• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD FX Overclocking

A little help please, if someone doesn't mind? I'm new to overclocking and was hoping to get the 'omg it's so easy with this chip you'd be stupid not to' speed boost for my FX8350 system but I'm hitting thermal probs.

With specs as per sig I had my PC running at stock and it was perfectly fine. Toasty temps though (40oC idle and 70+ with throttling on load). I changed to a CoolerMaster Seidon 180 AIO cooler, and swapped the rad fan for a Corsair SP120 for its higher static pressure. I also added a 120mm fan to the front panel of the case, which is just a cheap and cheerful BitFenix Neos ATX Tower. I then zip-tied an ultraslim 120mm fan to the waterblock tubing, which is thus just above, and directed at, the socket and VRMs.

At stock volts and speeds, idle is now 32 - 38oC (our new build house is hot anyway) and load is around 50oC.

I followed the basic guide here for overclocking. Set the BIOS as per the instructions, and upped the multi to 22 for a 4.4GHz clock on stock volts. It boots to Windows, and RealBench runs no problems... except temps hit the thermal limits (62oC core) within a few minutes. :(

Does anyone have any hints? With all these fans and a water cooler I would have thought the thing would have at least been good for 4.5GHz or so without boiling alive. The case is cheap, granted, but it's nicely laid out with seemingly decent airflow and is properly cable managed with a modular PSU.


Temps a few minutes into RealBench (sorry I didn't have CPU-Z open but it was at stock volts and everything else as per the linked guide):


I hit STOP as soon as the temp hit 62oC and it dialled back down quite quickly (as it should). The guide I followed calls for all the cooling and safety stuff to be off so I obviously watched to make sure no damage was done. Can anyone offer any ideas/pointers/wisdom? Throwing the whole bundle on the MM and grabbing a shiny 5820k bundle instead might be considered.... :D
 
The AIO should be able to handle a mild OC so I would hazard a guess that its your case not supplying enough cool air if your AIO is setup as exhaust

Case cooling is important, how is your AIO setup? if its exhaust then its getting hot air from the case and not cooling the CPU efficiently, where is your fan on the AIO plugged in? If its plugged into the CPU header and is PWM you could set that to "Normal" or above in the BIOS to keep the speeds up if it isn't already

Also, what LLC are you using? You might need to tweak this to stop the voltage flying around, if I remember rightly this helped with my temps when I started OCing my 8350, mine is set to Medium on a Gigabyte board
 
Relentless81 said:
The AIO should be able to handle a mild OC so I would hazard a guess that its your case not supplying enough cool air if your AIO is setup as exhaust

Case cooling is important, how is your AIO setup? if its exhaust then its getting hot air from the case and not cooling the CPU efficiently, where is your fan on the AIO plugged in? If its plugged into the CPU header and is PWM you could set that to "Normal" or above in the BIOS to keep the speeds up if it isn't already

Also, what LLC are you using? You might need to tweak this to stop the voltage flying around, if I remember rightly this helped with my temps when I started OCing my 8350, mine is set to Medium on a Gigabyte board
Click to expand...

The AIO is exhaust, yes, with the 120mm intake at the front. There is room for 2x 120mm at the front but I'm out of headers. The AIO's pump is plugged into CHF1 so it gets 100% all the time (obviously, being the pump), and the rad fan is plugged into CPU_FAN with a profile set to 50% speed until 50oC then 100%.

CPU LLC is ultra high (75%) and NB LLC is auto. Here are the relevant settings from the guide:

settings.png


ETA: Tried medium and high LLC and both BSOD before Windows with CLOCK_WATCHDOG_TIMEOUT. I'm back at stock (yay BIOS profiles) for now until I learn more. I've saved my OC settings so I can just flip back whenever. I did notice the stock volts are 1.32v and I can get 4.4GHz stable on this (albeit too hot), but the mobo is running auto at 1.4v+. I might clock the CPU down to 1.32 for now just to give more headroom on load temps.
 
Last edited:
You need to look at whats the best LLC for your motherboard as every board is not the same, also cpu packet is fine upto 70c, its the core which iss not ment to go over 62c but you can go over it but not for a long time while stressing, as while gaming just say it would only hit 50c or so.

Steps i take,
1st. reset bios to stock, load into windows open cpu-z take picture, upload on here.
2nd. find best LLC.
3rd disable what its says in the picture.
4rd. set v-core to 1.4volts and multi to 22.5 for 4.5ghz, run realbench for 15mins, if fails or locks up, up the volts to 1.4125 while checking temps.
 
fx63007850 said:
You need to look at whats the best LLC for your motherboard as every board is not the same, also cpu packet is fine upto 70c, its the core which iss not ment to go over 62c but you can go over it but not for a long time while stressing, as while gaming just say it would only hit 50c or so.

Steps i take,
1st. reset bios to stock, load into windows open cpu-z take picture, upload on here.
2nd. find best LLC.
3rd disable what its says in the picture.
4rd. set v-core to 1.4volts and multi to 22.5 for 4.5ghz, run realbench for 15mins, if fails or locks up, up the volts to 1.4125 while checking temps.
Click to expand...

As I understood it the CPU0 and CPU0 Package temps (same reading) was the core temp, and the CPU reading under the motherboard sensors was the socket temp? That means the temp showing 62 was the CPU cores? There used to be a 10oC difference between them both (which is common I believe) until I added the 120mm fan pointing at the AIO block/socket/VRMs.

How do I find the best LLC? Since it fails to boot at anything other than ultra I guess that's the one to stick to, but since you're advising more volts than I'd been using I guess that could change. Sorry for being such a n00b. :o

ETA: CPU-Z at stock as requested.

cpuz-stock.png
 
Last edited:
Leave the LLC at ultra high.

For the volts if your stock volts are 1.32 try set it to 1.35 for a 4.5 and open hwinfo64 but let us see the vcore while running realbench to see if the voltage is changing.

For the temps look on the page before as i wrote about it.
 
fx63007850 said:
Leave the LLC at ultra high.

For the volts if your stock volts are 1.32 try set it to 1.35 for a 4.5 and open hwinfo64 but let us see the vcore while running realbench to see if the voltage is changing.

For the temps look on the page before as i wrote about it.
Click to expand...

Thanks so much for all the advice. :) My VID is 1.337 volts (awesome irony, such amaze) so I went for 1.35v with a 22.5 multi. It froze after 5 mins of realbench and Overdrive said ~8oC thermal margin.

I bumped the vcore one notch and this time it froze out 6 minutes in with between 2.5 and 3.5 oC thermal margin with all fans on 100%. I've just bumped vcore once more to get back into Windows and write this.

As it was so close to max on the last bench, I imagine this latest vcore bump will push it over the edge. I'll run it for a few seconds to show the volt differences...

IDLE:

4-5GHz.png

LOAD:


Hmm, confusingly this didn't crash at all (expected with more vcore), but also didn't get as hot as fast. It had about 8oC margin for most of the test. At 9 minutes into the 15 minute test however the margin jumped from 2oC and became -1oC and hence in the red. I killed the test immediately. This is the last screenie I managed before it happened.

Not sure what to do now. Does this mean I can't hit 4.5GHz stable in this case/setup? If I drop down to 4.4 then that should allow less volts = more margin, right? On the plus side it seems the vcore didn't change between idle and load, which I think is a good thing?

Sorry again for being a n00b. :o
 
Rainmaker it might be worth investing a fan controller for your case fans and beefing up your case cooling, Bitfenix Recon is cheap and you can run up to 5 fans with 1 temp probe using software called Phoebetria

Once you have better case temps your AIO will cool your processor a bit more effectively

Fx the VRTR1 temps don't go over 80 in games, I had to drop my GPU volts to stock though :( Its not stable if I touch the voltage in games so I'm back to using CCC with 100mhz OC on the core and 50 on the RAM, very frustrating as my goal was 60FPS in Heaven in Ultra settings which I hit 60.1 FPS stable but in some games it misbehaves

CPU is happy at 5ghz though :)
 
Relentless81 said:
Rainmaker it might be worth investing a fan controller for your case fans and beefing up your case cooling, Bitfenix Recon is cheap and you can run up to 5 fans with 1 temp probe using software called Phoebetria

Once you have better case temps your AIO will cool your processor a bit more effectively
Click to expand...

The fan controller I understand. Beefing up case cooling though? I assume you mean more fans. I'm already running Corsair SP120s (the high static pressure ones), so I assume you simply mean adding more of them? I could add an extra one to the rad (push/pull), one more underneath it for exhaust and an extra one on the front for intake. Is that about right? Gratz on the 5GHz, I'm jelly. :D
 
@ rainmaker them temps are fine and another way to check if your stable is to play some games which like using all cores, such as bf4, crysis 3, advance warfare, shadow of mordor, gta v and so on, if it does not crash check your temps and compare it to realbench temps and you should see that it does not hit them temps.

With your aio have you tried to put the fan at the rear of it to pull the hot air out the case instead of pusing the hot through the rad.

@Relentless81, thats good that its lower than what you was stressing at over 100c, yeah i get that aswell with win10, i think it mite be to do with afterburner as if i set my volts to 1.2 @1150/1500 with a profile in AB, hwinfo64 reports my volts as stock but @1150/1500, when i use the overlay AB shows my vcore @1.2, it could just be a bug.
 
Last edited:
fx63007850 said:
@ rainmaker them temps are fine and another way to check if your stable is to play some games which like using all cores, such as bf4, crysis 3, advance warfare, shadow of mordor, gta v and so on, if it does not crash check your temps and compare it to realbench temps and you should see that it does not hit them temps.

With your aio have you tried to put the fan at the rear of it to pull the hot air out the case instead of pusing the hot through the rad.
Click to expand...

Really? 73oC is fine? I thought after 72oC things started to melt/explode? Granted games wouldn't get that high but isn't realbench just an encoding stress test? I encode and run VMs regularly and wouldn't want to toast the chip by running it unattended if that's the case. If I've misunderstood and this really is fine then I'm a happy bunny. Thanks so much again, I really do appreciate you taking the time.

Re: the AIO, no I didn't think of that. :o I'll give it a try and see what happens.
 
Rainmaker said:
Really? 73oC is fine? I thought after 72oC things started to melt/explode? Granted games wouldn't get that high but isn't realbench just an encoding stress test? I encode and run VMs regularly and wouldn't want to toast the chip by running it unattended if that's the case. If I've misunderstood and this really is fine then I'm a happy bunny. Thanks so much again, I really do appreciate you taking the time.
Click to expand...

I had my cpu @80c once but still alive @4,8ghz and yeah its just encoding if you do encoding a lot of the time, you could run 1 just to check if stable and temps wise, if it still that hot while doing that i back it down to 4.4ghz, till you could get more airflow in the case.
 
Rainmaker said:
The fan controller I understand. Beefing up case cooling though? I assume you mean more fans. I'm already running Corsair SP120s (the high static pressure ones), so I assume you simply mean adding more of them? I could add an extra one to the rad (push/pull), one more underneath it for exhaust and an extra one on the front for intake. Is that about right? Gratz on the 5GHz, I'm jelly. :D
Click to expand...

I don't bother with push/pull I read a lot into it and the it makes very little difference if you have good case airflow so I run my AIOs like Fx mentioned, the fan pulls the air out of the case from behind the rad, this makes the rad easier to clean which helps a lot with temps too

I have 3 x 120mm intakes, my only exhausts are on the rads which are 1 x 120mm rad and fan cooling the CPU and 1 x 280mm rad with 2 140mm fans cooling the cpu

The fan controller controls the intake fans speed, all it does it take the temps from the VRM area of the motherboard using 1 probe, at 32c = 800rpm, 35c = 1400rpm and 38C = 1600rpm

Its a great way to temps/noise
 
You must log in or register to reply here.
Back
Top Bottom