Is my Conroe dying? *updated*

[LucJoe]

*see update in Post #24* Much better temps with CM Hyper 212+

E6850
Gigabyte EP35-DS3L
4gb g.skill ddr2 1000

I decided to try to push my processor tonight and read some threads of people going to 3.8ghz easily with the stock heatsink. So off I went, using http://www.tomshardware.com/forum/245679-29-guide-overclocking-ds3l-updated-2008 to guide me.

I set up an overclock of 3.6ghz @ 1.46v set everything up and rebooted. I noticed first the temps were the same and then realized the overclock didn't even take (apparently). I decided to run Prime95 torture test at stock speed and very quick ran up to tjmax and throttling back to 6 multiplier. This is not oc'ed!

Anyone have any expert advice here? I seem to be experiencing a good amount of vdroop as well. My processor should be running 1.35v stock and apparently vdroop is common for this motherboard...

 

[Arkaign]

It's almost certainly your mobo. Sorry about that, but P35 mobos are rampant with that problem.

Also, people's ridiculous ideas aside, 3.8ghz on a stock heatsink with a 65nm C2D is INSANELY unwise.

 

[LucJoe]

It's almost certainly your mobo. Sorry about that, but P35 mobos are rampant with that problem.

Also, people's ridiculous ideas aside, 3.8ghz on a stock heatsink with a 65nm C2D is INSANELY unwise.

Well I was only going for 3.6

Honestly though I should be safe with SOME overclock. Thoough I'm definitely not against grabbing a cheap aftermarket heatsink to squeeze some life out of this.

Would it be completely crazy for someone who doesn't know what they're doing to attempt this vdroop fix mod? http://www.overclockers.com/forums/showthread.php?t=558347

 

[Spikesoldier]

p35 is capable of pushing cpus pretty far, but not extreme 500fsb oc's like p45's.

3.7 with a e6750 was doable with ip-35e, but i agree, not with the stock heatsink.

 

[LucJoe]

p35 is capable of pushing cpus pretty far, but not extreme 500fsb oc's like p45's.

3.7 with a e6750 was doable with ip-35e, but i agree, not with the stock heatsink.

I think we're missing the main problem here...

I'm @ Stock! 3.0ghz and my processor hits Tj. Max (95C!) with Prime95 testing.

 

[podspi]

I think we're missing the main problem here...

I'm @ Stock! 3.0ghz and my processor hits Tj. Max (95C!) with Prime95 testing.

I think the main problem is you need a new cooling solution! Even if your CPU was dying, it wouldn't get hotter without you supplying more juice to it first. It would just become unstable.

 

[taltamir]

I think the main problem is you need a new cooling solution! Even if your CPU was dying, it wouldn't get hotter without you supplying more juice to it first. It would just become unstable.

bingo. I see no reason to think that the CPU in question is dying... just that it is being poorly cooled.

Don't believe every chump that tells you how anyone can get this awesome super uber OC with stock HSF; it might be doable but it is rare.

Start your OC adventure with the cooling solution... oh, and make sure you set voltage to manual. Auto =! default, auto means that the mobo will use the default voltage at default clockspeed, and overvolt the hell out of your CPU if you try to overclock.

 

[WildW]

I set up an overclock of 3.6ghz @ 1.46v set everything up and rebooted. I noticed first the temps were the same and then realized the overclock didn't even take (apparently). I decided to run Prime95 torture test at stock speed and very quick ran up to tjmax and throttling back to 6 multiplier. This is not oc'ed!

Are you still running with 1.46v VCore? It doesn't appear to be the case from your screenshot, but you talked about some droop too. 1.46 is a LOT of volts for your CPU and could be generating some significant extra heat, regardless of clock speed.

 

[tweakboy]

You might need to use offset ,

I know I use -15 offset .. gl

Also for vdrop you can fix that with a pencil volt mod. You have to find on the net pic of your mobo and which transistor you color in. then you will get 0.0 v drop like me... thx

 

[LucJoe]

bingo. I see no reason to think that the CPU in question is dying... just that it is being poorly cooled.

Don't believe every chump that tells you how anyone can get this awesome super uber OC with stock HSF; it might be doable but it is rare.

Start your OC adventure with the cooling solution... oh, and make sure you set voltage to manual. Auto =! default, auto means that the mobo will use the default voltage at default clockspeed, and overvolt the hell out of your CPU if you try to overclock.

Yeah, I'm just a little worried about generating so much heat at stock speeds. I'll grab a hyper212 or something and see what happens

Are you still running with 1.46v VCore? It doesn't appear to be the case from your screenshot, but you talked about some droop too. 1.46 is a LOT of volts for your CPU and could be generating some significant extra heat, regardless of clock speed.

No, like I said the overclock didn't work. Apparently the changes I made in BIOS got reverted or something. I'm trying to figure out my heat problems at stock speed before trying again.

The CPU is set to 1.35v in bios now but as you can see is only getting ~1.3 (drops even lower sometimes)

You might need to use offset ,

I know I use -15 offset .. gl

Also for vdrop you can fix that with a pencil volt mod. You have to find on the net pic of your mobo and which transistor you color in. then you will get 0.0 v drop like me... thx

Not sure what you mean by offset.

As for the pencil mod, look at my post up a few.

 

[taltamir]

I have no idea what tweakboy is talking about either (what offset?)... I am not sure he does either.

As for pencil mod to eliminate vdroop... thats a load of BS. Anyways, your vdroop isn't that bad, certainly wouldn't account for the thermal protection throttling that you are seeing.

 

[Cerb]

Apparently the changes I made in BIOS got reverted or something. I'm trying to figure out my heat problems at stock speed before trying again.

If it can't finish the POST, it will go back to default for most of the CPU and RAM settings (AFAIK, all the GB P35 boards have this behavior, which saves you from resetting the CMOS all the time when OCing). You should see the setting you chose, but then either "auto" or the actual value of the default setting next to it.

You can get a pretty nice HSF for <$40, these days, so definitely try that (make sure to have some alcohol, and if the cooler comes w/ no TIM, get some).

 

[LucJoe]

If it can't finish the POST, it will go back to default for most of the CPU and RAM settings (AFAIK, all the GB P35 boards have this behavior, which saves you from resetting the CMOS all the time when OCing). You should see the setting you chose, but then either "auto" or the actual value of the default setting next to it.

You can get a pretty nice HSF for <$40, these days, so definitely try that (make sure to have some alcohol, and if the cooler comes w/ no TIM, get some).

Just ordered a Hyper212 for $30 along with http://www.amazon.com/ArctiClean-3-5...1417976&amp;sr=8-2 this kit for cleaning/thermal compound.

Will report back in 2 days!

 

[Morg.]

I don't remember which CPU that is, but you should have no trouble getting it past 4 ghz with a decent cooler.

 

[Arkaign]

I don't remember which CPU that is, but you should have no trouble getting it past 4 ghz with a decent cooler.

I believe you are mistaking this for the 45nm Penryn. Getting 65nm C2D past 4ghz on air is almost unheard of, even with the best of the best models.

 

[OCGuy]

OP you have to be thinking about the 45nm chips.

 

[bkiserx7]

My 6420 (65nm) used to push 3.9, but on water mind you. The E3200 (45nm) in that board now runs nice and cool at 3.8 on a CM 212+

 

[tweakboy]

I have no idea what tweakboy is talking about either (what offset?)... I am not sure he does either.

As for pencil mod to eliminate vdroop... thats a load of BS. Anyways, your vdroop isn't that bad, certainly wouldn't account for the thermal protection throttling that you are seeing.

offset to get proper temp.

CPU's mobos vary

I use -15 offset

You figure out what your offset is. Wheter you need to change it or not.

Im guessing,, -15c offset now go google in offset cpu and quit nit picking people.

 

[taltamir]

Temperature measurement offset! Why didn't you say so...

there is absolutely nothing the user needs to do with the RAW data coming off of the CPU's built in power sensor in regards to offset the values in order to get the real ones. Any recording software knows to do that automatically.

The reason there is even an offset is because CPUs do not report their current temperature, they report how many degrees they are below their max safe operational level. So, Core2, Nehalem, and Phenom2 each have a different "max temperature" and thus would need a different mathematical formula (a simple subtraction) to get the temperature from the value reported by the CPU...

Now, an "offset" is only ever needed when a new CPU family is just released and software has not been updated to detect them and use a correct formula, the latest formula for that company would be used instead which would make the reported value probably incorrect. Sometimes there are differences even within a family, for example some core2 chips meant for mobile have a lower TJMax value. If you are using a processor whose TJMax is 95 C and measuring the temp with software that was thinks it has a TJMax of 105C, then the reported temp will be 10 degrees higher then the actual measured temp.
New architectures are rapidly and easily added to measuring software as soon as they become available; and the program you are using (core temp 0.99.8) correctly detected the values for your processor and is displaying the true temperature.

So, bottom line there is absolutely nothing you, the user, need to do about it.

 

[Morg.]

I believe you are mistaking this for the 45nm Penryn. Getting 65nm C2D past 4ghz on air is almost unheard of, even with the best of the best models.

Errr...

I really thought the e6300 and e6600 were 65nm and I've oc'd both past 4 ghz on air - even one with a shit cooler. (the e6300)

And I checked .. those C2D were indeed Conroe's.

So I'm yet to find a Conroe C2D that won't do 4ghz (the e6600 reached it easily w/ 1.45 vcore more or less, got a lot of drop on that cheap asus board (set to 1.575 or 1.6 I don't remember) and supposedly the P55 is shit at OC but w/e)

Oh and, the temps on that E6600 are just fine .. doesn't get past 70&#176;C on prime95. (I've got to check, because since my cardboard ghetto 170mm fanmod (shit case fan from thermaltake who said it was a 200mm the bastards) - I think the temps changed a bit.

Also I just checked those 45nm you were talking about, my brother got one of these and it went to 4.5ghz easy on air .. don't know where you get your numbers tbh.

 

[betasub]

^ orly? The original C2D E6300 (Conroe) is 1.86GHz at stock. That's 7x266, so getting to 4GHz would require a FSB>571 (and great dividers/memory too!). The higher Conroes with 9+ multipliers could hit 3.6+GHz, but 4GHz was definitely unusual.

The newer Pentium E6300 (Wolfdale) is 2.80GHz at stock, and with 14x200, getting to 4GHz only requires an easy FSB~286.

 

[T101]

Get a better heatsink, the stock intel heatsinks are not really made for heavy use. It will permit the cpu to run as hot as its specs allow.

 

[Morg.]

^ orly? The original C2D E6300 (Conroe) is 1.86GHz at stock. That's 7x266, so getting to 4GHz would require a FSB>571 (and great dividers/memory too!). The higher Conroes with 9+ multipliers could hit 3.6+GHz, but 4GHz was definitely unusual.

The newer Pentium E6300 (Wolfdale) is 2.80GHz at stock, and with 14x200, getting to 4GHz only requires an easy FSB~286.

Meh .. I'm checking with my lil bro to see if his e6300 was a conroe .. but for the e6600 I can guarantee it I run it all the time like that with a fz120.

 

[LucJoe]

UPDATE!

Much better temps now, and I'm running at 3.6

I'm idling at 26-27 (was 59-60 before) and load up to ~50 (was throttling at 95 before)

Very happy with the 212

Question though... after a while in Prime95 it shut down my 2nd core due to an error (first core kept running)

It said "FATAL ERROR: Rounding was 0.5, expected less than 0.4"

Is this something to worry about? Needs a little more vcore maybe?

 

[betasub]

^ Yup, that's a Prime95 error indicating it has detected a calculation error i.e. you are NOT Prime95 stable (FWIW). As your temps are fine, try bumping the Vcore up a step.