Core i7/X58 Overclocking Thread

[ReverendMaynard]

3 years? I dont plan to keep my current Ci7 longer then one year (westmere release)

I'm on a 3 month cycle and have been since the 939 Opty's

[SteveRo]

quote - All the new Low Voltage DDR3-2000+ CL7 Ram coming out operates at 1.65 QPI Volts. Along with 1.65 Dram volts.

Both Corsair and G. Skill said let the QPI volts fly, but just watch your NB Temps.

These Ram Modules will not Operate at their specs unless the QPI is at least 1.55v

I'm running the G. Skill Now at DDR3-2050 at CL8

Kensek - post the high v memory info you found -triple channel I assume?
Can we find some folks that are using this and track how they work out ...

[SteveRo]

double post - sorry

[Tobit]

I'm on a 3 month cycle and have been since the 939 Opty's

Heh, I'm upgrading to an i7 from a Celeron 331.

[ReverendMaynard]

Heh, I'm upgrading to an i7 from a Celeron 331.

the increase in performance is going to hit you allot harder than it hit me lol.

[Kensek]

quote - All the new Low Voltage DDR3-2000+ CL7 Ram coming out operates at 1.65 QPI Volts. Along with 1.65 Dram volts.

Both Corsair and G. Skill said let the QPI volts fly, but just watch your NB Temps.

These Ram Modules will not Operate at their specs unless the QPI is at least 1.55v

I'm running the G. Skill Now at DDR3-2050 at CL8

Kensek - post the high v memory info you found -triple channel I assume?
Can we find some folks that are using this and track how they work out ...

Already posted in various other Threads, but here is the G. Skill response.



Dear customer



All DDR3 2000 module requires QPI Voltage 1.55v to run.

You need to have good air flow on NB chipset. 1.35v is way too low, although it is the recommend the safe voltage.

As long as NB chipset temp is not too hot, it is ok.



Thank you

GSKILL SUPPORT

[Splave]

^so you can send your CPU and or Mobo to G.skill when it gets burnt?
I assume it cant be that bad, Ive been to 1.55 QPI.vtt no problems

[Kensek]

The new Corsair DDR3-2000 GT that runs at 7-8-7-20 takes 1.65v of QPI.

[dejanh]

Already posted in various other Threads, but here is the G. Skill response.



Dear customer



All DDR3 2000 module requires QPI Voltage 1.55v to run.

You need to have good air flow on NB chipset. 1.35v is way too low, although it is the recommend the safe voltage.

As long as NB chipset temp is not too hot, it is ok.



Thank you

GSKILL SUPPORT

I can so see this happening...

Operator: "Hi, welcome to GSkill customer support, my name is Ashley, how may I help you?"
Customer: "Well, I have your DDR3 2000MHz modules and I was running my QPI according to your specifications but I seem to have burnt my CPU as my computer no longer boots."
Operator: "Oh, I understand. But how is this related to our memory?"
Customer: "Well, your instructions said to run the QPI at 1.55V so I did and I think my CPU may be fried."
Operator: "I understand, but unfortunately as there is no problem with the memory this is outside of our warranty policy so we cannot do anything about it."
Customer: "@#$%$# @#$#@#, I was told to run the QPI at 1.55V!"
Operator: "Yes sir, I understand, but the problem is not with the memory. I am really sorry."

Then after you contact Intel and say, "my chip burnt" and you tell them you used the "1.55V QPI DDR3 2000MHz memory" and they will say "sorry, that is not supported by our chip." Good luck getting your warranty honored. IMHO, just one reason no to play with fire because when you do you usually get burnt.

[Hot'n'Chilly]

3.4ghz now (tested stable on Prime for a few minutes just a bit ago).

CPU Ratio: 17
BCLK: 200
DRAM: 1203mhz
UCLK (UnCore): 3008mhz
QPI: 7218 MTs

CPU Voltage: 1.2375v
PLL Voltage: 1.9c
QPI (unCore) Voltage: 1.3625v
DRAM Voltage: 1.62v

I tried the 18x CPU Multi Ratio but it would blue screen out of Prime95, so I had to go back. I suppose I'm stuck here until I think of a way to crank something up without crashing. Any ideas?

[Extelleron]

3.4ghz now (tested stable on Prime for a few minutes just a bit ago).

CPU Ratio: 17
BCLK: 200
DRAM: 1203mhz
UCLK (UnCore): 3008mhz
QPI: 7218 MTs

CPU Voltage: 1.2375v
PLL Voltage: 1.9c
QPI (unCore) Voltage: 1.3625v
DRAM Voltage: 1.62v

I tried the 18x CPU Multi Ratio but it would blue screen out of Prime95, so I had to go back. I suppose I'm stuck here until I think of a way to crank something up without crashing. Any ideas?

Raise your CPU voltage..... 1.2375V is not enough for your chip @ 3.6GHz. What kind of cooling do you have / how are temps?

You should really run either LinX in 64-bit mode for a few hours or Prime95 for 12hrs+ to declare a system stable. Running Prime95 for a few minutes is useless in determining stability in my experience. Prime will run stable for 30 minutes yet LinX will produce BSOD in a minute or so.

Not related to you but the voltage scaling on these CPUs (at least mine) really sucks...... I can get 3.8GHz 8-hr LinX stable at 1.264V as I posted earlier today, yet LinX produces BSOD within minutes when running 4GHz @ 1.36 volts. This 920 will simply not do 4GHz whatsoever.

[Hot'n'Chilly]

I'm using a Noctua NH-U12P air cooler. My temps are good so far, 34-40 on RealTemp.

About raising voltage though: I thought we weren't supposed to raise CPU Voltage above 1.25v so I backed off. Would the full 1.25v be enough to get me there, or will it be one of those deals where I go a bit above?

[Extelleron]

I'm using a Noctua NH-U12P air cooler. My temps are good so far, 34-40 on RealTemp.

About raising voltage though: I thought we weren't supposed to raise CPU Voltage above 1.25v so I backed off. Would the full 1.25v be enough to get me there, or will it be one of those deals where I go a bit above?

If you want the processor to last 10-20 years running constantly, then perhaps 1.25V is a good limit. But really there is not going to be danger in raising the voltage above that if you plan on using the processor for its usable life span. Raising the voltage and increasing frequency are going to reduce the life span of the CPU, perhaps significantly, but it is highly unlikely that you will actually be using the CPU by the time it would die anyway.

I would recommend that you not exceed 1.40V on air cooling; as long as you stay below that you should be fine. I would try skipping to 3.8GHz and see what you can do there. Hopefully you will be able to get that stable w/ under 1.3V at load. Unless you have a golden chip 4GHz is going to be out of your reach without insane voltage.

[nordien]

I am happy so far. The Core i7 920 is from the 3837A507 batch. Keeps rather cool.



The next step is bclk above 205.....

[Chri$ch]

Try the same with HT on

btw i got today this one..

[nordien]

[QUOTE=Chri$ch;3538302]Try the same with HT on

btw i got today this one..

Need to get Watercooling first.....

[Hornet331]

Try the same with HT on

btw i got today this one..

Not bad, but its on WC right?

[Hot'n'Chilly]

If you want the processor to last 10-20 years running constantly, then perhaps 1.25V is a good limit. But really there is not going to be danger in raising the voltage above that if you plan on using the processor for its usable life span. Raising the voltage and increasing frequency are going to reduce the life span of the CPU, perhaps significantly, but it is highly unlikely that you will actually be using the CPU by the time it would die anyway.

I would recommend that you not exceed 1.40V on air cooling; as long as you stay below that you should be fine. I would try skipping to 3.8GHz and see what you can do there. Hopefully you will be able to get that stable w/ under 1.3V at load. Unless you have a golden chip 4GHz is going to be out of your reach without insane voltage.

I just got 3.6ghz off a CPU Voltage of 1.25v, heres the details.

CPU Ratio: 18x
BCLK: 200
DRAM: 1203
UCLK: 3008
QPI: 7218

CPU Voltage: 1.25v
PLL Voltage: 1.9v
QPI Voltage: 1.3625v
DRAM Voltage: 1.62v


Its stable after a few passes of Prime (haven't had time for extended Prime testing) and games fine. Any suggestions for where I should go from here? Maybe just keep cranking the CPU Voltage up while trying to hit 20x200? Can I count on hitting 20x200 (4ghz) on after-market air or would that be pushing my luck???

[ReverendMaynard]

Try the same with HT on

btw i got today this one..

close to my batch (can't really make yours out though).

B = golden. Been telling people this for weeks now lol.

[Chri$ch]

i have now 2 B batch CPUs..

3837B072 = 4Ghz @ 1.264v
3838B079 = 4Ghz @ 1.248v (try now lower vcore)

[ReverendMaynard]

i have now 2 B batch CPUs..

3837B072 = 4Ghz @ 1.264v
3838B079 = 4Ghz @ 1.248v (try now lower vcore)

How about 4.2ghz HT on? 4ghz is child's play lol.

[theonlybabyface]

How about 4.2ghz HT on? 4ghz is child's play lol.

I agree. Is this with HT and Turbo? Anyone can turn off the features and get 4+ Ghz at low volts....

[bench2die]

How about 4.2ghz HT on? 4ghz is child's play lol.

I like 4.3 (those settings would not pass prime lol it passed it at 4.2 but I upped the voltage some and it passed prime but temps are WAY to high.)

[dejanh]

i have now 2 B batch CPUs..

3837B072 = 4Ghz @ 1.264v
3838B079 = 4Ghz @ 1.248v (try now lower vcore)

Damn man, while I do have a high voltage chip even for A batch, your results are just nuts. I need 1.44V to be stable at 4.0GHz. Pathetic

I did not have an option to cherry-pick chips though...

I agree. Is this with HT and Turbo? Anyone can turn off the features and get 4+ Ghz at low volts....

He has all threads on. That's what makes it impressive.

[ReverendMaynard]

I like 4.3 (those settings would not pass prime lol it passed it at 4.2 but I upped the voltage some and it passed prime but temps are WAY to high.)

Nice, now get that thing under a cascade where all 965's belong