Search the archive:
YaBB - Yet another Bulletin Board
 
   
 
Page Index Toggle Pages: 1
Send Topic Print
Heat dissipaton vs voltage - 4.11 GHz (Read 1151 times)
Jun 9th, 2011 at 5:26am

Speed of flight   Offline
Colonel
Chasing the elusive "faster
than yesterday" goal.

Gender: male
Posts: 150
*****
 
So after a while of being happy at my OC yield, I have a question for a knowledgeable person...
I've run a trusted utility to monitor temps at 100% CPU load. Never gets above 50*-51*. I'm @ 1.515V, with a 50mv boost @ full load (jumping it to 1.615+).

I know that I'm already at a limit of sorts, but I wonder if raising the voltage further (considering the temps) will have detrimental consequences with only heat dissipation, or if there are other concerns, like the transistor material not liking voltages this high or above? other considerations?

So my question is, is the voltage really limited by the ammount of heat that can be dissipated, or are there other things I should be worried about?

Ultimate point: can I go even further? Cheesy (evil "mad scientist" laugh)

Parts/specs in my signature, and any specific problems someone has seen or heard of will be greatly appreciated...
 

Asus Crosshair V Formula-Z
AMD 8350 @4.65 GHz on H100i (226.8 x 20.5)
8 GB DDR3 1814 MHz CL8
ATI 6870 HD Radeon 1 GB
Antec 850 W PSU
Cooler Master HAF 932
500 GB and 200 GB HDDs
Windows 7x64
VRS F/A-18E Superbug, PMDG 747-400 & -8 and MD-11, Captainsim 777, Iris F-14A&B and A-10, Area 51 C-5M Super Galaxy and C-17, loads of others.
IP Logged
 
Reply #1 - Jun 9th, 2011 at 6:31am
NNNG   Ex Member

 
Increasing the CPU voltage increases the temperatures of both the CPU and the components on the motherboard that provide power to the CPU. Even if you have adequate CPU cooling then increased voltage will mean more strain on the power regulation components on the motherboard. My CPU for example, can handle a lot more than 3.2ghz (Q6600s are known for lasting for years at 1.5v, 3.6ghz), but I'm afraid the motherboard won't be able to handle it. Increased voltage will also increase detrimental effects on the CPU such as electromigration, this still occurs at low temperatures but as far as I know high temperatures make it worse. I really don't know if there's a specific voltage that if exceeded will kill your CPU instantly or very quickly, but I do know that increased voltage will make the CPU age faster. It might be possible that there is such a point though, since I know some 2nd generation Core 2 based chips would die quickly if the voltage was even a little bit outside of a range. I don't know much about AMDs 45nm SOI process so I cannot give you anything specific to your chip.

http://www.youtube.com/watch?v=OnbBSiXyFHw&feature=related

As far as I know, 4.11ghz is impressive for a Phenom II 965 and you are unlikely to get much more out of it since, as far as I know, going further would need a lot more voltage for little gain. Also I would steer clear of using features that increase the voltage under load or things like Load-Line-Calibration (LLC) unless you know exactly what they do because there has been some concerns about voltage spikes using such features: http://www.anandtech.com/show/2404/5 (the link is for an intel processor but as far as I know AMDs are pretty much the same for the purposes of the discussion).

Of course, you could go overclock as far as you can just for the lols and then replace it with a new AMD 8 core (when those are released later this year) or Core i7 2600K  if you kill it.

If you want a better answer then you're best off asking someone who either has an Electronic Background ( like me in a couple of years Tongue ), or asking someone who is intimately familiar with overclocking Phenom IIs.
« Last Edit: Jun 9th, 2011 at 9:50am by N/A »  
IP Logged
 
Reply #2 - Jun 9th, 2011 at 3:50pm

Speed of flight   Offline
Colonel
Chasing the elusive "faster
than yesterday" goal.

Gender: male
Posts: 150
*****
 
Quote:
Increasing the CPU voltage increases the temperatures of both the CPU and the components on the motherboard that provide power to the CPU. Even if you have adequate CPU cooling then increased voltage will mean more strain on the power regulation components on the motherboard. My CPU for example, can handle a lot more than 3.2ghz (Q6600s are known for lasting for years at 1.5v, 3.6ghz), but I'm afraid the motherboard won't be able to handle it. Increased voltage will also increase detrimental effects on the CPU such as electromigration, this still occurs at low temperatures but as far as I know high temperatures make it worse. I really don't know if there's a specific voltage that if exceeded will kill your CPU instantly or very quickly, but I do know that increased voltage will make the CPU age faster. It might be possible that there is such a point though, since I know some 2nd generation Core 2 based chips would die quickly if the voltage was even a little bit outside of a range. I don't know much about AMDs 45nm SOI process so I cannot give you anything specific to your chip.

http://www.youtube.com/watch?v=OnbBSiXyFHw&feature=related

As far as I know, 4.11ghz is impressive for a Phenom II 965 and you are unlikely to get much more out of it since, as far as I know, going further would need a lot more voltage for little gain. Also I would steer clear of using features that increase the voltage under load or things like Load-Line-Calibration (LLC) unless you know exactly what they do because there has been some concerns about voltage spikes using such features: http://www.anandtech.com/show/2404/5 (the link is for an intel processor but as far as I know AMDs are pretty much the same for the purposes of the discussion).

Of course, you could go overclock as far as you can just for the lols and then replace it with a new AMD 8 core (when those are released later this year) or Core i7 2600K  if you kill it.

If you want a better answer then you're best off asking someone who either has an Electronic Background ( like me in a couple of years Tongue ), or asking someone who is intimately familiar with overclocking Phenom IIs.


Good info, and thank you for your prompt response.

I really don't need to throttle it further, but I am a limit pusher, and that's pretty much the only reason I ask.

You really did give me the level of info I was looking for. If I push it any further, I'll keep everyone posted. I don't think I'll see a gain in performance, but that's not the only reason we do this, is it... Wink

I recently re-arranged a few things, and replaced the heat-sink fans that seem to keep the thing very cool, and I can't help but think that I can try to give it a little more. But in light of this new info, I may not. Dunno.

Good luck with the electronics classes. It should be a blast! I started an aeronautics school last year, and am training to be an A&P mechanic (airframe and powerplant). Electronics, as they apply to avionics, was quite an involved subject that we only touched on. I found it quite involved, and if you want to go that way, brush up on the math. Freaked me out! It has been nearly 20 years since high-school math. Eek! You seem a very intelligent guy, though. shouldn't be that big a deal, and good luck! Cool

Again, thank you.
 

Asus Crosshair V Formula-Z
AMD 8350 @4.65 GHz on H100i (226.8 x 20.5)
8 GB DDR3 1814 MHz CL8
ATI 6870 HD Radeon 1 GB
Antec 850 W PSU
Cooler Master HAF 932
500 GB and 200 GB HDDs
Windows 7x64
VRS F/A-18E Superbug, PMDG 747-400 & -8 and MD-11, Captainsim 777, Iris F-14A&B and A-10, Area 51 C-5M Super Galaxy and C-17, loads of others.
IP Logged
 
Reply #3 - Jun 10th, 2011 at 3:41am

Speed of flight   Offline
Colonel
Chasing the elusive "faster
than yesterday" goal.

Gender: male
Posts: 150
*****
 
So as it sits, I remember how hard a time I had pushing the thing to where it is now. Mainly, the difficulty lies with me telling the MB to go ahead and give the CPU 1.65V, and with a 50mV load line calibration, gets it to 1.7V.

Here's the problem with me. The temps hover at 53 max. It takes AIDA64 and super pi @ 8million digits to get that temp. Otherwise, it goes to 51 -52, and stays there.

Even if that isn't the only concern, dagnabbit, I can't help but think it can do more!!!

But I digress. I think that this is the limit for this hardware. I moved the fans from the uncontrolled 4 prong connector to the MB chassis and optional fan sockets for better fan control. Asus had a great utility for Q-fan control that allows for higher fan speed @ higher temps. Try as I might, the MB temp ONLY goes down when under load (at least for where the sensor is...)

I'm callling it phase 2. So far at 4.16, and V@ 1.65. Getting a little scary...
Once I had it to 4.2, but the volts were way out of control, and it wouldn't bench. So, I'll keep trying other things, and see what I get. Wish me luck.
 

Asus Crosshair V Formula-Z
AMD 8350 @4.65 GHz on H100i (226.8 x 20.5)
8 GB DDR3 1814 MHz CL8
ATI 6870 HD Radeon 1 GB
Antec 850 W PSU
Cooler Master HAF 932
500 GB and 200 GB HDDs
Windows 7x64
VRS F/A-18E Superbug, PMDG 747-400 & -8 and MD-11, Captainsim 777, Iris F-14A&B and A-10, Area 51 C-5M Super Galaxy and C-17, loads of others.
IP Logged
 
Reply #4 - Jun 11th, 2011 at 12:03am

Speed of flight   Offline
Colonel
Chasing the elusive "faster
than yesterday" goal.

Gender: male
Posts: 150
*****
 
So @ 4.16, I'm @ 1.65, and with load line enabled, 1.7+.
If I cook the processor, I can replace it for $200+-. I'm not gonna keep it there, because it won't bench at any higher settings. I remember @ 4.2, windows wouldn't even start.

Here, I played BFBC2 for 4 hours, and put the crunch on it with aida. it finally blue screened, but that's a hefty load.
I'll sit @ 4.11 and call it good.
 

Asus Crosshair V Formula-Z
AMD 8350 @4.65 GHz on H100i (226.8 x 20.5)
8 GB DDR3 1814 MHz CL8
ATI 6870 HD Radeon 1 GB
Antec 850 W PSU
Cooler Master HAF 932
500 GB and 200 GB HDDs
Windows 7x64
VRS F/A-18E Superbug, PMDG 747-400 & -8 and MD-11, Captainsim 777, Iris F-14A&B and A-10, Area 51 C-5M Super Galaxy and C-17, loads of others.
IP Logged
 
Page Index Toggle Pages: 1
Send Topic Print