E6600, can't get stable at 3.6ghz (400x9) FPO#L632B086

Elrein · 17 Feb 2007 at 16:27

Its_Me said:
http://www.xtremesystems.org/forums/showthread.php?t=107030

Thanks

Cir108 · 17 Feb 2007 at 22:04

".... but my load temps were around 70C!"

The weak point in your setup is the cooling and 70C says it all. Within practical designed limits, it is the heat that kills, not so much the voltage. The voltage is merely the “driving force” whereas the heat, a result of due physical processes within the electronics/electrics translated to inefficiencies, will build up to the level which will destroy/destablise the energising item [CPU circuitry in this instance]. When humans get too hot, it also spells trouble as well as machinery. The evacuation of heat [to maintain the inefficiency of the process] was not adequate to maintain balance/stability. The fact that Orthos Blend fails so predictably over a long term confirms that any niggling change [due to unforseen changes as the test progressses] will push the overall system stability “over the edge”. Will extreme cooling @ extreme frequency be possible over an extended period? Not even Intel knows because the Error Probablity envelope is beyond theoretical predicitions. Only practical testing will tell … on a CPU by CPU basis. That’s the con in “Overclocking-to-the-Max”. It’s all about the luck of the draw. Pouring in money to find your CPU limits. Intel is extremely happy about their hobbist CPU limit testers.

jbloggs · 17 Feb 2007 at 22:40

smsmasters said:
Anyway, I upped the vcore to 1.5125V (1.45V droop under 100% load) and I've been running orthos for 20mins so far...

Here's a screenshot. I think 1.5125V should be safe? ~70C load.

I would suggest trying to do something to lower that temperature, I think it is too high!

The Laughing Ma · 18 Feb 2007 at 01:42

6 hour Orthos test to test my undervolt stability which it passed fine only for it to then go and lock up whilst.... get this.... creating a short cut for the new version of CPU-z.

Says it all really.

smsmasters · 18 Feb 2007 at 01:50

I decided to stay at 356x9 1.3V stable to keep my temps down.

Idle 33C
Load 54C

smsmasters · 20 Feb 2007 at 13:12

I found out my side intake fan was blowing air out instead of in.

Now my temps at load are 62-64C as reported by TAT when running orthos.

Idle is 30C with EIST.

I also did the pencil mod and the cpu core voltage is a steady 1.43V at idle and load.

I guess I'll be sticking with 400x9 (3.6ghz) @ 1.43V.

bfar · 20 Feb 2007 at 16:31

smsmasters said:
I found out my side intake fan was blowing air out instead of in.

Now my temps at load are 62-64C as reported by TAT when running orthos. Idle is 30C with EIST.

I also did the pencil mod and the cpu core voltage is a steady 1.43V at idle and load.

I guess I'll be sticking with 400x9 (3.6ghz) @ 1.43V.

Glad u found your sweet spot dude

Just a thought - they say 400fsb is where the NB is stressed highest on that mobo. Have you tried 401 as a matter of interest?

smsmasters · 20 Feb 2007 at 16:39

bfar said:
Glad u found your sweet spot dude

Just a thought - they say 400fsb is where the NB is stressed highest on that mobo. Have you tried 401 as a matter of interest?

Yes, I've tried 401 but the memory latency tests in everest were slower. About 60ns compared to 55.1ns.

I've been running orthos for 4 hours now 15/15 rounds with no errors. I think it's pretty much stable enough for my needs.

bfar · 26 Feb 2007 at 14:43

smsmasters said:
Yes, I've tried 401 but the memory latency tests in everest were slower. About 60ns compared to 55.1ns.

I've been running orthos for 4 hours now 15/15 rounds with no errors. I think it's pretty much stable enough for my needs.

Yes the latancy test would be faster at 400fsb - you're probably on a 1333mhz strap, and the NB is overclocked as high as it can go. If you go >400fsb, the strap defaults back to 1066mhz.

I couldn't get 4GB running stable at 400fsb, but all is fine at 401. So the memory controller is running close to stock. I think in my case the extra mem makes up for it anyways. But bear in mind you're really pushing your NB, which might be why you had problems at 3.6

Starfall · 27 Feb 2007 at 11:27

Cir108 said:
The weak point in your setup is the cooling and 70C says it all. Within practical designed limits, it is the heat that kills, not so much the voltage. The voltage is merely the “driving force” whereas the heat, a result of due physical processes within the electronics/electrics translated to inefficiencies, will build up to the level which will destroy/destablise the energising item [CPU circuitry in this instance].

When humans get too hot, it also spells trouble as well as machinery. The evacuation of heat [to maintain the inefficiency of the process] was not adequate to maintain balance/stability. The fact that Orthos Blend fails so predictably over a long term confirms that any niggling change [due to unforseen changes as the test progressses] will push the overall system stability “over the edge”.

Will extreme cooling @ extreme frequency be possible over an extended period? Not even Intel knows because the Error Probablity envelope is beyond theoretical predicitions. Only practical testing will tell … on a CPU by CPU basis. That’s the con in “Overclocking-to-the-Max”. It’s all about the luck of the draw. Pouring in money to find your CPU limits.

Intel is extremely happy about their hobbist CPU limit testers.

Nice post

What is your background? Thermo? EEng?

The Asgard · 27 Feb 2007 at 20:00

Using a 632B here and mine runs rock solid with 1.46 volts at 3.6 orthos stable. 1600/800 1T 4-4-4-12 Synch P5N32 E SLI 680

Most 6600 seem to do 3.6 now but the key is doing 3.6 at low volts. Still waiting to find a good one after selling my gem to Jay, lol.

Still working well m8 ??

jongeeone · 27 Feb 2007 at 20:13

I just cant get my head around why some people buy more then 1 chip because they cant get a decent clock when 2.8ghz is plenty to start off with. its mental!

smsmasters · 27 Feb 2007 at 20:25

The Asgard said:
Using a 632B here and mine runs rock solid with 1.46 volts at 3.6 orthos stable. 1600/800 1T 4-4-4-12 Synch P5N32 E SLI 680

Most 6600 seem to do 3.6 now but the key is doing 3.6 at low volts. Still waiting to find a good one after selling my gem to Jay, lol.

Still working well m8 ??

My cpu is running well. So fast. I can encode a whole album in 20 seconds to ogg format. I'm simply in awe at the power.

I found the lowest stable voltage for my e6600 @ 3.6ghz is 1.43V. Anyone got any ideas how to make it stable at lower voltages?

Remember: "With great power comes great responsibility."

smsmasters · 27 Feb 2007 at 20:26

jongeeone said:
I just cant get my head around why some people buy more then 1 chip because they cant get a decent clock when 2.8ghz is plenty to start off with. its mental!

It's because in this world we live in, some people are eager to get as much as they can for what they pay for.

Shocky-FM · 28 Feb 2007 at 13:22

jongeeone said:
I just cant get my head around why some people buy more then 1 chip because they cant get a decent clock when 3.2ghz is plenty to start off with. its mental!

Some people are just that pathetic... They will buy 2, 3 CPU's and keep the best clocking chip then brag on the forums how easy they are to overclock...

bfar · 1 Mar 2007 at 10:56

smsmasters said:
My cpu is running well. So fast. I can encode a whole album in 20 seconds to ogg format. I'm simply in awe at the power. I found the lowest stable voltage for my e6600 @ 3.6ghz is 1.43V. Anyone got any ideas how to make it stable at lower voltages?

Remember: "With great power comes great responsibility."

Better cooling.

marscay · 1 Mar 2007 at 11:49

smsmasters said:
My cpu is running well. So fast. I can encode a whole album in 20 seconds to ogg format. I'm simply in awe at the power. I found the lowest stable voltage for my e6600 @ 3.6ghz is 1.43V. Anyone got any ideas how to make it stable at lower voltages?

Remember: "With great power comes great responsibility."

only option = find a better processor, there's not really anything you can do bar exotic cooling.

3.6ghz @ 1.43 you should be pretty happy with that.

not as nice as my 6600 @ 3.7 1.38v but hey you're not far away

Nelly · 1 Mar 2007 at 12:37

marscay said:
only option = find a better processor, there's not really anything you can do bar exotic cooling.

3.6ghz @ 1.43 you should be pretty happy with that.

not as nice as my 6600 @ 3.7 1.38v but hey you're not far away

dont suppose you have the batch# for that E6600 by any chance, thats awesome!

did you try for 3.8ghz? if so what voltage did it take. . . ?

smsmasters · 1 Mar 2007 at 19:02

marscay said:
only option = find a better processor, there's not really anything you can do bar exotic cooling.

3.6ghz @ 1.43 you should be pretty happy with that.

not as nice as my 6600 @ 3.7 1.38v but hey you're not far away

What VID does coretemp say for your cpu?

Mine says 1.325V VID, I guess your cpu has a lower voltage at default.

marscay · 2 Mar 2007 at 15:41

Nelly said:
dont suppose you have the batch# for that E6600 by any chance, thats awesome! did you try for 3.8ghz? if so what voltage did it take. . . ?

L627B156 most 27&28 Rev B chips i've seen are great clockers.

haven't found the max yet been waiting to finish building the rig.