Is my overclocking Okay so far?

Toggle sidebar Toggle sidebar
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
All my system specs are in sig.

8350 with H100i.

I am at 4.5 Ghz (22.5 Multi)
at 1.4V

I raise multi up by .5 and run prime95 to test for stability, and add 0.0125V until it is stable.

Is this correct so far? My temps so far are maxing out at 36
 
Solution
Jake Wenta


I've already stated them :??:
I gave you waterblocks, fan ideas, thermal paste changing, heatsink changing. There's no miracle in cooling. But if you want to go to extremes, there's always peltier which cools below ambient xD. But that takes alot of studying and safety precautions and another PSU before attempting.
Sort by date Sort by votes
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160


IKR, this H100i is amazing. Some guy said I couldn't OC to 5ghz with it. I'm taking that as a challenge :)
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160


around 22. It goes below that at times.

What is the max voltage for this processor?

I read somewhere its 1.55
 
Upvote 0 Downvote
Kelthar

Kelthar

Honorable
Mar 27, 2013
640
0
11,360
As with all processors, it varies even between same models due to manufacturing.

To be safe I'd stay at 1.5v, although people have mentioned 1.55V on other websites.

If you do a safe overclock increase, little by little, you should notice when you reach the limit. But I wouldn't go as far as 1.55V, personally.
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
Okay, now another question.

If I can't go up anymore in voltage, can I try mixing FSB overclocking with the multiplier?

Also, if I lower multiplier and raise FSB to get same as original Ghz, do I have to change voltage at all?
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160
For some reason I can't believe it's maxing @ 36*C. How long do you run the test-and are you running torture test? Go to options, torture test, make it run Min FFT 4096 and as much RAM as you can. Run it for about 30minutes.
And make sure to NOT note the water temp, core temp BUT the CPU/Socket Temp. (And watch the VRM temperatures, they tend to get hot too-at 70ish degrees Celsius, you should start to think of adding a fan to blow at them)
Also you have LLC at high or higher. As lower than that and you experience Vdroop. The voltage decreases under load. With High, it does slightly, higher it increases the voltage under load. This should help you achieve higher OC's.
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160


FSB or BCLK affects the north bridge. HT, NB, RAM, CPU frequencies increase. So increasing the bus you'll have to check RAM speed, probably adjust ram timings and voltage. Adjust NB and NB frequency. Add CPU/NB and NB voltages. And maybe RAM voltage, if you will increase its frequency.
Best way to start OC'ing is finding your highest stable Bus speed first, then RAM and then play with the multiplier. (Just from my experience)

This may help too:
http://www.overclockers.com/forums/showthread.php?t=688663
Remember to mark benchmark ratings, as higher OC doesn't always mean better performance.
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160



Maybe yes, maybe no. Only you will know. Running the torture test maxed for 30+minutes should give you an idea if it is stable. I can not tell you whether or not it needs more voltage.
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160


Ok. I was wondering if in general if it needed it be changed from the get-go.
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160


Only your stability will justify that. You run the torture test to test stability and thermal limits. If it fails you'll need to alter things. For example, adjusting a multiplier usually means increasing the CPU voltage. Increase RAM frequency or tightening timings may require additional RAM voltage and CPU/NB (pretty much the memory controller for AMD-VCCIO for Intel). Increasing NB frequency you may require higher NB voltage (not same as CPU/NB). High Bus will meaning adjusting multipliers, frequencies, possibly timings, and the previously stated voltages. It all depends what you do. Don't forget to increase the LLC to prevent vdroop. High tends to be pretty close under load as idle, and very high and extreme increase the voltage dramatically underload. So take this into consideration. Here's a nice post about it and how it affects it-ddiffers from motherboard to motherboard of course:
http://linustechtips.com/main/topic/24019-load-line-calibration-why-overclockers-should-care/

His high has voltage higher under load than idle, where on my motherboard LLC on high drops my voltage by about .02 under load. CPU-Z, it'll tell you the voltage, then start Prime95 and you'll see the voltage change.
Forgot to note, OpenHardwareMonitor is a good program for temperatures. It'll tell you the lowest, current and highest temperatures achieved. (There's also graphing and stuff :))
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
Ok, I have another question. It is stable at 4.7Ghz now. 22.5x 210
I have llc at 75%[low]My motherboard is backwards.

Bios Voltage: 1.475


Idle voltage: 1.452/1.464
Load voltage: 1.416/1.404

Now about the VRM's. I have a fan pointed at them. The heat sink reads upper 50's to 60 when hit with a heat gun. How do I tell exact what they are and what is safe?
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
Here is a SS after 30 sec of prime95 (the max was from a longer test)




The VR T1 and T2 I think are the VRM's and north bridge, since I have the rev. 4 motherboard and it has those heat sinks linked:

gigabyte_990fxa_ud3_rev4_1.png
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160
The VRM's are under the Heatsink to the left of the CPU socket. And 81*C is VERY hot! I would becareful. And your motherboard is NOT backwards. If you look at your idle and load voltages, under load the voltage decreases (vdroop). You do not want this. You are not using the correct test either.
Click options, torture test, custom -Min FFT 4096, and memory-as much as you have free.
And it can not be too stable if it says the max Voltage for VN VIR was 1656 :lol:
I don't use that program though, I prefer OpenHardwareMonitor personally.

You have to understand the heatsink is taking the heat to cool the component. So it won't read 100% as if you were to target the VRM's directly. Also, if you have something blowing at the VRM's, you're blowing hot air off the heatsinks helping to cool them, but the Heatsinks will be cooler than the VRMs.
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160



But stock LLC is at 100%. What do you suggest then for stability and for more overclocking and the VRM temps?
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160


LLC on stock should usually be Auto. You want high or very high. Not sure what it's called on your board particularly. However, for more overclocking-you'd need to cool your VRM's more. Possibly throw in a custom loop for it-they have blocks for VRM's. If you're thermally limited, then that is it. Pushing it more would just fry your hardware. Also you didn't show you max CPU temperature in that picture. So I have no idea what it was, but most likely 70*C, as a HOT VRM (very close to the CPU) heats up the CPU and vice versa. You want to stay under 65*C for 24/7 OC, and 70*C for as little usage. 62*C is the FX safe temperature though-so I'd recommend staying at that. And if you're VRM's are at 81*C, you have wayy to much current/voltage. low 70's is as high as I'd go.
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
My cpu is at 48 max. Im going to improve cooling to VRM's and lower voltage down a bit.

But the LLC on the board was 100% at stock. when I set it to 75%, it says "low" and when I set it to 0% it says "extreme"
 
Upvote 0 Downvote
Jake Wenta

Jake Wenta

Honorable
Mar 13, 2013
696
1
11,160


For some reason, I can not believe that. Are you sure it is the CPU socket temperature and not the CPU core temperature? Because a VRM at 81*C would make the CPU ~65-70*C

A cooler CPU means a cooler VRM, a cooler VRM means a cool CPU. It's almost like a hybrid cooling. But it also works the other way, hotter VRM's->hotter CPU.

For VRM's, this should fit your board:
http://www.frozencpu.com/products/15212/ex-blc-1071/EK_Gigabyte_X79_UD7_MOSFET_Liquid_Cooling_Block_-_Acetal_EK-FB_KIT_GA_X79_UD7_-_Acetal.html?tl=g30c293s728

And you do not want extreme. The Voltage gain at extreme under load is immense. And without watching you're voltage's, you'll end up in one conundrum.
 
Upvote 0 Downvote
shadow32

shadow32

Honorable
Aug 8, 2013
881
0
11,160
Okay, I ran your test settings.

I lowered voltages to 1.4625

Under load it is 1.428. The board is still set to low 75% LLC.

The Core is 45C and the VRT2 (VRM's) are around 66C.

How do I find the socket temp?

Edit: HWinfo shows:
Motherboard: 31C
North bridge: 42C
CPU: 44

If the picture I showed you is my motherboard, the northbridge and VRM's heat sink are linked, doesn't that mean they should be the same temp?
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom