Need help with overclock....

W

wasteomind

Gawd
Joined
Aug 13, 2004
Messages
522
Hello.

I've been working on a new computer I built recently.I'm kind of new to this generation of O/C. Previously I was using a AMD 2100XP processor and PC2700 mem. Needless to say the O/C ability of that system isn't close to this one. I've been trying to overclock my core 2 duo 6600 and 2 gigs of ddr800 memory as best as possible. My goal for the processor is 3.6ghz which would give me a nice 50% overclock increase. Memory I'm not really sure what would be optimal for that cpu speed.

This is my system:


Thermaltake Armor Series VA8000BWS Black Aluminum / Steel ATX Full Tower Computer Case - http://www.newegg.com/Product/Product.asp?Item=N82E16811133154

EVGA 122-CK-NF68-AR LGA 775 NVIDIA nForce 680i SLI ATX Intel Motherboard - http://www.newegg.com/Product/Product.asp?Item=N82E16813188009

Intel Core 2 Duo E6600 Conroe 2.4GHz 4M shared L2 Cache LGA 775 Processor - http://www.newegg.com/Product/Product.asp?Item=N82E16819115003

ZALMAN 9700 LED 110mm 2 Ball CPU Cooler - http://www.newegg.com/Product/Product.asp?Item=N82E16835118019

G.SKILL 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual Channel Kit Desktop Memory - http://www.newegg.com/Product/Product.asp?Item=N82E16820231065

Creative 70SB046A00000 7.1 Channels 24-bit 192KHz PCI Interface Sound Blaster X-Fi XtremeGamer Fatal1ty Professional Series - http://www.newegg.com/Product/Product.asp?Item=N82E16829102005

Western Digital Raptor WD1500ADFD 150GB 10,000 RPM 16MB Cache Serial ATA150 Hard Drive - http://www.newegg.com/Product/Product.asp?Item=N82E16822136012

SAMSUNG
SAMSUNG Black 18X DVD+R 8X DVD+RW 8X DVD+R DL 18X DVD-R 6X DVD-RW 12X DVD-RAM 16X DVD-ROM 48X CD-R 32X CD-RW 48X CD-ROM 2M Cache IDE DVD Burner With LightScribe Technology - http://www.newegg.com/Product/Product.asp?Item=N82E16827151136

eVGA 768-P2-N831-AR GeForce 8800GTX 768MB 384-bit GDDR3 PCI Express x16 HDCP Video Card - http://www.newegg.com/Product/Product.asp?Item=N82E16814130072

OCZ GameXStream OCZ700GXSSLI ATX12V 700W Power Supply 100 - http://www.newegg.com/Product/Product.asp?Item=N82E16817341002






Right now I seem to be stuck around 3.4ghz. I've been testing with Orthos beta, but it generates errors around 10 minutes at a rated fsb of 1540 (385fsb).

Printed on the cpu before I installed it was what i believe is the stepping info : L632A734

I was hoping to get some advice to help push my system to the 3.6 goal, and get some suggestions on how to optimize my ram settings.

Some screen shots of my settings follow. I would also welcome, Anything else that seems out of place or non-optimized in my bios.

Thanks for any advice:

BIOSsettings.jpg


cpuz-orthos-intel.jpg
 
I've got a pretty similar setup. I didn't see anything way out of line on your bios settings. Your vcore should be fine for 3.6 @ 1.4v for a b2 chip, but if your getting errors on small fft's it may want more. At 1600fsb I only needed to run the northbridge at 1.45v also. The nb is hot enough as it is. Your setup should be capable of 3.6 without too much of an issue. Past 3.6 I've found it wants a heck of alot more vcore to run stable and the temps shoot way up on air cooling. Going by your screenshots I take it your errors were on small fft's? If so your chip most likely wants more juice. Give it 1.425v and test it again. Mine runs alright on 1.40 but chips can be different. As long as you stay 60c or less at load you're fine. Most apps you run don't load a cpu like orthos does anyway. Some go higher but I don't like to play games with heat. Always nice to have head room when ambient temps rise.
 
It won't stay orthos stable at 3600mhz fsb of 400 at 1.45v. I pushed the vcore up to 1.55v in the bios, which i guess is about 1.5v real?

The temps shoot way up to 66c at full load, and in CPU-Z the voltage is showing 1.213v

That can't be right. This time though, Orthos didn't immeadiately fail.

I guess the first question is, is this amount of heat/voltage safe?

and second is there something else in my bios that is holding me back from O/Cing at a lower voltage? A memory timing or something?

What strikes me is your system and mine are very close, but yours O/C's so much better. Can the chips really be that much different?
 
Why don't you try lowering the multiplier to 8 (it's painful to sacrifice a multiplier... but you might as well see if it works) and setting the fsb to 450? See if that does anything for you. I have an E6600 but I think the ds3 doesn't like Micron D9s so I was only able to push the fsb to 440 while keeping 4-4-4-12 timings. When I tried overclocking with the x9 multiplier (at 3.5 ghz), it required more vcore and wasn't even stable at 1.5V. But, at 440x8, I'm stable at 1.4625V (or something VERY close to that, I can't check my bios right now but I'm pretty sure that's right). I didn't get the best processor, but it's decent. I think OCing is part luck with getting good components. If you're able to get to 3.6 at 1.4v, I think it's simply because you got an amazing processor. Anyway, my knowledge of overclocking is limited, but these were just my experiences...
 
That's one awesome RIG you have there?


1: Did you install the Nvidia chipset fan?

2: Have you tried a 400+ FSB already? If no...


Set your FSB to 1620 and your RAM to 810mhz

FSB and RAM unlinked. Set your CPU Voltage to 1.5V.

If the above doesn't work, try disabling LINKBOOST and GPU EX.

Also try to bring the HT multiplier to 3X (For testing purposes)

All your other settings look fine to me.
 
Yep I installed the chipset fan.

I've tried a 400fsb but doesn't seem to want to be stable at 1.5v or less. Worried to take the vcore higher, and at 1.5v full load is 67c. Not sure thats where I want to be.

What will disabling linkboost and gpu ex do for my cpu overclock?

and sorry to sound noobish but what is the HT multiplier that you suggested i change to 3?

Also what disadvantages are associated with changing my cpu multiplier from 9 to 8 to achieve a higher fsb oc?
 
Truth is that not all chips can do 3.6ghz, and not all boards can handle a high FSB.

The Linkboost and GPU EX settings apply to your Video Card and the PCI-E lane, and since we are trying to find the reason of your cpu's "lock", we can disable that for now, @ least while testing.

The HT multiplier handles system's bandwidth size. I said to bring it down, just in case it makes a difference since honestly I don't see anything in the settings you are already using that could be stopping you from going any higher...
 
Lets say that the HT multiplier being lowered allows for a higher FSB clock, what disadvanage does that leave me with having a lower HT multiplier since I will have a lower bandwidth?
 
orion23 said:
That's one awesome RIG you have there?


1: Did you install the Nvidia chipset fan?

2: Have you tried a 400+ FSB already? If no...


Set your FSB to 1620 and your RAM to 810mhz

FSB and RAM unlinked. Set your CPU Voltage to 1.5V.

If the above doesn't work, try disabling LINKBOOST and GPU EX.

Also try to bring the HT multiplier to 3X (For testing purposes)

All your other settings look fine to me.
Click to expand...

Tried all of this, with different combinations of Linkboost, GPU EX, and HT multi @ 3. It would always cause orthos to error within 3 minutes.
 
I've seen a few people complaining about week 32 chips... Not all chips can do 3.6ghz....

Have you tried upping the CPU Voltage? Try 1.55V (in windows it should be about 1.5V Real)

The only other thing I can suggest is to change your RAM's timings to "SPD" or manually set them to 5-5-5-15 (for testing purposes).

Also play with the SPP, MCP voltages a bit. Here is a guide (from Nvidia) about overclocking the 680i boards...

http://www.nvidia.com/docs/CP/45121/nforce_680i_sli_overclocking.pdf
 
try 430x8 and keep the RAM divider at 1:1, then bump the chipset voltage a little bit.

puts your CPU @ 3440, and RAM @ 860
 
Thank you to everyone who has tried to help me with this, I think i'm coming to terms with the realization I didn't get a chip realisticly capible of the 3.6ghz o/c I had hoped for.....kinda sad, but not the end of the world.

I'd like to ask a few more questions if anyone is still willing to help me.

I've had the core voltage up to 1.55v in the bios, shows as 1.51v in bios actual, but 1.23something in CPU-Z.

I've got the cpu running at 3.6ghz (400fsb) 6 hours double orthos (two instances) stable with 1.50v. Usually in its sixth hour it errors.

Thing is it runs about 65c. I'm concerned thats too hot, and it still isn't as stable as I would like.

How hot is too hot? I've noticed most people like to stay mid 50's for temps. Am I putting myself at risk having a processor running at 65c?

My memory settings I have kept at 4-4-4-12 since thats what the ram was rated for on newegg, and it appears stable at those timings. Would 5-5-5-15 give my processor the head room it needs to hit 3.6ghz at a lower cpu voltage to lower core temps? Or are those completely unrelated?

Finally, I've seen the suggestion to lower the multiplier, and have thought this myself too. If due to core temps being high, I need to live with a say, 3.5ghz o/c, would it be more benefital to have that 3.5ghz at a 9x multiplier with say 390 fsb?

Or is it better to have it with a 8x multiplier with a 438 fsb? Assuming thats possible.

Again thanks for helping with my noobish questions.
 
Hey man, I'm still a bit noobish myself, but I've been reading everything I can find on the Intertron for about two months now, and I do have a few comments/ideas. In no particular order:

* 65C does seem to be the temp ceiling for what most OCers are comfortable with, but you're nowhere near frying your proc at those temps. According to Intel, 80-85C is the threshold of the danger zone. Of course, if you're seeing load temps in the 70s you'd probably want to change some settings or modify your cooling hardware to bring them down, but there's no reason to freak out and immediately shut down your PC or anything. I've seen a few posts from noobs who had temps in the 90s (and even one who claimed 115C) without permanently damaging their CPUs. (These temps were due to improperly installed cooling hardware, obviously.)

* What happens if you set vCore back to stock, or maybe just a smidge above? I ask because most C2D chips aren't very voltage-hungry until you get far above their rated clockspeed, and you ain't really there yet. Some kids are even getting better results undervolting their chips.

* Are you sure the CPU is the bottleneck right now? My gut says it's your RAM or the timings you have set. To test this: in Orthos click the "Test:" pulldown (which defaults to "Blend" I believe) and change it to "Small FFTs" to stress the CPU. Let that run a few hours (assuming no errors), then stop it and switch the pulldown to "Large, in-place FFTs" to stress your RAM. I strongly suspect you'll pass the CPU test with flying colors but get errors on the RAM. If that's the case you still have plenty of tweaking to do, but at least you'll know where to put your effort.

* I know your RAM is rated for 4-4-4-12, but you probably want to loosen it up substantially while you're trying to find your system's OC sweet spot. You said in your OP that you used to be an AMD OCer and you're new to tweaking the C2D. From what I've read, AMD's architecture benefits WAY more from tight mem timings than Intel's. Most benchmarks show C2D systems get relatively slight performance gains (like 4-6%) from tight timings versus loose. So while you're still tweaking, I'd loosen it to 5-5-5-15 or even higher. Once you've found a good stable OC with a high clockspeed that you're satisfied with, you can tighten up the timings somewhat.

This post is already too long so I'm going to leave it at that for now. Hope this helps. Share your results and further questions, and good luck!
 
What do you mean by Double Orthos?

You don't need to run 2 instances of orthos. Why are you doing that?

Is that the reason why orthos failed on the above test?

I know prime95 has to be run 2 times, but orthos is Dual Core ready.

65c is fine since you won't normally push your overclock that high other than when running orthos or another stressing application. 80-85 is supposed to be the bad side.

Also, try a 401 FSB (400 is not a good number for the board)
 
I was under the impression that running two instances of orthos would further stress the cpu for resources at max load? Its interesting though, I've had two instances running at a 3.6ghz, one orthos failed at 7 hours, the other continued running for 16 hours until I stopped it.

This was at a core voltage of 1.51v with a FSB of 400. Temps were around 65c still.

Why do you say 400fsb is not a good number? I was under the impression that it was a sweet spot (when huge gains are noticeable) number like overclocking a 133mhz chip to 166mhz and 200mhz were back in the day.

Does breaking the 400mhz barrier (even stepping to 401) give a better overall system performance more so than say a 375mhz FSB regardless of processor speed? Or is the resulting speed of the processor what is more important to the system performance?
 
Orthos by itself is already a KILLER application and the people who run it for hours are IMO, PARANOID. In reality, most users will not stress a computer 100% on both cores for 8 hours straight.

Anyway, one run of orthos is more than enough. I never heard of anyone trying to run it 2 X the way you were doing it.

THAT MUST BE YOUR PROBLEM. You're really killing your beautiful RIG.

And about the 400 FSB, it is believed by many that the motherboard will work its hardest @ 400 FSB or near that. Supposedly, @ 401+ the new strap will kick in....

In fact, many users cannot get some boards to post between 350 and 399-400.

401+ fixes that!
 
The 680i is an Nvidia chipset, I don't think it has that 400Mhz strap. The 965 based boards do.

Getting a little OT here, but I'm paranoid and run two instances of Orthos, one blend, one small. I'm starting to doubt that practice though, not because Orthos isn't a stressful program, I just wonder what exactly is it stressing - just the FPU or the whole CPU? My thinking is you'd want to run through all of the CPU's instructions - MMX, SSE/2, etc. and make sure they pass. Because Intel/AMD sure as hell aren't running Orthos to test their CPUs at the factory :)
 
theelectic said:
The 680i is an Nvidia chipset, I don't think it has that 400Mhz strap. The 965 based boards do.

Getting a little OT here, but I'm paranoid and run two instances of Orthos, one blend, one small. I'm starting to doubt that practice though, not because Orthos isn't a stressful program, I just wonder what exactly is it stressing - just the FPU or the whole CPU? My thinking is you'd want to run through all of the CPU's instructions - MMX, SSE/2, etc. and make sure they pass. Because Intel/AMD sure as hell aren't running Orthos to test their CPUs at the factory :)
Click to expand...

Do you run them AT THE SAME TIME?

Or separate runs?
 
I've been reading in some other threads that the chipset heatsink/fan unit on the EVGA 680i boards has had issues where an abundance of thermal putty was used to attach the unit straight from the factory.

Could that cause higher temperatures than normal on the cpu, resulting in more vcore needed for stability and more heat being produced?

I really don't want to pull that unit off if I don't have to, but looking at the pictures I've seen from this thread, I kinda concerned that I should....

http://www.evga.com/community/messageboard/topic.asp?TOPIC_ID=24455
 
Well memory = 4-4-4-12 timings 1000mhz @ 2.3v is 7 and half hours memtest86 stable.

Seems like I have decent ram, perhaps I could push it some more, but not sure how much would be too much.

More or less I'm pretty sure its not my memory holding back my overclock after trying a few different timings with my cpu overclock.

Still most stable 3.6ghz oc I've gotten is 1.51v @ 400x9. Still runs at 65c.

I know I've asked many times now, but no one seems to give the same answer. Is 65c too hot? Some people say 80c is too hot and 65c is fine.

I saw some intel spec sheet that said 60-61c is max temp. http://processorfinder.intel.com/details.aspx?sSpec=SL9S8

another here:
http://download.intel.com/design/processor/datashts/31327803.pdf

Also noticed in this pdf that there is a section that says the chip can be installed and removed from the socket 15 times. Page 40 section 3.1.4 what does this mean? Will it break or cease to function after a 15th time?

But then I saw another Intel sheet that confirmed 80c was too hot but never declared between 60-80c was not acceptable. (wish I could find the link)

Others here say 60-70c is fine. Over on evga forums they say 60-80c is fine.

Think 60c or less is where I want to be, guess I will have to clock my processor back down to get an acceptable temp.

Just kinda disappointing when I hear others getting 3.6 easy on 1.4v or less.

Other than that, perhaps reseating, or changing the orientation of the cpu fan might help the cooling issue....
 
Hmmm that does put things a little more at ease with me.....

HardOCP reviewers, and the core temp program seem to think 77c is hot, but not too hot for this chip.

interesting....
 
You must log in or register to reply here.
Back
Top