Quad Core Thread: Overclocking, Cooling, Motherboards, Troubleshooting & General Info

[fat jez]

No, AUTO gives it whatever it feels is needed, and sometimes it gets it wrong.

This is why, in the guide, I mention tweaking the voltages. Quite often, you can shave off a chunk of voltage by changing back to manual vcore.

Thanks, I'll have a play when I get home from work. Just need to work out what voltage the CPU is currently using, WITHOUT using the virus that is Asus Probe. I got the dreaded AsIO.sys missing error, even after removing the Probe software and it took me most of last night to get rid of it, going through the registry and removing all references to Asus and then doing the same on the C drive.

Clunk, which chipset blocks are you using?

Cheers,
Stephen

[bwgames]

Thanks, I'll have a play when I get home from work. Just need to work out what voltage the CPU is currently using, WITHOUT using the virus that is Asus Probe.

On that subject, is there a vendor-independent program that people would recommend for reading voltages?
CoreTemp is recommended for temperatures, but what about voltages?

I have to say, this is one aspect where I like Gigabyte - voltage modification in the BIOS is given as + / - the stock voltage, rather than just a list of voltages in themselves.

[Clunk]

Maybe a table of volts vs OC might help...

It wont work unfortunately because some chips need silly amounts and some need none, or just a tiny amount for a decent overclock

[Clunk]

On that subject, is there a vendor-independent program that people would recommend for reading voltages?
CoreTemp is recommended for temperatures, but what about voltages?

I have to say, this is one aspect where I like Gigabyte - voltage modification in the BIOS is given as + / - the stock voltage, rather than just a list of voltages in themselves.

For Asus, I use Asus probe 2.

There will often be a slight drop in voltage shown in windows due to droop (sometimes a large drop on Asus boards).

Just try a few different ones, and use the one that you prefer that is reasonably close.

I would suggest sticking with the manufacturers software if you are unsure.

[Clunk]

Thanks, I'll have a play when I get home from work. Just need to work out what voltage the CPU is currently using, WITHOUT using the virus that is Asus Probe. I got the dreaded AsIO.sys missing error, even after removing the Probe software and it took me most of last night to get rid of it, going through the registry and removing all references to Asus and then doing the same on the C drive.

Clunk, which chipset blocks are you using?

Cheers,
Stephen

Asus probe has always been ok for me for voltages. The only problem I had with it was that it didnt play nice with Vista x64, apparently it does now, but I've yet to try it

Latest version of Asus Probe 2.


Edit: Still looking for recommendations for Coolers, Waterblocks, Motherboards etc for use with Quad Core CPUs to add to the first post!

[fat jez]

Asus probe has always been ok for me for voltages. The only problem I had with it was that it didnt play nice with Vista x64, apparently it does now, but I've yet to try it
[/SIZE]

Yeah, that was the one I tried, but I was getting a popup window every time I rebooted my machine or started it up complaining it couldn't find AsIO.sys, which continued even after I removed it. Took me all last night to get rid of it. I read a few posts on the net from people who had reported this to Asus who claimed it wasn't even their software that caused it! (Even though it's part of Asus Probe!).


Clunk, which chipset block are you using on your motherboard?

Cheers,
Stephen

[Clunk]

Sorry, I meant to say before, its a bog standard DangerDen Maze4 chipset block with 1/2" barbs, all my stuff is here.

Not sure about the asIO thing, not used it for a while, but I will try it later and see if it does it on this vista x64 install.

[fat jez]

Sorry, I meant to say before, its a bog standard DangerDen Maze4 chipset block with 1/2" barbs.

Not sure about the asIO thing, not used it for a while, but I will try it later and see if it does it on this vista x64 install.

Make sure you take a system restore snapshot first, just in case! It didn't do it to me first time, but after I removed it and put on a later version, I got the problem.

As for your chipset block, what happened to the heatpipe cooling from the south to north bridge and onto the MOSFETs? Any pics so I can see what you've done?

Cheers,
Stephen

[Clunk]

Pics to it all above in the link

I have a ghost image of the install if it goes tits up

[daza]

I'll recommend the Quad gt again, as its 24hr stable at 355x9 (3.2ghz) bridge is at 1.45v and the cpu is at 1.37v, only problem is the mosfets need some serious cooling if your overclocking.

[Kilmatead]

The 19th-century explorer Richard Burton had a decisively negative opinion towards firearms. Not on the grounds of violence, mind you, he was not averse to regular "social altercations", as it were, but simply because they were unsporting. Any woman, child (and these days many do), or adult commoner could fire a gun and win a duel (this is the 19th century) without the slightest skill, forebearing, or (dare it be) 'elan'. He continued throughout his life to adhere to the sword, a man's weapon of choice, as it required training, experience, spirit, and a sort of "earned respect", if you will, to master such a weapon of honour.

This attitude could be moved to the modern by a general distain for those who buy Lamborghini's and Ferrari's without even a driver's license to boast of.

Richard Burton wouldn't have liked me; I'm a newbie with a gun. Don't hate me.

As the proud owner of an admittedly ridiculous QX6800 (running on a Commando), it was just the result of one of those spur-of-the-moment type of days when you have more money than sense, and no girlfriend in sight to tell you not to, if you know what I mean. Ah well, what's done is done.

This has led me thusly to your thread.

I've spent the last week happily reading Clunk's guide, related posts, etc. And I must thank him for even the simplest suggestions about setting memory latencies in the BIOS manually in lieu of just leaving everything to AUTO. I also spent my life's most frustrating day trying to adhere to his suggestions about balancing the heat-sink on the processor for even temps. Ultimately successful (thanks again) all was made clear when viewing his pictures of his own system: that is not the heavy blunted-fingered calloused hand of a stone-worker in those pics. Being a stone worker myself, it's pure hell trying to twist thumb-screws on a Tuniq Tower 120 mounted through the Commando's already ott heatsinks with my stubby fingers. Maybe major manufacturers just employ children and elves to assemble their systems.

Anyway, being a neophyte at overclocking I decided to try simple things. Copying a suggestion from a magazine, I raised the multiplier from 11 to 13, went FSB to 277, and 1.55V - and hey presto instant 3.6G. Of course my temps went through the roof (on air) so I put an end to that pretty quick, but it was fun to see the numbers for the fleeting moment. (I used to have a C2D 6600.)

In the guide you intentionally left out Extreme model stats, primarily concentrating on raising FSB frequencies and so on, but I settled to just leaving it at 266 and upping the clock to 12. This results (obviously) in a nice happy 3.2 with no real threat to temps or sanity. My question is rather simple: what's the difference between what I did, and say an FSB of 290 with multi of 11 (or any other mathematical similarity)? The end result seems the same. I did encounter a few times (when experimenting) that while the numerical computation would say the system was pushing 3.4 or so, real world tests showed the system running slower, which may have been the memory out of sync or something. All very confusing. And whats-her-name, Sylvia, Samantha, Sandra, quite happily tells me I'm running 7% above reference, but it just doesn't 'feel' right.

Am I confused? Or can intuition still count for something in this mathematically precise age? Burton's spirit might condemn me for what I bought, but his outlook might applaud my questioning the so-called 'proof' of numerical obstinacy.

Thanks for your time. (And all the suggestions!)

[Clunk]

The 19th-century explorer Richard Burton had a decisively negative opinion towards firearms. Not on the grounds of violence, mind you, he was not averse to regular "social altercations", as it were, but simply because they were unsporting. Any woman, child (and these days many do), or adult commoner could fire a gun and win a duel (this is the 19th century) without the slightest skill, forebearing, or (dare it be) 'elan'. He continued throughout his life to adhere to the sword, a man's weapon of choice, as it required training, experience, spirit, and a sort of "earned respect", if you will, to master such a weapon of honour.

This attitude could be moved to the modern by a general distain for those who buy Lamborghini's and Ferrari's without even a driver's license to boast of.

Richard Burton wouldn't have liked me; I'm a newbie with a gun. Don't hate me.

As the proud owner of an admittedly ridiculous QX6800 (running on a Commando), it was just the result of one of those spur-of-the-moment type of days when you have more money than sense, and no girlfriend in sight to tell you not to, if you know what I mean. Ah well, what's done is done.

This has led me thusly to your thread.

I've spent the last week happily reading Clunk's guide, related posts, etc. And I must thank him for even the simplest suggestions about setting memory latencies in the BIOS manually in lieu of just leaving everything to AUTO. I also spent my life's most frustrating day trying to adhere to his suggestions about balancing the heat-sink on the processor for even temps. Ultimately successful (thanks again) all was made clear when viewing his pictures of his own system: that is not the heavy blunted-fingered calloused hand of a stone-worker in those pics. Being a stone worker myself, it's pure hell trying to twist thumb-screws on a Tuniq Tower 120 mounted through the Commando's already ott heatsinks with my stubby fingers. Maybe major manufacturers just employ children and elves to assemble their systems.

Anyway, being a neophyte at overclocking I decided to try simple things. Copying a suggestion from a magazine, I raised the multiplier from 11 to 13, went FSB to 277, and 1.55V - and hey presto instant 3.6G. Of course my temps went through the roof (on air) so I put an end to that pretty quick, but it was fun to see the numbers for the fleeting moment. (I used to have a C2D 6600.)

In the guide you intentionally left out Extreme model stats, primarily concentrating on raising FSB frequencies and so on, but I settled to just leaving it at 266 and upping the clock to 12. This results (obviously) in a nice happy 3.2 with no real threat to temps or sanity. My question is rather simple: what's the difference between what I did, and say an FSB of 290 with multi of 11 (or any other mathematical similarity)? The end result seems the same. I did encounter a few times (when experimenting) that while the numerical computation would say the system was pushing 3.4 or so, real world tests showed the system running slower, which may have been the memory out of sync or something. All very confusing. And whats-her-name, Sylvia, Samantha, Sandra, quite happily tells me I'm running 7% above reference, but it just doesn't 'feel' right.

Am I confused? Or can intuition still count for something in this mathematically precise age? Burton's spirit might condemn me for what I bought, but his outlook might applaud my questioning the so-called 'proof' of numerical obstinacy.

Thanks for your time. (And all the suggestions!)

Are you really C3PO?

Anyway, performance will be very similar at the same clock speeds regardless of the multiplier. The only benefit that you might see from using a lower multi and higher FSB is that you will have more memory bandwidth, which may or may not benefit you.

As for things not feeling right, well, theres an easy way to tell if things are running right, and thats do an overnight stress test with something like prime95 beta, orthos (if itll run) or 4x SP2004 (and change the affinity yourself).

You might be right though in your suspicion that there is something wrong. As you may have noticed, the more you tinker with things, the more you get a feel for what is going to work, and you can anticipate when something is going to go wrong..sometimes

One thing that could possibly be wrong, is that your memory had reached its limit, or possibly you started with the RAM too high?

Post your bios settings, pics will do, and we will see what we can do.

[bigpun69]

i've set my vcore voltage to 1.45 and its stable now at 3.25GHz, does that voltage sound high?

[Kilmatead]

Undoubtedly C3PO has more dextrous fingers than I for poking around humming mobos.

My 'problem' (if so it be called) is less physical and more philosophical. I once had a physics professor who banned (on pain of failure) any student from asking a question beginning with "Why". He was quite happy to respond to any query of "How..." because that's what physics does: it describes HOW things work, not WHY they work. I'm having the same trouble with your guide.

I'm only too happy to screw things up. If my computer is running smooth then I'm not learning anything. It's boring. Thanks to your guide I have more "test" programs than I can shake a stick at. Took awhile to figure out at 'affinity' stuff in Prime95, and Sandra is a little over-detailed for beginners (if that's the 'Lite' version what in God's name do you get if you throw money at them?), but not un-useful. Fun, regardless. (Was forced to learn enlightening things about resetting CMOS too, when having too much fun though.)

My memory kept defaulting to 800 (when I know it's PC2 8500C5 1066 capable). So I messed around with the settings - et voila, DDR2 1066 enabled - which I must say did make a decent boost to real world application. I was happy. Incidentally the speeds weren't printed on the chips as per your guide, had to use this interweb thingie to discover 5-5-5-15. Which brings me back to one aspect of my 'problem'.

What the heck do these numbers mean to a beginner? I can grasp that 4-4-4-12 might be "better/faster" than 5-5-5-15 but you go on to suggest things like 4-3-4-8 (or something) and the lights go out. Most people understand that 1066 is 'faster' than 800, fair enough - but abstract latency timings, and WHY they are better, is confusing to newbies at the best of times.

Don't get me wrong, I'm not criticizing... I've learned more in the last week from mucking about with your guide than I have in the last year, and I'm all the happier-mind-boggled for it! I merely suggest that in future revisions (you strike me as a personality who loves to revise [that's for the C3PO jab] [smile!]) you don't just say "Now we're going to try this..." without offering some suggestion (i.e. context) as to WHY we should want to as opposed (like me) to just raising the multiplier and accomplishing the same thing. My fuzzy brain is loose on the push-pull relationships involved with this stuff, but we're learning. And don't even get me started on voltages. (Strangely, say for memory, as you suggest to manually set it [mine is rated for 2.2v] the BIOS gets very upset if I try to do this and prints the number in bright red - apparently to scare me off - but it reverts to a pleasant white text up to 1.95v. These things make me reach for more Twix.) Obviously, and unfortunately, you can't be a guru expert on any and all contrasting hardware possibilities, but is a small collegiate treatise upon Latency too much to ask? (Cough.)

Anyway, the struggle continues! (In a Galaxy far, far... oh get off it already.)

[Clunk]

i've set my vcore voltage to 1.45 and its stable now at 3.25GHz, does that voltage sound high?

If the temps are ok, and you are happy to run that voltage, then go for it

[smellybumlove]

just got mine installed, yes its not a go

anyhoo on my freezer pro its running cooler than the e2140!!!!, i got some artic silver on her and its idling at 46-48 and thats at 360x9 and 1.4v


hitting 62 under load on orthos, but it only does 2 cores??