E7400 is the best shot though you are better off looking for the highest clocked E7x00 or E8x000 cpu and putting it in the board. as far as performance your not going to notice it with cpu speed being the same, while a higher fsb with the same speed is better. L2 size once past 2mb total doesnt offer any improvement that you would be able to notice in gaming or day to day use. Unless your doing heavy calcs like running the SuperPi 1m game, thats about the only area the more L2 will make a difference in performance.

Click to expand...

It varies depending upon which model of Core 2 Duo you get.

My E7500 (2.93GHz), 2 different E8400 E0's (3GHz), my E8500 C0, and a Pentium E6700 Dual-Core (3.2GHz) all hit 3.6Ghz with stock voltage no problem. My E6700 Core 2 Duo couldn't do 3.6GHz without a massive voltage increase though (1.46v). My E4700 (stock 2.6GHz) could run at 3.6GHz with a very slight voltage boost to 1.36v (the stock voltage on that particular CPU was 1.35v). My E8400 C0 required a very small voltage increase to 1.28v (VID is 1.25v). I guess what that says is that if the stock frequency is greater than 2.6GHz you shouldn't have as much problem with voltage requirements for an overclock to 3.6GHz, and if the CPU is 2.9GHz or better at stock then hitting 3.6GHz with stock voltage is nearly guaranteed in my experience.

As for hitting 3.8GHz with stock/default voltage, both of my E8400 E0 samples hit that with less than stock voltage (I under-volted them in my BIOS) (1.16v and 1.216v, respectively (VID's are 1.2625v and 1.25v), my Pentium E6700 dual-core hit 3.8GHz with less than stock voltage as well (1.26v, stock was 1.325v), the E8500 C0 also ran at 3.8GHz with less than stock voltage (1.17v, stock is 1.20v on that one). The E7500 needed a slight boost to make it there though and required an increase to 1.31v (stock is 1.30v).

While it may be possible that some of my LGA775 CPU's are unusually good-performing samples with lower voltage requirements than is the norm, I highly doubt that all of them would be.

My source: I've benchmarked and overclocked on the LGA775 platform quite a bit with air cooling.

aargh, so close

ok so i got hold of an E7600 cpu (3.06 11.5x 1066 MT/s) and decided to give that a try with some pad modding

i first tested the E7600 at stock and it works fine. then i did the BSEL mod on it to get it to 1333 MT/s. what happened though is, although my pc did boot up (ie. i can go into the bios etc), as soon as the pc attempted to load windows, the pc would crash (either some garbled message in dos, or it would just reset). at this point i thought, ok let's try to raise the voltage. the stock voltage of my E7600 was reported as 1.2875v. so next i used the interactive excel spreadsheet which is available on the net called "C2D VMod Calculator.xls" to determine the type of VID (volt) mod to perform. i noticed that to jump from 1.2875v to 1.3875v was a relatively easy mod compared to some of the others (i just had to join 2 pads together... '4' and 'S'), as shown below:



here's where things get weird though...
i then attempted to boot up the pc and it worked, i got all the way to the windows desktop, and managed to get the following data:



something to notice is, both cpuz and core temp report my volts at 1.2875v ... im not sure what that means.. could they be reporting incorrectly, and if not, is it just coincidence that my pc booted up ok, and the cpu is indeed at 1.2875v (and that my VID (volt) mod has been unsuccessful?)

finally, after running the pc for about 10 minutes with the heavyload software stressing out the cpu, i got the screen below, and then shut down my pc. i havnt turned it on since as im not sure how to proceed. does anyone have any thoughts as to what is going on, or any advice on how to proceed. im more than happy to raise volts further (i know my temps are a little high but that's something i would resolve at the end (new heatsink/fan if necessary etc)

ps. thanks for reading this guys

The VID dialog box in Core Temp only indicates the default voltage that is programmed into your CPU, it does not indicate what your CPU Voltage/vCore is actually set to.

Core VID in CPU-Z is the same thing, just the default setting, and not necessarily what it's actually set to. Which is a bit weird, because in every other version of CPU-Z I've ever used it had a Core Voltage dialog box there instead, which did indicate the current voltage setting for the CPU vCore. I've never actually used the Version 1.71.1 of the software that you're running there. I'm on Version 1.70.0.

I'd recommend you try installing and running CPUID HWMonitor/Hardware Monitor to see what it says about your Core Voltage.

cool ok i will have a try tomorrow with CPUID HWMonitor/Hardware Monitor
if it turns out i am indeed at 1.3875 volts, would you recommend i raise it up just a bit more in the hope of avoiding a blue screen of death again? (and if so do you have a recommendation for how much of an increase in volts should be adequate) please note if anything goes wrong (damage) i wouldn't complain or moan here of course. it's all good fun and im not using anything expensive and wouldnt be upset etc (well maybe a little bit haha )

That BSOD code is usually a driver or memory error, and by seeing that big red artifact on your screen I'd be inclined to say driver.
If you're only on air cooling, I wouldn't go much past 1.4v

Mr.Scott said:

That BSOD code is usually a driver or memory error, and by seeing that big red artifact on your screen I'd be inclined to say driver.
If you're only on air cooling, I wouldn't go much past 1.4v

Click to expand...

if it was some sort of driver error going on, would i have any options to resolve that or would that possibly be a dead end so to speak? i havnt mentioned that the pc i am using for this overclock is running windows xp 64bit.. but i wouldnt have a problem switching to windows xp 32bit if necessary (or possibly windows 7 as a last resort if i had to)

kitt said:

if it was some sort of driver error going on, would i have any options to resolve that or would that possibly be a dead end so to speak? i havnt mentioned that the pc i am using for this overclock is running windows xp 64bit.. but i wouldnt have a problem switching to windows xp 32bit if necessary (or possibly windows 7 as a last resort if i had to)

Click to expand...

Try a different GPU driver. Might not be the driver at all. What's the GPU temp? Could be a bunch of different things.
I was just noting the code and the artifact. Artifacts generally aren't caused by a CPU overclock. That's the only reason I mentioned it.

run memtest86+ to see if you have any ram errors taking place, this would be just to make sure things are good to go ram wise. i would suggest using either xp-32bit if your under 4gb of ram, if you over that go to win7-64bit if you have it. i recall in the past people complaining about how bad windows xp-64bit was with drivers and random issues. i think there might be that possibility of something going on in xp-64bit if trying different drivers doesnt help.

there is also a possibility the power supply might be a bit overloaded, depending on the GPU being used.

*edit*
TechTweaker,
i hear ya, i stopped reading what peoples results where with core 2 cpus around the time i7's and the last models for core 2 came out. i just never had any luck with my E7200 hitting 3.6ghz on stock voltage, it did do a superpi 1m run at 4.6 i think or [email protected], might have been higher. my memory is so foggy these days about some of my oc's.