Core i7/X58 Overclocking Thread

[Elektrosadist]

we'll see. My setup is i think very good and im absolutly positive that i can keep my low temps. I have so much water in my circuit and such a big resevoir plus huge radiator that i think it can take it.

But heck i'm up for whatever comes. And if the I7 C0 is too hot for me i will directly put it on ebay and get a D0.

[The Water Dog]

Elektro: D0 can be hotter then C0 just so you know, D0 does use less volt but some of they are made in hell i think:P

[Chicken Patty]

Elektro: D0 can be hotter then C0 just so you know, D0 does use less volt but some of they are made in hell i think:P

I have also noticed that. Less voltage, but run hotter

[Formann]

Too much D0 talk here now

920 (3841A768) on air, Cooler Master Hyper212



I did a run @ 4620Mhz, but Paint.NET crashed when I tried to save the screenshot.

[Formann]

I have also noticed that. Less voltage, but run hotter

Since the beginning of time, all my best chips have ran hotter than the poor OC´ing ones.

More leakage = more heat = higher clockrates

[YMAA]

COOLING : Water single loop - D-tek Fusion, Swiftech 355, swiftech 1dual + 1single rad, 4870 in same loop with EK Supreme block, xspc restop, 3/8 tubing

How cold is it where you live? Your ambient temps must be insanely cold to get that low of a temperature on 360mm thin radiator with a 4870 in the same loop. What, do you dump the rads in ice water?

[ReverendMaynard]

I have also noticed that. Less voltage, but run hotter

this is because the D0's are actually using the voltage rather than leaking it.

[Sniper]

How cold is it where you live? Your ambient temps must be insanely cold to get that low of a temperature on 360mm thin radiator with a 4870 in the same loop. What, do you dump the rads in ice water?

If you look at the 4.5 screenshot, you'll see that he has turbo throtling witch will keep the load and temp down quite a bit.

[Anth Seebel]

new PB for max validation, single stage, 1.49v , HT Enabled

[glide 1]

If you look at the 4.5 screenshot, you'll see that he has turbo throtling witch will keep the load and temp down quite a bit.

yes this is true, my multi's keep throttling between 20 -21 while running prime keeping the temps/load from shooting up. Room is +- 22-24deg

[glide 1]

How cold is it where you live? Your ambient temps must be insanely cold to get that low of a temperature on 360mm thin radiator with a 4870 in the same loop. What, do you dump the rads in ice water?

Room temp was +- 22-24deg when i ran the test with ac vents blowing towards case. As sniper mentioned above in realtemps screenshot, the system was throttling 20 -21 keeping temps from shooting up too much. Mid 70's max temps is still pretty good though, i think cold ambient helps a lot.

[rge]

Ambients are definitely key on i7, cant wait til winter when I can bench mine with cool ambients. Though using A/C vent works too, it can get you down to teens in C.

You can pretty much gauge ambients by looking at RT minimum which usually = idle temp. Load TDP varies significantly chip to chip, idle watts however is similar just because 10-20% difference at idle is still only a few watts.

Stock, idle, power savings, undervolted the best you can get is about 6-8C over ambient on i7 with water, or ambient is RT minimum temp - 6-8C on most.

At 4.2 and 1.25vcore you will be idling at ~25W, and gradient will be >12C over ambient, or ambient is at least 12-14C below RT minimum temp.

At 4.5 and 1.42Vcore idle will be at ~50W and core temps at least 16C over ambient, or ambient is at least 16C below RT minimum temp.

Does not work perfectly as above in all...but it will be reasonably close.

[Chicken Patty]

Ambients are definitely key on i7, cant wait til winter when I can bench mine with cool ambients. Though using A/C vent works too, it can get you down to teens in C.

You can pretty much gauge ambients by looking at RT minimum which usually = idle temp. Load TDP varies significantly chip to chip, idle watts however is similar just because 10-20% difference at idle is still only a few watts.

Stock, idle, power savings, undervolted the best you can get is about 6-8C over ambient on i7 with water, or ambient is RT minimum temp - 6-8C on most.

At 4.2 and 1.25vcore you will be idling at ~25W, and gradient will be >12C over ambient, or ambient is at least 12-14C below RT minimum temp.

At 4.5 and 1.42Vcore idle will be at ~50W and core temps at least 16C over ambient, or ambient is at least 16C below RT minimum temp.

Does not work perfectly as above in all...but it will be reasonably close.

RGE, can you explain the above in more simple words. not familiar with what you mean by gradient. Maybe thats what is throwing me off.

[rge]

Gradient refers to the gradual temperature difference from hot core to the ambient (or IHS, etc). These relationships can be calculated by intel supplied formulas, or can be measured. The gradient is dependent on power, the higher the power, the higher the gradient or temp differential between hot core and ambient.

Here is thermal imaging of gradient from hot core to cooler area, see figure 2.
http://domino.watson.ibm.com/comm/re...nnovation.html

And here is a pic (from stanford paper) of sensor graph of pentium showing gradient or temp differential, in this case from hot spot to ambient (38C) on pentium at ~78W load.


Intel defines this gradient with a formula based on TDP. Higher watts, the higher the gradient or temperature difference between hot core and ambient.

At even 3-6 watts, undervolted, idle, underclocked, there will be still at least a 6-8C difference or gradient between hot core and ambient, for example 28-30C core would be ~ 22C ambient, given excellent cooling/water or very high end air/fans.

When you get up to ~45+ watts, for example core i7 D0 at idle, 1.4 vcore, 4.4+ghz, there will be a minimum gradient or temperature difference between hottest core and ambient of at least ~15-18C.

For example if I boot up at 4.5ghz, 1.4 vcore my average realtemp idle temperature of 4 cores is 44C. My ambient intake temps is 27C, so I have a 17C temperature difference (or gradient) from core to ambient, which is typical for 50 Watts. Others maybe ~10-20% difference, ie few C one way or another given similar cooling. But not possible to have for example an 6-8C temp difference between idle core and ambient (gradient core to ambient) for 40-50W.

[unclewebb]

The data sheet for the Core i7 CPUs shows this linear relationship. As the watts go up, so does the Tcase temperature which is the temperature measured at the center of your CPU. This graph is based on the OEM cooler. With after market cooling or water, the slope of the line will be different but the relationship will stay the same. More watts = higher temps.

Intel® Core™ i7 Processor Extreme Edition and Intel® Core™ i7 Processor Datasheet, Volume 1
June 2009 p.73



When running a stress program like Prime 95 Small FFTs, the core temperature can be a good 20C beyond the Tcase temperature.

[Chicken Patty]

thanks guys. Now I understand exactly what you mean

so for example in below screen shot. THe average between all four cores is 42ºc. My ambient on the CPU right now is 29ºc. This means I have a gradient of 13ºc??? this is idle of course.

[rge]

chicken patty, yep you got it, and that is typical ambient to core gradient at that OC.

And thanks unclewebb for the graph, makes it clearer.

[Chicken Patty]

chicken patty, yep you got it, and that is typical ambient to core gradient at that OC.

And thanks unclewebb for the graph, makes it clearer.

woohooo, thanks to you and everyone else who helped and to unclewebb

[ReverendMaynard]

The last 4 posts in this thread are just pure win.

[Chicken Patty]

The last 4 posts in this thread are just pure win.

s**t, I learned something new, something interesting

[TheCool]

These are some insane clocks. Is there a consensus on which boards provide the best results aside from the EVGA classified? It's not that I don't think that board would do a good job, it's just that it's a bit too expensive for my budget.

[Chicken Patty]

These are some insane clocks. Is there a consensus on which boards provide the best results aside from the EVGA classified? It's not that I don't think that board would do a good job, it's just that it's a bit too expensive for my budget.

I've seen some great results with this board believe it or not. over 40k in vantage and dual gtx295's

Only the CPU water cooled, everything else stock air cooling

not bad price either
http://www.newegg.com/Product/Produc...82E16813135229

this one too, seen pretty good results with, only three ram slots, but hey if you get 3x2gb ram sticks, you'll be fine

http://www.newegg.com/Product/Produc...82E16813186163

[ReverendMaynard]

s**t, I learned something new, something interesting

actually I did as well, rge and uncle are top notch peeps.

[Chicken Patty]

actually I did as well, rge and uncle are top notch peeps.

i have heard many times about gradient and temps and stuff, but nobody actually took their time to explain to me. Finally someone did and I appreciate it alot

[glide 1]

The last 4 posts in this thread are just pure win.

i agree, big thanks to unclewebb and rge especially, and to all who have contributed here. i'm slowly understanding all this and the importance of getting good temps with decent clocks....