E6600 Overclocking Problems

Hi Guys :D I've been sniffing around these boards while overclocking my machine but not registered until today as I need a bit of help now and was hoping someone would be kind enough to offer some advice:surrender:

You can see my config from the sig, I'm also using the Asus Silent Square Pro which is working wonders at the moment, the problem I'm having is random reboots when my CPU is clocked above stock, I just get a BSD even when not under load and when doing boring things like checking email! I have used Orthos and CPU-Z to check the clock and speedfan + coretemp to check temps. Orthos doesn't like working in blended mode on my machine, it throws up an error and closes at random intervals so I've been running the CPU test instead and after running for 1 hour the highest my temps have spiked is 48 degrees although they usually stick at around 46 for the duration when clocked to 355X9. I'm using the overclocking guide on this site as it's the best I've found so far (Thanks Guys :D) but am still having problems, could anyone help pretty please!!

Please bear in mind that I haven't done any overclocking since the days of jumpers and using a pencil on the L1 bridge so my knowledge is somewhat hazy on the area :confused:

I should add, the FSB's I've tried and found unstable range from 290 up to 355, I was hoping that the board would just push through the lower end problems as they shouldn't be an issue with the chip but that didn't seem to work out :o

It could be to do with certain things you need to disable in the bios to overclock effectively.

It might be in here

http://forums.hexus.net/showthread.php?t=103676

I havn't really read it but it should help.

thanks mike :D that was the guide i was talking about, i did try and post the link but can't as i'm a noob :(

i have disabled all the cpu features as in the guide, also changed the memory settings. at the moment i'm wondering if it could be due to a dodgy ram module but am unsure of how to test that

Did you manually enter the ram timings and voltages?

Is the cooler mounted properly?, double check this.

How are the temps load and idle?

Run memtest for a few hours.

Is your 4/8 pin EPS cable from the PSu plugged into the board?

Memtest86+
runs from a floppy or cd. Start by pressing [F8] key if using cd, when u startup.
or
Memtest HCI
Or
OrthosThis is a general stability test but running it with large ffts will test nenory more than cpu.
or
OCCT
Another good stability tool. this takes 30min to run but is a good quick way of determining that your stable.

so in summary 1st 2 to check memory
last 2 general stability - if they fail it may be cpu or memory issues.

welcome to the forum btw:D
luck

Quote:

---

Originally Posted by Clunk

Did you manually enter the ram timings and voltages?

Is the cooler mounted properly?, double check this.

How are the temps load and idle?

Run memtest for a few hours.

Is your 4/8 pin EPS cable from the PSu plugged into the board?

---

I've manually entered the RAM settings as per the guide linked above (5-5-5-15, can't remember the rest off the top of my head and I'm at work so can't check)

The cooler is properly mounted, I was very anal about this as I've burnt out a chip before through my own idiocy!

temps went up to 48 degrees at a 3.2 (355x9) clock whilst under load but are usually around 46 under load, 28-30 degrees ish on Idle.

Not run memtest yet but will do as soon as I get home :)

The 4/8 pin EPS cable (I'm assuming you mean the heatsink fan cable) is plugged in and working as I've checked the fan speeds etc in the BIOS.

No, not the fan cable, there are 2 wires from the PSU that plug into your motherboard. One os the main 24 pin ATX lead, the other is a smaller 4 or 8 pin cable, and it usually goes at the top of the motherboard.

The memtest thing will help you rule out other stuff. memory problems are a ballache to diagnose properly, so if it is ram, its sometimes just better if one of the sticks is dead, rather than dying :)

You 100% sure that you have manually entered the correct voltage for your ram?

Quote:

---

Originally Posted by Clunk

No, not the fan cable, there are 2 wires from the PSU that plug into your motherboard. One os the main 24 pin ATX lead, the other is a smaller 4 or 8 pin cable, and it usually goes at the top of the motherboard.

The memtest thing will help you rule out other stuff. memory problems are a ballache to diagnose properly, so if it is ram, its sometimes just better if one of the sticks is dead, rather than dying :)

You 100% sure that you have manually entered the correct voltage for your ram?

---

ah, that cable! yeah, that's plugged in and working fine

i'm really hoping it's not ram as that'll be a complete pain, i'm hoping I've just been a noob and overlooked something silly!

i'm not 100% about the ram voltage, I've entered the ram settings on the guide but if there's something else i should be doing then I've missed it...

Quote:

---

Originally Posted by Clunk

You 100% sure that you have manually entered the correct voltage for your ram?

---

Just a general observation and is loosely linked to the thread:-

Surely with more voltage and the ram runs hotter?

On auto on DS3P the voltage for Geil is 1.85v...yet minimum supposed to be 1.9v and should be 2.1 on the stick, yet on every test I've done e.g orthos, my system is stable (erm well haven't done memtest).
So if I have a decent OC and it's stable why increase mem voltage?

Quote:

---

Originally Posted by magicmonkey

ah, that cable! yeah, that's plugged in and working fine

i'm really hoping it's not ram as that'll be a complete pain, i'm hoping I've just been a noob and overlooked something silly!

i'm not 100% about the ram voltage, I've entered the ram settings on the guide but if there's something else i should be doing then I've missed it...

---

have you, by chance, set your ram to run at 800mhz and then started to overclock the cpu?

You need to make sure that you have entered the correct voltage for the ram. take one of the sticks out, itll say something like 1.9v or 2.1v, go into the bios, there will be a setting to change the dram voltage, possibly near the bit where you change the fsb.

Quote:

---

Originally Posted by excalibur2

Just a general observation and is loosely linked to the thread:-

Surely with more voltage and the ram runs hotter?

On auto on DS3P the voltage for Geil is 1.85v...yet minimum supposed to be 1.9v and should be 2.1 on the stick, yet on every test I've done e.g orthos, my system is stable (erm well haven't done memtest).
So if I have a decent OC and it's stable why increase mem voltage?

---

Yes, the ram will run hotter, and dont call me shirley :D

I wasnt actually telling YOU to up your voltage, I was telling magicmonkey to check that his is correct.

if your system runs stable at lower voltage than the ram needs, then great, no need to change it, but you have to remember that it is rated at what it is rated at for a reason, and setting it to the rated voltage is just one less thing to worry about when fault finding. :)

Quote:

---

Originally Posted by Clunk

Yes, the ram will run hotter, and dont call me shirley :D

I wasnt actually telling YOU to up your voltage, I was telling magicmonkey to check that his is correct.

if your system runs stable at lower voltage than the ram needs, then great, no need to change it, but you have to remember that it is rated at what it is rated at for a reason, and setting it to the rated voltage is just one less thing to worry about when fault finding. :)

---

BTW does the principle of a faster side bus equals faster computer, or am I confused about something in the past:-


Which is faster....... cpu@9 X 300 = 2.7ghz

or

cpu@6 X 450 = 2.7ghz

Quote:

---

Originally Posted by Clunk

have you, by chance, set your ram to run at 800mhz and then started to overclock the cpu?

You need to make sure that you have entered the correct voltage for the ram. take one of the sticks out, itll say something like 1.9v or 2.1v, go into the bios, there will be a setting to change the dram voltage, possibly near the bit where you change the fsb

---

the ram was set to 1:1 so that should be ok, i haven't changed the voltage though as i want to try and leave that alone as much possible. i'll give this a go when i get home later, fingers crossed...

Quote:

---

Originally Posted by Clunk

Yes, the ram will run hotter, and dont call me shirley :D

I wasnt actually telling YOU to up your voltage, I was telling magicmonkey to check that his is correct.

if your system runs stable at lower voltage than the ram needs, then great, no need to change it, but you have to remember that it is rated at what it is rated at for a reason, and setting it to the rated voltage is just one less thing to worry about when fault finding. :)

---

BTW does the principle of a faster side bus equals faster computer, or am I confused about something in the past:-


Which is faster....... cpu@9 X 300 = 2.7ghz

or

cpu@6 X 450 = 2.7ghz

At equal clock speed the higher fsb will give more performance.

Without trying it, I would have said that the 9x300 would give better performance, just because @ 400-460 ish FSB, there is a large hit in memory bandwidth due to the changeover of chipset straps at 400.

Try it and see. 400 onwards to about 460= drop in memory bandwidth.

Quote:

---

Originally Posted by magicmonkey

the ram was set to 1:1 so that should be ok, i haven't changed the voltage though as i want to try and leave that alone as much possible. i'll give this a go when i get home later, fingers crossed...

---

You need to set the ram voltage to what it is supposed to run at. Its not going to harm it, and you wont be running it out of spec.

Quote:

---

At equal clock speed the higher fsb will give more performance

---

plus if memory can handle it that will also be faster

Thanks for the advice guys, I'll give it a try and let you know how I get on later tonight :D

Well, I've got home (finally, that was one hell of a long day!) and upped the RAM voltage to 2.10 from 2.0 as per the RAM specs, run memtest to 100% coverage and found no errors so clocked the CPU to 355x9 (3.2ish). So far it's running fine but I've only been up for about 20 minutes, running Orthos still returns an error though, could anyone tell me what this means :

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 1 minutes 27 seconds - 1 errors, 0 warnings.
Execution halted.

I'm a bit perplexed! Other than that little glitch, running on CPU stress only I'm getting temps of up to 39 degrees under load which I'm quite impresses with so have to recommend the Asus silent square pro for anyone looking for a new heatsink :D

Anyway, any advice on what that error might mean in terms of problems would be welcomed, I'm off to search google until my eyelids close of their own accord...

If orthos was set to test using small ffts - cpu only then it's most likely you need more vcore (cpu voltage)

what is your cpu voltage set to in bios ??

Orthos was running a blended test when it gave me the errors, no errors while running small ffts - cpu only though, that runs fine which is why I was thinking RAM although memtest seems to have ruled that out... I'm just about to up the voltage one notch to the CPU, will post the numbers in a minute...

It's most likely a memory error.
when u say you ran memtest which 1 was it Memtest86+ or Memtest HCI??

Reason i ask is that both have their strengths & weaknesses:-

Memtest86+
Strength:-
tests nearly all memory as it runs in it's own enviroment
wide variety of tests
weakness
doesn't test memory as it is used in windows only test 5 & 8come close to emulating that. This often means that you can pass this memtest but still hve issues in windows.

memtest hci
advantage
runs in window so complements 86+ weakness. To be most effective run 2 instances.

disadvantage
only tests windows free memory - u need the full version to run like 86+, though in faiurness it isn't that dear.

Only by running both for longish periods do you get comfort that memory is ok.

my own preference i think would be to run 86+ 1st test 5. if this errors then tweak your voltage +/- a notch & then rerun stest. If that runs clear then run test 8 for an hour or so.

then i'd use HCI memtest in windows:-
testing as:-

From Manual:-
if those are good then go into xp & run hci overnight
it works in % coverage
100% coverage (a quick test to make sure your RAM is functioning reasonably)
Test for 1 hour (detects the majority of memory errors)
Test overnight (this will catch virtually all errors in your memory)



luck:)

Ok, have upped the CPU voltage to 1.35 and I'm still getting the same error, I'm on 12 hour shifts at the moment so I'll start memtest running in windows before I go out tomorrow and see what happens when I get back...

Quote:

---

Originally Posted by Clunk

Without trying it, I would have said that the 9x300 would give better performance, just because @ 400-460 ish FSB, there is a large hit in memory bandwidth due to the changeover of chipset straps at 400.

Try it and see. 400 onwards to about 460= drop in memory bandwidth.

---

Well it is difficult to get too excited about OC with just an arctic pro.......IMHO it's the weakest link for me and starts struggling over 2.95ghz to get CPU temp below 65c during heavy CPU test load, on a cool day.

Mind you with three different temp measuring utilities I'm not sure what to trust, the bios gives very low temps compared to core temp...............I suppose if I get crashes on a hot day playing something like COD2 then I will know what one is accurate.

You can't possibly compare temp in bios with coretemp. that's like comparing an idle temp with a load temp & asking why is it dfifferent.

when looking in bios there is virtually no load on cpu, cortetemp in windows on the other hand hmay have a load that varies between 0~100% in a nmatter of seconds.

coretemp is probably about the best for windows, using relatively little resource small desktop footprint, but whatever takes your fancy.

I prefer TAT to coretemp as it gives me the worst case scenario.. Coretemp reads a few degrees lower.

Must vary with build coretemp for me reads a degree or so higher,

Quote:

---

it gives me the worst case scenario..

---

is another reason i prefer coretemp:mrgreen:

Quote:

---

Originally Posted by Supershanks

You can't possibly compare temp in bios with coretemp. that's like comparing an idle temp with a load temp & asking why is it dfifferent.

when looking in bios there is virtually no load on cpu, cortetemp in windows on the other hand hmay have a load that varies between 0~100% in a nmatter of seconds.

coretemp is probably about the best for windows, using relatively little resource small desktop footprint, but whatever takes your fancy.

---

Well I'm thinking that Gigabyte's easytune is reading the temps in the bios as you click on a button with the same name i.e PC health........erm maybe it doesn't.
Temp in Tat is very high when using their CPU stress part........but some guys reckon it's not a real world test (it's worse) and orthos should be used instead.
Anyway I'm using core temp only because the guys in the know here are using it, and not because of detailed research by myself.

Well, I ran memtest while at work and came home to find that my machine had rebooted itself and 'recovered from a serious error' Looks like it's time to return the RAM and see what happens with the next batch. Do you think it's best to contact the people I got it from or Geil directly? Or should I carry on to find out which stick it is before returning it?