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

FX-8350 F@h unstable

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

csimon

Member
Joined
Apr 30, 2005
Hi, this is my first thread here so be gentle.

I'm trying to get my system F@h stable for 24/7 folding. I'm not sure what I'm doing wrong but every time I think I've got things stable, F@h will shut down a few days later. I bump up the voltage but it happens again and again. I'm wondering if it may be a digi+ vrm setting that needs to be adjusted or perhaps something else. I've started back at the original voltages and began adjusting the vrm settings.

I'd appreciate suggestions but keep in mind that anything I change probably won't show instability right away. Usually it takes a day or two before shut down.

Here are my current settings:
BIOS settings:
Current CPU Speed: 4500mhz
CPU Ratio: 22.5
CPU Bus: 200
Mem Freq.: DDR3-2133mhz
CPU/NB Freq.: 2400mhz
HT Link Speed: 2600mhz
DRAM Timings: 9-11-10-28-1T
CPU Voltage: 1.45v
CPU/NB Voltage: 1.3v

Digi+ VRM settings:
CPU LLC: Ultra High
CPU/NB: Regular
CPU O-C Protection: 130%
CPU/NB O-C Protection: 130%

The rest are defaults.

Does anyone else do F@h with this type of setup, and if so what are your experiences? Specs are in sig below.

Thanks,
Chris
 
Last edited:
give screen shots of cpuz all tabs..
and hwmonitor at load..
folding is kinda funny it will find some errors that prime doesn't...
 
give screen shots of cpuz all tabs..
and hwmonitor at load..
folding is kinda funny it will find some errors that prime doesn't...
Here are screens of CPU-Z & HWMon as requested. This is w/ SMP folding only and not GPU. I don't run GPU until the evening while I'm not using the machine, but CPU temps are the similar then, only case gets a few degrees warmer but remain very reasonable. Here are other details of the build:

Case = Lian Li PC-V2000b
CPU Block = Enzotech Sapphire Series SCW-REV.A
Radiator = Thermochill PA160.1 single pass high flow
Pump = Laing D5-12V variable speed

Sorry about my CPU-Z being broken, I think one of my benchmark programs borked it but not sure. See attachments. Am I missing anything?
 

Attachments

  • first tab.png
    first tab.png
    30.5 KB · Views: 560
  • mainboard.png
    mainboard.png
    23.6 KB · Views: 557
  • memory.png
    memory.png
    26.9 KB · Views: 553
  • spd.png
    spd.png
    25.9 KB · Views: 555
  • caches.png
    caches.png
    29 KB · Views: 554
  • graphics.png
    graphics.png
    25.8 KB · Views: 552
  • hwmon1.png
    hwmon1.png
    49.9 KB · Views: 570
Last edited:
well first ur almost at max temps when days get hotter it will be worse..
that's a pretty small rad for that cpu..
I would try dropping the htt down to bout 2200
may have to ease up slightly on the mem...I have that same set and cant quite get what ur getting...
 
Csimon, I agree with Boucher about your temps that may be one cause of your instability. Speaking for my system, I have a tough time with my dram and instability when I start approaching 2000 MHz. The first issue I would work on is your heat load.
 
Ok, I've lowered my cpu/nb, cpu/nb voltage, and mem to the following. I can probably get the temps down a little with a good dusting. I will let the temps rise for about 15mins and check before I do that. This is current:

BIOS settings:
Current CPU Speed: 4500mhz
CPU Ratio: 18
CPU Bus: 250
Mem Freq.: DDR3-2000mhz
CPU/NB Freq.: 2250mhz
HT Link Speed: 2500mhz
DRAM Timings: 8-10-9-27-1T
CPU Voltage: 1.45v
CPU/NB Voltage: 1.1875v

Digi+ VRM settings:
CPU LLC: Ultra High
CPU/NB: Regular
CPU O-C Protection: 130%
CPU/NB O-C Protection: 130%

I haven't adjusted VRM settings yet.
What about the cpu/nb voltages? Shouldn't stock be enough or should I raise it a little?

Thanks for all of the responses so far, I'm listening.

EDIT: Temps weren't effected by this, I'll dust and get back soon.
 

Attachments

  • cpu.png
    cpu.png
    36.8 KB · Views: 552
  • memory.png
    memory.png
    29.6 KB · Views: 543
  • HWMon.png
    HWMon.png
    23.8 KB · Views: 551
Last edited:
mine (the 8350) folds for wcg 24/7 at 4.2 and have been working on more when I have time...
difference is mine is wc/ed to overkill (ek hf cpublock,swifty655,swiftyminires,2xBI sr1 480,s....
 
folding is kinda funny it will find some errors that prime doesn't...

Prime95 is not the best anymore for core testing. Linpack uses more of the chip than Prime95 does. I'm not surprised that Prime95, especially at its default settings finds nothing.
 
Csimon are you running push pull on the radiator? I know that on my chip when I start getting close to 63 on the package, things start getting screwy. Your chip may it like anything above 60. If you are only running 1 fan you may want to get another or drop the OC and CPU v and see if lowering the temps helps with stability.
 
Csimon are you running push pull on the radiator? I know that on my chip when I start getting close to 63 on the package, things start getting screwy. Your chip may it like anything above 60. If you are only running 1 fan you may want to get another or drop the OC and CPU v and see if lowering the temps helps with stability.
mandrake, I am pushing a cougar 120 vortex pwm from the front of the lower compartment of the case. That compartment is exhausted but 2 80mm fans and a PSU 140mm fan, but no direct pull. I can elaborate more on this if need be.
In order to add a fan to this rad I would need another shroud. I regret not buying one while Danger Den was still in business. I do have plans for a new case and watercooling setup after tax refunds.

Anyhow, the dusting seems to have payed off nicely. Aso, I am wondering if I really need 1.45v for 4.5 ghz? Problem is, no matter what I set now it may not surface issues for a few hours/days.
 

Attachments

  • cpu.png
    cpu.png
    38.1 KB · Views: 535
  • mem.png
    mem.png
    27.6 KB · Views: 529
  • hwmon.png
    hwmon.png
    18.7 KB · Views: 530
Last edited:
Prime95 is not the best anymore for core testing. Linpack uses more of the chip than Prime95 does. I'm not surprised that Prime95, especially at its default settings finds nothing.

F@H will still find errors Linpack doesn't too. If I want the absolute most stable OC I always use F@H to test it as it has found errors many times for me that nothing else did.
 
cpu/nb is 1.25v on mine..
ht link speed on mine is 2200

Awesome temps you got there. I did raise my cpu/nb voltage just to be sure. Current settings below:

BIOS settings:
Current CPU Speed: 4500mhz
CPU Ratio: 18
CPU Bus: 250
Mem Freq.: DDR3-2000mhz
CPU/NB Freq.: 2250mhz
HT Link Speed: 2500mhz
DRAM Timings: 8-10-9-27-1T
CPU Voltage: 1.45v
CPU/NB Voltage: 1.25v

Digi+ VRM settings:
CPU LLC: Ultra High
CPU/NB: Regular
CPU O-C Protection: 130%
CPU/NB O-C Protection: 130%

I'm wondering if the O-C Protection settings are really necessary?
 
I read the title of this topic and, based on my experience, knew without even opening it that the problem would be heat and running too high of overclock.

I started out folding on my 8350 at 4.5ghz too, since I could get 5.0 running for short bench sessions and 4.7 stable for normal computing. That lasted about.. two minutes into each folding session. The temps would either hit 60+ and I had to shut down, or FAH would have the client throw errors and quit, losing all my work. I pulled it back and back. Finally found that depending on how serious you are about keeping the room air cool and the case open, either 4.2 or 4.3ghz is the sweet-spot for 24/7 folding without much worry.

Your temps are the main concern. 60+ can damage the processor. Anything over 53-54c and I notice the processor throttles back and you end up with a slower TPF and lower PPD projection than if you had been running at a little slower clock speeds. Also, anything in this temp range and that throttling tends to almost always freeze the FAH client the first time that you touch the mouse or open anything other programs at all, even just HWMonitor. The key here is that these processors really weren't designed for 100%, 24/7, round-the-clock workloads like this. Even running it at stock speed is probably fortunate to have no stability issues. A mild OC is definitely what's called for here. Better to lose 1-2% from your PPD projections than lose WU's that crash after running for long periods...
 
I'll try lowering the voltages a tad and see what I get.

Current settings:

BIOS settings:
Current CPU Speed: 4500mhz
CPU Ratio: 18
CPU Bus: 250
Mem Freq.: DDR3-2000mhz
CPU/NB Freq.: 2250mhz
HT Link Speed: 2500mhz
DRAM Timings: 8-10-9-27-1T
CPU Voltage: 1.425v
CPU/NB Voltage: 1.2v

Digi+ VRM settings:
CPU LLC: Ultra High
CPU/NB: Regular
CPU O-C Protection: 130%
CPU/NB O-C Protection: 100%
 
I read the title of this topic and, based on my experience, knew without even opening it that the problem would be heat and running too high of overclock.

I started out folding on my 8350 at 4.5ghz too, since I could get 5.0 running for short bench sessions and 4.7 stable for normal computing. That lasted about.. two minutes into each folding session. The temps would either hit 60+ and I had to shut down, or FAH would have the client throw errors and quit, losing all my work. I pulled it back and back. Finally found that depending on how serious you are about keeping the room air cool and the case open, either 4.2 or 4.3ghz is the sweet-spot for 24/7 folding without much worry.

Your temps are the main concern. 60+ can damage the processor. Anything over 53-54c and I notice the processor throttles back and you end up with a slower TPF and lower PPD projection than if you had been running at a little slower clock speeds. Also, anything in this temp range and that throttling tends to almost always freeze the FAH client the first time that you touch the mouse or open anything other programs at all, even just HWMonitor. The key here is that these processors really weren't designed for 100%, 24/7, round-the-clock workloads like this. Even running it at stock speed is probably fortunate to have no stability issues. A mild OC is definitely what's called for here. Better to lose 1-2% from your PPD projections than lose WU's that crash after running for long periods...

How can I tell when my cpu is throttling? Are you running F@h stable at 4.7ghz? What are your temps?
 
from what musicjunkie says im also finding 4.2-4.3 for the sweet spot for folding..any higher to achieve stability doesn't seem to be worth what little gains versus the more voltage and temps...
 
Back