• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Computer freezing instead of dropping cores with high OC

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Kenrou

Member
Joined
Aug 14, 2014
So what makes this happen ? Temps are dropping hard where i live so i am experimenting again with OC and for some reason, no matter what voltage or settings, if i go over 4.8ghz i start getting freezes during stress testing. I have always had a very stable system with my previous MB (Crosshair V Formula non-z), so the freezes are new. ATM i have 4.8ghz 1.432v/1.452v which is stable for gaming/video coding/Internet stuffs, and gives me nice low temps ~57c core/package on IBT standard 20 runs (please don't pick at this like i said its stable for everyday use :rolleyes:). I know its high (as it should) with air cooling but i still have some space to manoeuvre as socket doesn't go above 65c.

I have tried up to 1.5v and it doesn't want to go to 4.9ghz on the Saber (used to on the Crosshair). Temps don't have time to hit the "safe" 62c/72c before the freeze. VRM's are in the low 80's, so there should be no problem there, so i am thinking maybe memory ? thoughts ?

AMD FX-8370, 4.8ghz (24 x 200)
CPU 1.432v/1.452 - CPU/NB 1.250v - CPU/VDDA 2.65V
NB 2600mhz / HT Link 2600mhz
 
Its just a 'harder' error. NO ideas though. I doubt its the ram as you are changing the multiplier only, right? So that would not affect the ram speed and its stability.
 
I'm thinking the sabertooth has a lower vrm shutdown temp than the crosshair5. the crosshair shuts down at 90c if you don't disable monitoring.
I'll have to take a peek at the sabertooth.
try a little more air on the vrm then some more vcore.
we have seen a number of later 8xxx chips that will just hit a clock wall.
 
This one's VRM's has been at 95c+ in the 1st few days i tried OC with FSB and no freeze, even asked in another thread what would be its upper limits and someone replied 110c, i shall investigate more. Not keen on installing the Suite just for the thermal radar, disabled the extra sensors in HWInfo64 because they were using ~5% of CPU time (methinks software still slightly bugged) :(
 
my early 8350 will do 4.8 on fair vcore but 5.0 takes much more cooling of the cpu and a very large leap in vcore to get it done.
you want to hit the cpu with all the cold you can, drag the thing out into the garage like I do, and this helps reduce the vcore needed to push onward.

I'll have to drag my sabetooth out of storage and find out where the vrms start to protect it self for me.
 
Yep seems to be the same for all 83** i have seen so far, up to 4.7 its all peaches and roses (im stable at 1.392v/1.416v ~52c with fans at half), from then on... Still trying to fix my Crosshair, the LLC on this one is maddening (or maybe i just got spoiled).
 
With my ram at 1920 I need 1.48v @ 4.8 gHz, 4.9gHz requires ~1.52. From what I understand a dropped core means you're close, but a system hang/freeze means you're nowhere near the voltage the chip needs for stability.
 
With my ram at 1920 I need 1.48v @ 4.8 gHz, 4.9gHz requires ~1.52. From what I understand a dropped core means you're close, but a system hang/freeze means you're nowhere near the voltage the chip needs for stability.

If I go by that 1.5v is far away from what the chip needs for 4.9ghz ?

Going to take the suggestion of sticking a 80mm on the VRM and see if it helps, maybe it's just accumulated heat.
 
I did put the stock AMD fan on my vrm, dropped vcore 1 temps from 82c down to ~62, it might help you out. I also use current probe instead of thermal. If I recall correctly, 4.9 was stable at 1.5v with my ram at 1600.
 
Besides the fan tomorrow i can switch my ram with wife's 2x4gb Vengeance LP 1600mhz, found out a couple weeks ago in another thread that mine might not be 100% compatible (there's 3 "cousins" on the manual but no sign of these) ? ED said that shouldn't be a problem as i am just upping multi/voltage but just in case i would like to make sure :)
 
If the setup just locks up from my experience when this has happened to me. It was either the board couldn't give the chip enough clean voltage, this was when I was trying to Oc my 8350 on a M5A99X evo or I just wasn't giving the chip enough volts for the Oc on my Sabertooth or CHV. When I've crashed due to heat it's been BSOD and Reboots not a Freeze.

I'll also add, on any cooling at some point the chip isn't going to take any more voltage without crashing due to heat. This will happen even with the temps below what we consider safe. I've had more then one chip under sub ambient water BSOD on me due to the amount of voltage I was trying to feed it, even though my temps under load were only in the 30c range.
 
I have only seen the high voltage bsod's over 1.62, not below 1.6.
I have found a few of my notes with the 8350 in the sabertooth at over 5.0.
these are the settings that helped me into the mid 5's.
disable cpu spread spectrum.
disable epu power saving mode.
cancel oc tuner.
cpu current capability to 140%.
cpu power phase control to extreme.
cpu voltage frequency to manual and max it out.
cpu vdda to 2.5
vddr to 1.25.
 
disable cpu spread spectrum. - check
disable epu power saving mode. - check
cancel oc tuner. - check
cpu current capability to 140%. - is at 130% atm will change (never went past 130% because of the red label)
cpu power phase control to extreme. - check
cpu voltage frequency to manual and max it out. - is that safe on air ?
cpu vdda to 2.5 - currently at 2.650v, helped stabilizing 4.8ghz
vddr to 1.25. - ? will change and report back
 
cpu current capability to 140%. - is at 130% atm will change (never went past 130% because of the red label)
Capability. Its like taking off a glass ceiling on how much power the CPU can handle. This PLUS voltage can potentially damage a chip. But if you turned this up to 3000% and didn't touch voltage nothing would change.

cpu voltage frequency to manual and max it out. - is that safe on air ?
Yes, it is a frequency, not a voltage change. It may increase temps, but not a lot at all, particularly when compared to changing voltage.
 
That did the trick, no more freezing. Re-testing now :)

Edit : no luck, temps raise so much that the CPU becomes unstable and starts freezing again at around 1.488v. 4.8ghz on air is not too bad I suppose :)
 
Last edited:
If the setup just locks up from my experience when this has happened to me. It was either the board couldn't give the chip enough clean voltage, this was when I was trying to Oc my 8350 on a M5A99X evo or I just wasn't giving the chip enough volts for the Oc on my Sabertooth or CHV. When I've crashed due to heat it's been BSOD and Reboots not a Freeze.

For me, even though not with FX, too little Vcore usually causes me to get an error message. And a freeze would mean I went way too far.

I had a later-2000s T-bred lock up, starting at around 2415 Mhz and possibly at 2400 Mhz when temp past 55 C.

I have gotten a BSOD due to heat before with my Barton. (KMODE_EXCEPTION_NOT_HANDLED with Windows 2000 in 2007)
 
Looks to me like you need to pump up them CPU_NB volts all that DIGI + stuff is really not going to help that much at normal overclocks unless you're trying to run one of the 9xxx chips that tend to need a bit more power. Also from what I have observed raising the current % will allow the board to pull more at the same voltage settings. So you're only fooling yourself when you adjust them. When I was testing my chips I never changed anything but the LLC and ran well into the mid 5G range without issue.
From what I see in your posts you're running 16GB of ram on a 2600 NB @ 1.25 volts. That will never fly. Even without changes in speed to the NB when the CPU goes faster the NB needs more power to keep up. IIRC you weren't that stable before were you. I'm certain I recall all the IBTAVX business and you couldn't get it to run. The standard IBT isn't going to work that system.
 
Johan is pretty much spot on. Hard lockups are usually related to unstable CPU/NB speed. Try dropping CPU/NB to 2200 or 2400 without any other changes and retest.
I have never really got CPU/NB speeds long term stable over 2400MHz. There are certain multiplier combos that might work better if you feel like playing with the BCLK.
 
Back