Overclock.net banner
1 - 20 of 27 Posts

TheDocToR10

· Registered
Joined
·
16 Posts
Discussion starter · #1 ·
Hey guys!
I joined this forum a day ago mainly because of this problem I'm writing about. In August last year I built this new rig under my profile and it works fine, no problem. Well, at least while not OC'ing. The thing is, every time I try to OC the CPU it boots up normally, it even works for half an hour but then locks up and I have to plug the power cord out of the wall and reset it that way because nothing else works. At 4GHz it's pretty stable but still does the same thing. On 4.2 it's almost instant after I load up OCCT. On stock everything's working fine. Three main components I think might be the cause of this are (from most probable to least): RAM, MOBO, CPU.
My RAM runs at 1333MHz stock and has 9-9-9 timings, but is also tested at 1600MHz with 10-10-10 timings. Today I found out I was actually overclocking my RAM to 1600MHz while leaving the timings at 9 which I thought may have caused the system to freeze. Unfortunately, that wasn't the problem. I even ran MemTest86 and everything worked fine.
My main concern is if my motherboard is faulty, but I don't think it is because than something would be wrong even on stock... And 4+2 power phase is definitely enough for 4.2GHz on this CPU (my friend has 4.5GHz on a 4+1 power phase mobo, same CPU).
Other than that, it could only be my OS. I'm running Windows 7 Ultimate 64-bit. Also, my mobo BIOS version is 1903.

If you know anything that can help, please share it.
smile.gif


Thanks in advance
wink.gif
 
Well... Using anything less than 8+2 for an 8 core is going to be the first thing people bring up. (Even me... heh...) Anyway, what volt are you at when you are locking up? The 8350 is higher binned, but I can get a 22x multi with 1.3875v on my Sabertooth.
 
Discussion starter · #4 ·
I'm at 21X multiplier for 4.2GHz and I am raising the voltage and adjusting it. It's almost 100% stable on 4GHz with 1.272V, so I think that 1.31 - 1.33 (that's where I was when it locked up) should be enough for 4.2GHz. Also my cooling is Noctua NH-D14.
 
Discussion starter · #6 ·
Quote:
Originally Posted by Dillmiester View Post

You might want to look at the memory the timings or voltage.
Yeah that's the most probable solution to me right now as I haven't tested the memory much on different settings. I'll try something more with that and see what I come up with.
 
Discussion starter · #8 ·
Quote:
Originally Posted by AcEsSalvation View Post

Try increasing NB voltage. I wonder if that isn't getting enough for CPU and RAM
I'll definitely try out your advice. I'm from Croatia and it's nighttime here so it'll have to wait until tomorrow. Will report back the results.

Cheers
smile.gif
 
Are you messing with the LLC settings in the BIOS? That can help as well if the 2.0 boards have it. I have the same board and CPU except mines the EVO version. (Still not sure on the difference.)
 
this definitely sounds like a voltage issue on your problem. my guess is one of 2 things, A. your not feeding it enough vcore. B. you might not have enough juice from your power supply. I know PC Power and Cooling makes a good power supply but I think 500 is probably a bit lowish for that 8350 for overclocking it. was the power supply purchased brand new or did you have it left over from another rig? its definitely a voltage issue, if your ram is rated for ddr3-1600 and you were downclocking it to ddr3-1333 that wont be the issue. try to bump your vcore up and maybe your cpu nb volts up to see if it stabilizes. make sure you run the recommended voltage for the ram. if all else fails see about borrowing a bigger and better power supply if you can from someone else and see if that snubs out your freezing issue.
 
Discussion starter · #11 ·
Alright guys, I tried out your advices with increasing the voltage. I bumped it up on NB, RAM and CPU more than usual. I was leaving the voltage setting on NB and RAM on "Auto" before. Now the problem seems to be solved. System is extremely stable even under heavy 100% load using OCCT for 15 minutes and with 3 tabs (including youtube videos) in Firefox opened. Before it would lock-up under a minute in OCCT. I think it could be NB, but will continue testing later. The PSU isn't the problem.
smile.gif


Thanks guys so much for help, I would never think of increasing NB voltage because I always thought that "Auto" was giving it enough juice... Doesn't seem like it.

Also, can someone explain to me what LLC is actually for? Is it better to leave it on enabled or disabled? Or Auto? Those are the only 3 settings I have for it. Tnx
wink.gif


Cheers!
thumb.gif
 
Quote:
Originally Posted by TheDocToR10 View Post

Alright guys, I tried out your advices with increasing the voltage. I bumped it up on NB, RAM and CPU more than usual. I was leaving the voltage setting on NB and RAM on "Auto" before. Now the problem seems to be solved. System is extremely stable even under heavy 100% load using OCCT for 15 minutes and with 3 tabs (including youtube videos) in Firefox opened. Before it would lock-up under a minute in OCCT. I think it could be NB, but will continue testing later. The PSU isn't the problem.
smile.gif


Thanks guys so much for help, I would never think of increasing NB voltage because I always thought that "Auto" was giving it enough juice... Doesn't seem like it.

Also, can someone explain to me what LLC is actually for? Is it better to leave it on enabled or disabled? Or Auto? Those are the only 3 settings I have for it. Tnx
wink.gif


Cheers!
thumb.gif
Great to hear it's resolved. Sometimes there isn't enough voltage on the NB for the memory controller on the chip. The CPU does need to talk to the RAM
thumb.gif

LLC (Load Line Calibration) is a way to prevent VDroop. When put under stress, the voltage changes. Sometimes it can spike wildly (not too dangerously as long as it isn't above recommendations) and can cause a lockup. When the chip needs more volts than what it is getting and cannot get enough fast enough - lockup. LLC keeps the volts higher; close to what will be needed when under load. This does result in loss of efficiency and higher temps... but this is OCN
devil.gif
 
Discussion starter · #13 ·
Quote:
Originally Posted by AcEsSalvation View Post

Great to hear it's resolved. Sometimes there isn't enough voltage on the NB for the memory controller on the chip. The CPU does need to talk to the RAM
thumb.gif

LLC (Load Line Calibration) is a way to prevent VDroop. When put under stress, the voltage changes. Sometimes it can spike wildly (not too dangerously as long as it isn't above recommendations) and can cause a lockup. When the chip needs more volts than what it is getting and cannot get enough fast enough - lockup. LLC keeps the volts higher; close to what will be needed when under load. This does result in loss of efficiency and higher temps... but this is OCN
devil.gif
It is
smile.gif

I also noticed that when I raised what was needed the PC became noticeably faster. I mean, when I had it on 4.2 without increased NB voltage, windows was loading quite slower than now. That's also a nice change
smile.gif

I usually leave the LLC on disabled and put the voltage on something that will always be enough for the CPU to work. I'd like to try pushing it to 4.5 cause I can with my cooling, but not sure if mobo will allow it.
biggrin.gif
 
Quote:
Originally Posted by TheDocToR10 View Post

I also noticed that when I raised what was needed the PC became noticeably faster.
I immediately know I have a bad OC when Windows takes an additional 15 seconds to let me enter my password. It's not guaranteed that it is a bad overclock, but most of the time it is. Right now I have my 8350 on my Sabertooth at 22x 200 with 1.3875v. I can do 22.5x for a 4.5GHz OC with 1.3925v, but I do put all my cores under stress in daily situations, so temperatures are an issue. I did just clean my H100 vent and the fan filters on my case, so I think I'm good at 4.4GHz for a little bit
 
Discussion starter · #15 ·
That
Quote:
Originally Posted by AcEsSalvation View Post

I immediately know I have a bad OC when Windows takes an additional 15 seconds to let me enter my password. It's not guaranteed that it is a bad overclock, but most of the time it is. Right now I have my 8350 on my Sabertooth at 22x 200 with 1.3875v. I can do 22.5x for a 4.5GHz OC with 1.3925v, but I do put all my cores under stress in daily situations, so temperatures are an issue. I did just clean my H100 vent and the fan filters on my case, so I think I'm good at 4.4GHz for a little bit
That's great!
smile.gif

I didn't know that unstable OC could slow down the system. Also, what power saving features can I leave enabled? Currently they are all disabled. What's APM? Or Application Master Mode it says?
 
Discussion starter · #16 ·
Well, unfortunately the problem isn't solved. Just came back from school, turned on the PC and after about half an hour locked up on a youtube video... Same settings as yesterday when I thought I solved it.
frown.gif
 
Quote:
Originally Posted by TheDocToR10 View Post

Well, unfortunately the problem isn't solved. Just came back from school, turned on the PC and after about half an hour locked up on a youtube video... Same settings as yesterday when I thought I solved it.
frown.gif
I'm starting to suspect the 4+2 VRM. I don't want to discourage you, but if applying more volts to NB, RAM, and CPU didn't give stability, then the last option is to turn on LLC and pray for the best. Good luck
 
Discussion starter · #18 ·
Quote:
Originally Posted by AcEsSalvation View Post

I'm starting to suspect the 4+2 VRM. I don't want to discourage you, but if applying more volts to NB, RAM, and CPU didn't give stability, then the last option is to turn on LLC and pray for the best. Good luck
The LLC is turned on... :/
There is one other thing I'm afraid of... I made a huge mistake about a month and a half ago, but I didn't think about it at all till now. I tried to update my BIOS through USB BIOS Flashback and it didn't work. The light didn't flash as intended and the system didn't boot. Then I started fraking out but found a way out and downgraded the BIOS to version 15xx I believe. It's now on v1903. Funny thing is, when I got the PC in August and Oc'ed it to 4.4GHz everything worked fine. But after this bios thing it started happening...
It' not possible that 4+2VRM can't reach 4GHz because almost 100% chips do reach at least 4.2, so I'm really not suspicious about the VRM.
What do you think?
 
Quote:
Originally Posted by TheDocToR10 View Post

The LLC is turned on... :/
There is one other thing I'm afraid of... I made a huge mistake about a month and a half ago, but I didn't think about it at all till now. I tried to update my BIOS through USB BIOS Flashback and it didn't work. The light didn't flash as intended and the system didn't boot. Then I started fraking out but found a way out and downgraded the BIOS to version 15xx I believe. It's now on v1903. Funny thing is, when I got the PC in August and Oc'ed it to 4.4GHz everything worked fine. But after this bios thing it started happening...
It' not possible that 4+2VRM can't reach 4GHz because almost 100% chips do reach at least 4.2, so I'm really not suspicious about the VRM.
What do you think?
I've been following a few threads dealing with overclocking problems, so I got yours mixed up with another, my mistake. I thought you were going for a higher overclock. I would reset your BIOS/CMOS and re flash it just to be sure. The ASUS easy flash is actually really easy to use, so it shouldn't be an issue.
 
Discussion starter · #20 ·
Alrig
Quote:
Originally Posted by AcEsSalvation View Post

I've been following a few threads dealing with overclocking problems, so I got yours mixed up with another, my mistake. I thought you were going for a higher overclock. I would reset your BIOS/CMOS and re flash it just to be sure. The ASUS easy flash is actually really easy to use, so it shouldn't be an issue.
Alright, so should I flash it with the latest version?
 
1 - 20 of 27 Posts