AMD Phenom II 1050T OC @ 3.7 problems

[craig_]

craig_, do you understand that when electronic devices start to pull a load that line voltage can drop? That's why the lights in a house can temporarily dim when the furnace or the AC kicks in.

I thought my $250 mobo should be taking care of that for me. Like I said, I'm completely new to all of this. I also thought that's what the LLC was supposed to do. I think I have it set 1 mark higher than normal.

 

[Johan45]

What happens when your system fails? Do you have a current backup image or something else to rely on?

Nope. There's always the repair utility. But I figure if it's that bad I just wipe and reinstall. No biggie

If you're not living on the edge you're taking up too much space.

 

[trents]

Yes, LLC is supposed to compensate for vdrop (voltage fall of at idle in Windows) and vdroop (voltage fall off under heavy load in Windows). In one sense it doesn't matter in that it will take whatever voltage it takes to keep the system stable whether LLC compensates or not or even whether you have or use LLC or not. If LLC isn't doing the job, can it and just add more voltage in bios. If you have to increase the voltage in bios to 1.55 (as an example) to wind up with 1.425 under full load in Windows then do it. You don't live in bios anyway.

 

[craig_]

Nope. There's always the repair utility. But I figure if it's that bad I just wipe and reinstall. No biggie

If you're not living on the edge you're taking up too much space.

HAHA that's great! For me, it takes at least two full days to re-install all software and utilities. That's time I could be working or playing. Installing software is no fun. Much easier to make a disk image and refresh it anytime you make major system changes.

 

[Johan45]

I don't keep much of any importance on my system drive, all my drivers are on a flash and win8 combined with a SSD takes about 10-15 minutes to install. Besides system image disks tend to pick up garbage along the way . The stuff that probably lent to the crash in the first place. If anything I'll make a new installation disk when the first service pack is released.

 

[craig_]

right, so I started playing in my bios again and found out how to use the manual setting for v-core voltage. It didn't make sense to me before. So now I've got my v-core set to 1.4 and north bridge set to 1.25. I upped the LLC to high which shows a 50% compensation. Under full load my v-core is 1.380 at idle 1.392. This looks alot closer to what I have set in bios, not so scary. Ran prime last night and it crashed in 15 min. (didn't drop cores, program stopped working) Testing again now with the values mentioned earlier. CPU is staying at 44 c. What is the cutoff temp on your northbridge? I don't have any cooling on my mobo, just case fans. Right now it's 60 c.

 

[craig_]

Question about cupid Hardware Monitor: What is the Powers listing, says package 141.60 W

 

[RGone]

Question about cupid Hardware Monitor: What is the Powers listing, says package 141.60 W

No one knows if that is a 'true' read out of the power the cpu draws to run the speed you have set or not. I doubt it can be completely accurate since the chips that do the sensor reading are not expensive at all. But it does indicate the cpu is drawing power for sure.
RGone...

 

[craig_]

Well I'm stable again in prime after changing from offset mode. I don't think that stability in prime was ever my problem though. Can you suggest a different stability test?

 

[craig_]

No one knows if that is a 'true' read out of the power the cpu draws to run the speed you have set or not. I doubt it can be completely accurate since the chips that do the sensor reading are not expensive at all. But it does indicate the cpu is drawing power for sure.
RGone...

That can't be very accurate... My chip should be drawing 125 w stock. I'll need to check that out with stock settings just to see what it says. 15 watts more for this oc seems like alot.

 

[RGone]

That can't be very accurate... My chip should be drawing 125 w stock. I'll need to check that out with stock settings just to see what it says. 15 watts more for this oc seems like alot.

Power consumed by a CPU is defined as P = C * V² * f
Where C is capacitance, f is frequency and V is voltage.

So if that sensor circuit or yourself actually solved that equation you would see that there is nO division going on a all. Only multiplication. So you up the Vcore and up the CPU Frequency at all, the real total watts or power consumed goes up pretty darn FAST.

Plus as far as I know, most of us put n0 faith in that power number since we doubt the ability of the program and sensor to be anywhere close to accurate. It is only a little gee-whiz that is there.
RGone...

 

[trents]

To illustrate part of what RGone is saying about the software, I've noticed that CPU-z reports wattage of the CPU to be somewhat lower than HWMonitor does. They are only rough indicators of the actual power being handled. But you can be certain that the amount power used by a CPU as it is overclocked rises disproportionately to the amount of overclock because of increasing electrical resistance.

 

[craig_]

Wow, very informative to both trents and RGone. I didn't realize that I'd be using that much more power wile overclocked.

 

[craig_]

Got to play medal of honor with no errors. Final v-core 1.404, cpu nb: 1.27 Temps are running about 48 - 49 top.

 

[craig_]

How do you know that you are finally stable? I keep getting these random program errors. I think the first time I passed 8 hours prime blended I was at 1.34 v v-core and below 1.25 cpu-nb. 2 days later I found that my game wouldn't play correctly. Upped my v-core and nb voltages, I could play my game. Good for one day, then, crash again. This time passed Prime 5 hours and Linx @ 4gb tests for 50 runs. No problems all day, then blue screen. Upped v-core and nb voltages again and dropped 1 core in prime after 5 minutes. Upped voltages again and this time I see that one core has been working on test 4 for 30 minutes. Now I'm at 1.416 v-core and 1.303 nb. Cpu Temp maxes out at about 50-52 degrees, northbridge idles at 45 and goes up to 60 -65 degrees.(is that safe?) Anyone have any Ideas where this instability is coming from? I'm wondering if something else is causing this.

 

[craig_]

This where I'm at currently....

 

[RGone]

While the memory dump files seem rather useless for some and are often seemingly short of real live troubles named in plain english, there often comes a time when it is time to take a look at the dump file created during a windows BSOD.

How to Configure Windows to Create a Dump File when you get a BSOD

Often when everything seems to be falling apart it is time to run Memtest86+ since normal trouble-shooting seems failing and has come time to pull out all the stops.

Honestly when I have an operating system that has BSOD'd three times on me it has been my practice to either wipe and RE-install the operating system or put an "image" of my system (made shortly after installation) onto my hard drive in place of the one that is acting up. Repeated BSOD's can and do corrupt the operating system and becomes then more likely to BSOD. A self-defeating situation.
RGone...

 

[trents]

Try giving your RAM and extra .05 volts.

 

[craig_]

Try giving your RAM and extra .05 volts.

Thanks, I ran memtest + and got an error 30 minutes in. Took your advice and upped my dram voltage by about .05. Seems like the test is moving along faster this time. Is it normal for the beep on the post test be a higher frequency after changing ram voltage?

 

[trents]

I would not think increasing the ram voltage would speed up anything. It just might help with stability, however.