Is this a safe 24/7 overclock?

A

Astralogic

Gawd
Joined
Apr 24, 2012
Messages
581
CPU Mulitplier = 44
Vcore = 1.29v
VCCIN = 2v
CPU Ring Voltage - 1.29v
Ring Bus Multiplier = 42

The CPU is a 4770K on an MSI GD65 GAMING board powered by a Seasonic M1211 750W PSU. The temps reach 80 degrees C under load.

I RARELY leave my computer on 24 hours a day but on occasion I do.

How long do you think my CPU will last like this?
 
It will last a long time like 10 years but at some point it might start needing more voltage or drop clock a bit. Ring voltage looks a little high but other than that looks great.
 
Bluesun311 said:
It will last a long time like 10 years but at some point it might start needing more voltage or drop clock a bit. Ring voltage looks a little high but other than that looks great.
Click to expand...

I guess I'll know when it needs more volts/lower clock when I start getting blue screens?
 
VCCIN looks like it's a bit higher than it needs to be as well as ring voltage.
Have you tried about 1.8V with highest LLC setting?

Also, seeing as ring bus has very little effect on performance it may be worth dropping it to about 40x if that allows you to reduce ring voltage a bit.
 
case sensitive said:
VCCIN looks like it's a bit higher than it needs to be as well as ring voltage.
Have you tried about 1.8V with highest LLC setting?

Also, seeing as ring bus has very little effect on performance it may be worth dropping it to about 40x if that allows you to reduce ring voltage a bit.
Click to expand...

I will change the VCCIN to 1.8, and change the LLC (if it's not already at max, I can't remember what it is set to, or even if I'll be able to find it, I'm an OC noob). Then I'll run an 8 hour stress test.

I'll change/test the ring bus after testing the VCCIN and LLC.
 
From what I can tell, LLC on MSI boards is called Vdroop Offset. So I changed the CPU VDroop Offset Control from auto to +100. All the other settings related to it I left on Auto. I then changed the volts to 1.8 and I have successfully booted :D I will now run an 8 hour stress test.

If I crash should I increase the volts until I'm stable?
 
Yeah thats a great oc. Will last a very long time. Cpus are built to run 24/7 at 100% and last years.
 
The 8 hour stress test succeeded. So before I try to reduce the VCCIN any further I will reduce the ring bus to 40 and see how low I can get the ring bus voltage.

Edit: I set the Ring Bus Voltage to auto, should I instead set it to a specific value?
 
Last edited:
Astralogic said:
The 8 hour stress test succeeded. So before I try to reduce the VCCIN any further I will reduce the ring bus to 40 and see how low I can get the ring bus voltage.

Edit: I set the Ring Bus Voltage to auto, should I instead set it to a specific value?
Click to expand...

I wouldn't try to reduce VCCIN any further, 1.8V is around stock anyway and it needs to be approximately 0.5V higher than the next highest voltage.

It's probably best to manually set Ring Bus Voltage to prevent the board from increasing it.
I would aim for 1.2V or below.

If the CPU can't pass a stress test at that voltage it is probably better to lower the ring bus multiplier rather than increase voltage because ring bus affects performance so little.
 
case sensitive said:
I wouldn't try to reduce VCCIN any further, 1.8V is around stock anyway and it needs to be approximately 0.5V higher than the next highest voltage.

It's probably best to manually set Ring Bus Voltage to prevent the board from increasing it.
I would aim for 1.2V or below.

If the CPU can't pass a stress test at that voltage it is probably better to lower the ring bus multiplier rather than increase voltage because ring bus affects performance so little.
Click to expand...

OK, I set the ring bus voltage to 1.2v instead of auto. I'll now stress test it.
 
The ring bus blue screened me at 40 so I've set it to 39, which I believe was the default. Stress testing again.
 
1.2V isn't going to do any harm but if you can go lower and stay stable then why not :)

Once you are happy with the overclock it might be a good idea to do a longer stress test (e.g. Prime95 for 24 hours) just make sure that the core voltage is NOT set to adaptive.

Other than that, well done and a nice overclock.
 
case sensitive said:
1.2V isn't going to do any harm but if you can go lower and stay stable then why not :)

Once you are happy with the overclock it might be a good idea to do a longer stress test (e.g. Prime95 for 24 hours) just make sure that the core voltage is NOT set to adaptive.

Other than that, well done and a nice overclock.
Click to expand...

Damn, I have both core and ring volt mode set to adaptive :/
 
Just set them manually to target voltage, enable (not auto just in case) C1E and EIST (speed step)

then go to power options in Windows control panel, change plan settings (whichever plan you are using) >
advanced power settings > Processor power management and make sure minimum processor state is 1%

Setting it to manual rather than adaptive will prevent the CPU from being over-volted further when running Prime95.
To check that the core voltage is still lowered when idle, use CPU-Z version 1.64 or older or use HWiNFO64.
 
Thanks for the help. I hope you don't mind giving me some more tips because I don't quite know what to adjust from here. After changing cpu core and ring voltage from adaptive I got crashes. What would you suggest to do from this point? Increase the core voltage? I've reverted the ring ratio and voltage back to deaults (auto), better to work on the cpu ratio alone first right?


I'd also like to OC my igpu, I use quicksync to record and stream so it might come in handy for me. See that there are only three settings to adjust it would seem to be a simple thing to do, but the CPU ratio and voltage comes first.
 
What is the core voltage currently set to and do you know what the BSOD error code was (probably either 124 or 101)?
I'm assuming the crashes were during a stress test. If they were at idle, disable C3 and C6/C7.

Yes it is better to work with one variable at a time.

Not too sure on iGPU overclocking as I don't use it myself.
Googling HD Graphics 4600 overclocking might turn up something useful.
 
The core voltage was set to 1.2 (previously 1.29) and the BSOD was the instant Windows booted up while it was loading so it wasn't when the CPU was idle. It must have been because the vcore was too low.
 
You will still need the same core voltage you needed before for stability.

Using manual rather than adaptive prevents the voltage from being increased further (to potentially unsafe levels, +0.1V) when the CPU is under certain loads e.g. Prime95.
For this reason, you may actually need to set a slightly higher voltage than before but this shouldn't be anywhere near as high as the maximum voltage in adaptive mode.

Try working upwards from 1.25V.
 
I was wondering why my temps were at 100 degrees....... then noticed I accidentally set my voltage to 1.7 *facepalm*
 
At least an hour, maybe more :eek:

I didn't get any errors or BSOD's though.

Do you recommend me changing any of these settings:

Intel C-State = Enabled
C1E Support = Enabled
Package C-State Limit = Auto

CPU Ratio Mode = Dynamic
Enhanced Turbo = Enabled
 
1.7v is probably your vccin. That's not safe value for vcore and something would have gone wrong.
 
Astralogic said:
At least an hour, maybe more :eek:

I didn't get any errors or BSOD's though.

Do you recommend me changing any of these settings:

Intel C-State = Enabled
C1E Support = Enabled
Package C-State Limit = Auto

CPU Ratio Mode = Dynamic
Enhanced Turbo = Enabled
Click to expand...

You can leave those settings how they are if you want your CPU to down clock when not fully loaded.

Also, hopefully you didn't do any damage to the CPU with the 1.7v.
 
ilal2ielli said:
You can leave those settings how they are if you want your CPU to down clock when not fully loaded.

Also, hopefully you didn't do any damage to the CPU with the 1.7v.
Click to expand...

Nah it seems fine. IS there any point in OCing the Ring Bus? I just did a benchmark with the Ring Bus OC on then did another with it off and the scores were the same.

Edit: I'm currently Oc'ing my iGPU. Is there a way to stress test both IGPU and CPU at the same time? I want to see what effect both OC's have on my temps when stressed simultaneously. Currently I am only able to stress test them individually.
 
Last edited:
Astralogic said:
I was wondering why my temps were at 100 degrees....... then noticed I accidentally set my voltage to 1.7 *facepalm*
Click to expand...

Astralogic said:
No it was definitely core voltage.
Click to expand...

Astralogic said:
Is there a way to stress test both IGPU and CPU at the same time? I want to see what effect both OC's have on my temps when stressed simultaneously. Currently I am only able to stress test them individually.
Click to expand...

Wow :D That's crazy voltage. These chips are amazing. I didn't think you could go that high but I checked my BIOS and it can do Vcore up to 1.9v! Vccin is normal to be at 1.8-1.9 so I just assumed. + another thing is that Hwmonitor shows vcore as vccin on a couple boards I've seen for at least the BIOS revision I was looking at. On mine I am using right now it has the Vcore as VIN4... and it just reads the VID as Vcore.

As for video+ cpu stress test some people just use multiple apps like prime95+furmark at the same time. That's completely unrealistic of course, but it still should technically work even if it cause temps to go (even more) berserk.
 
Is there any point in overclocking the ring bus at all or shall I just leave it alone completely?
 
You must log in or register to reply here.
Back
Top