Gigabyte GA-965P-DS3 v3.3 and E6400

KnightRid · Post by **KnightRid** » Wed Mar 28, 2007 7:20 pm

Ok, I am in heaven right now - as long as it keeps the speed

3GHZ!!!!!!

I am sa happy right now

Now i do have some questions...

Is it better to have the multiplier high and the FSB lowq, or the other way around?

How do I know when I shoudl raise the voltages, and on what should I raise - cpu, fsb, mcu ( mce, something liek that ), etc.

I cheated and used stuff i found on the net and raised the fsb and mcu/mce/mcsomething voltages a little bit, then lowered the fsb to 380 since 400 at 8x would not work

this is just awesome!!!

I really like overclocking when you can hit almost a 1ghz improvement ;)

Mike

Help, tips, etc REALLY wanted!!!

Post by **Apoptosis** » Thu Mar 29, 2007 5:41 am

I'd personally suggest lowering the multiplier and going higher on the FSB... It won't really help the bandwidth, but should help the floating point scores more. It all depends on what you are trying to do.

CPU-Z shows you at 1.33V on the CPU, but what is it set at in the BIOS?

Like you said 3GHz is killer and I'm glad you are reaching a level that you are happy with. Just imagine how happy I was to take quad-core over 4GHz... The performance numbers were insane.

KnightRid · Post by **KnightRid** » Thu Mar 29, 2007 7:36 am

ok will try to lower the multiplier to 6X and raise the fsb when i get some time today or tomorrow!!

right now here are the settings

8X - multiplier
380mhz - fsb
memory multiplier - 2.00
(G)MCH Overvoltage - +0.1V
CPU Voltage - 1.37500V - says normal is 1.32500V

everything else is still set to the default settings

Mike

all help is welcome

I am a newbie OC'er!!!!!! yell at me, tell me I am doing stuff wrong - thats how we all learn ;)

Post by **Apoptosis** » Thu Mar 29, 2007 7:58 am

try 430MHz FSB (1720MHz) with the setting you have now and see if there is any performance improvement from 380MHz with an 8x multiplier.

Both will give you ~3GHz, but run benchmarks at both to see which is best for you.

shadco · Post by **shadco** » Fri Mar 30, 2007 5:17 am

I'm a little surprised that you need that much vcore at that clock speed but then I'm used to 6400's

Mine are set at 1.312 in bios which seems to yield about 1.280 after the droop. I run them at 400x8 for 3.2Ghz

KnightRid · Post by **KnightRid** » Fri Mar 30, 2007 5:45 am

shadco wrote:I'm a little surprised that you need that much vcore at that clock speed but then I'm used to 6400's

Mine are set at 1.312 in bios which seems to yield about 1.280 after the droop. I run them at 400x8 for 3.2Ghz

Should I lower it? I do not have any idea what I am looking for wit hthe voltages - I jsut used what someone else posted they used on a website

Messed with the FSB yesterday still at 8X and got the max I can do - 385FSB 8X about 3.08 and the voltage says 1.328 in cpuz - too high, too low, raise the voltage in bios to get more overclocking?

What program is the best to test between the 8X and 6X - SuperPi?

Might do that today as long as superpi runs on vista

Mike

kenc51 · Post by **kenc51** » Fri Mar 30, 2007 6:51 am

I'd leave the voltage as is for now, then test to see what the max freq. you can get. test with Orthos (blend test) for ~30mins at a time.

Once Orthos fails, reboot and add more voltage.......keep doing this until either it gets too hot or there's too much voltage.

After you find the absolute max, you can then decide what speed to use for 24/7 use and then tweak the memory etc.

btw, I should be switching to C2D very soon too :D

KnightRid · Post by **KnightRid** » Fri Mar 30, 2007 7:57 am

Ok here is the pic of SuperPi calcs and times, with cpuz and easytune

Now - why does cpuz screw up with the multiplier sometimes? It is showing 6X but ti is set to 8x - thats why I had easytune open to show the actual speed.

Ok will try Orthos - figured Super Pi and FAH 24/7 would be good enough

heat is NOT an issue - well at least while i have the side panels off ;) CPU never has gotten above 35C and the system has been no higher than 38C

Well I hope those are good numbers

Mike

So adding more voltage will increase the overclock? so if I increase the voltage to ( just the cpu? ) a higher amount, then I should be able to get a higher fsb?

Newbie isnt a strong enough word for what I am

KnightRid · Post by **KnightRid** » Fri Mar 30, 2007 8:33 am

OK I raised the voltage and now it seems to be stable at 400mhz FSB!! - 3.2ghz!!!!

I saw here... http://www.hardforum.com/showthread.php?t=1138241 that people are using 1.57 voltages!! That seems awful high to me!

I raised it to 1.42500V.

What is the most accurate program to measure the temps on the cpu? Every program I try seesm to give me different numbers, which is real goofy.

Mike

Thanx for all the help so far!!!! But I need plenty more ;) I jsut get so damn impatient I start trying stuff - I just dont want to kill the cpu though!!

that would be REAL bad.

shadco · Post by **shadco** » Fri Mar 30, 2007 9:07 am

I use my stuff for folding, that means close to 100% cpu utilization 7x24 x 365.

I've settled on E6400's set at 400x8 with aggressive ram timings as the sweet spot for my mess. I then work to see how low I can get away with dropping voltages. I also drop voltage on North and Southbridges if the mobo allows and I've been able to run my ram at 1.9V's

Synthetic benchmarks don't mean much to me unless I'm at the beginning of a new setup where I'll use orthos and 3dmark 2001se to shake it out and get a read on temps and volts. Once I've found the mark I switch em over to Linux and put em to work.

Except for my Main Family box which is running xp and is used primarily for Photoshop and general web stuff.

shadco · Post by **shadco** » Fri Mar 30, 2007 9:09 am

KnightRid wrote:OK I raised the voltage and now it seems to be stable at 400mhz FSB!! - 3.2ghz!!!!

What is the most accurate program to measure the temps on the cpu? Every program I try seesm to give me different numbers, which is real goofy.
.

I use TAT Intel Thermal Analysis Tool

http://www.techpowerup.com/downloads/392/mirrors.php

KnightRid · Post by **KnightRid** » Sun Apr 01, 2007 4:57 am

Well I havent lowerd the multiplier as of yet - still trying to get it stable at 8X and 400 fsb.

I had to raise the voltage AGAIN because it was locking up. So far it has run since yesterday with no problems.

TAT program says both cores are around 49C. BIOS never shows them above 30C - I still dont get it.

Well if this voltage setting works for stability ( running smp folding to max the cores ;) ) then I will be trying the lowering multiplier and raising fsb starting later today or tomorrow.

My only concern with that is the memory. I am worried the memory wont handle the higher fsb

I guess I should have asked this before i bought just 800mhz memory huh :p

Mike

I feel all powerful now

If only Ohio State would not have cheated to win :p

shadco · Post by **shadco** » Sun Apr 01, 2007 12:22 pm

TAT measures the core tmps from the cpu. I don't have a clue what Gigabyte is measruing but I don't trust it.

Coretemp is another good tool.

http://www.thecoolest.zerobrains.com/CoreTemp/

KnightRid · Post by **KnightRid** » Mon Apr 02, 2007 5:11 am

So does anyone know how high I can take the voltage?

Can I just keep upping it until the cores get too hot - how hot is too hot anyway?

As I raise the voltage I can also raise the fsb higher - correct?

this could take a while to figure out what the max is with the multiplier at 8X - then I will do it again at 6x and run the prime thingy to see which is faster

Mike

I now know why water cooling is becoming so popular! Especially if you can take the temps down a LOT - at least 10C would be nice ;)

shadco · Post by **shadco** » Mon Apr 02, 2007 9:55 am

KnightRid wrote:So does anyone know how high I can take the voltage?

Can I just keep upping it until the cores get too hot - how hot is too hot anyway?

As I raise the voltage I can also raise the fsb higher - correct?

this could take a while to figure out what the max is with the multiplier at 8X - then I will do it again at 6x and run the prime thingy to see which is faster

Mike

I now know why water cooling is becoming so popular! Especially if you can take the temps down a LOT - at least 10C would be nice ;)

I'm not a fan of High Vcore voltage. I killed a few P4's back in the day and a mobo or 2.

I see regularly where folks run say 1.45 vcore in pursuit of high synthetic benchmarks but I fail to see the benefit in that anymore and I doubt they really run their systems 7x24 under heavy load with High vcore setting.

What's your goal??

Is this a gaming system??

Are you looking to establish some high benchmarks (I'm not judging if you are as I've perused the same) ??

Is this a system meant for folding??

A Video editing or photoshop batch processing system.

bumblyhumbly · Post by **bumblyhumbly** » Mon Apr 02, 2007 10:30 am

IMHO, higher FSB and lower multiplier is usually more stable. Some higher multipliers will bring on frequent system hangs. Some simply won't boot.

FYI - Intel FSB is validated with a +/- 5% on the chipset Vcore and Vtt volatges with chipset temps at 0C and 100C. I believe CPU's are same votages but a much lower tolerance on the temps. Stable FSB beyond 1600mhz with low multi's are possible with the right conditions.

KnightRid · Post by **KnightRid** » Tue Apr 03, 2007 6:05 am

To the question about what the system is for....

First and foremost I want this to be my main rig once i can get some more sata drives.

second i want to fold as fast as I can

third - video editing will be done, but not very much...just normal everyday computer usage.

I am running stable at 3.2ghz with the voltages raised, and the core temp has gone to 52C that I have seen but no higher - now this is with the case sides off! I want to find another 120mm fan before i try with the sides on.

I think I will start to mess with the lower multiplier and higher fsb today - seems i could be fiddling with this thing a LOT longer than I thought to get the max performance.

Mike

I dont want to see the cpu go over 60C, am I being to conservative?

Post by **Apoptosis** » Tue Apr 03, 2007 6:22 am

KnightRid wrote:So does anyone know how high I can take the voltage?

Can I just keep upping it until the cores get too hot - how hot is too hot anyway?

As I raise the voltage I can also raise the fsb higher - correct?

this could take a while to figure out what the max is with the multiplier at 8X - then I will do it again at 6x and run the prime thingy to see which is faster

Mike

I now know why water cooling is becoming so popular! Especially if you can take the temps down a LOT - at least 10C would be nice ;)

On air cooling 1.4-1.425Vcore should be good for daily use as higher than that just gets too hot.

I don't like seeing load temperatures over 63-65C. I have several Core 2 Duo's that start throwing errors and have BSOD when they get hotter than 65C.

When I overclock I max everything out with no voltage increases and once I reach the top I then increase the voltage and keep going up. This way you don't use too much voltage for the overclock you want, which lowers power consumption and reduces heat.

As for reading temps I use Core Temp Beta and TAT, but be careful with TAT as it can and will kill CPU's when run at full load for long periods when overclocked and overvolted.

stereo55 · Post by **stereo55** » Tue Apr 10, 2007 10:35 am

Sweet stuff . I was thinking of grabbing this board too , but Newegg seems to have version 1.3 . Anything drasticly good/bad about the older version(s) I should know about ?

http://www.newegg.com/Product/Product.a ... 6813128042

KnightRid · Post by **KnightRid** » Wed Apr 11, 2007 6:07 am

stereo55 wrote:Sweet stuff . I was thinking of grabbing this board too , but Newegg seems to have version 1.3 . Anything drasticly good/bad about the older version(s) I should know about ?

http://www.newegg.com/Product/Product.a ... 6813128042

This is the one i got - http://www.newegg.com/product/product.a ... 6813128012

I got the revision 3.3 - dont go by the box, actually look at the board if you buy this oine - it is stamped in the corner by the pci slots.

my box said a different revision, but the board is 3.3.

Mike