Would you trade 20% OC for 1BSOD / week?

Talking about my G51 w/ P7350. And there is no drop to 15% OC for no BSOD option. I've done a little charting:

~500+ hours without using SetFSB -> no BSOD (actually way more than 500 hours, but only counting hours since I first used the SetFSB program and then turned it off for control purposes)

~100+ hours with SetFSB OC'ed to 2.45GHz -> 3 BSOD

~100+ hours with SetFSB OC'ed to 2.05GHz -> 3 BSOD

So, seems to be all or nothing..

Never happens while gaming, usually while idling, surfing the web, using itunes, etc.. I've tried to rule out other stuff. I.e. without using SetFSB, I've OC'ed the graphics card using nTunes to pretty ridiculous levels and left it there for two days w/o BSOD. Also passed the windows memory diagnostic tool.

They are all IRQ_NOT_LESS_OR_EQUAL with no offending file details. Next time it happens I'll post a minidump (if someone is kind enough to take a look at it to confirm it's caused by overclocking, mucho thanks...).

Anyway, would you trade?

 

Very interesting results... How did you manage to test it ~100+ hours?

Anyhow, I think it would be logical to keep it at 2.45 as you have the same number of BSODs with CPU at 2.05

My 2c...


Best Regards, Leonid

 

Oh yeah. My gaming desktop restarts in like two seconds. I'd gamble on that anyday. Plus think about the gross amount of points you get Folding@Home.

 

I leave my notebook on 24/7, so just left it on those settings for a little over 4 days.

EDIT: i guess technically it's less than 100 hours b/c I woke up to a few BSOD's, so there was time overnight when it couldn't have BSOD'ed again

EDIT: heh I guess the title doesn't make much sense considering what I just wrote. I only use the laptop for like 5 hours per day, so in that sense 3/100 hours comes out to approx once / week while I am using it. BSODs when I'm not using it don't bother me at all since this isn't my torrent rig.

 

Yay (imperceptibly) higher power bill for (imperceptibly) helping out mankind. no thanks

 

Screw helping mankind. I'm in it for braggin rights. And it teaches me how to remote manage multiple machines.

 

Try this:

Overclock 20%.
Run ORTHOS on the "stress CPU" setting.
Then run ORTHOS on the "blend" setting.

If ORTHOS gives no errors on the "stress CPU" setting, but fails the "blend" setting then you know your ram is unstable.

You could use thaiphoon burner to flash a lower frequency onto the ram.

 

erm setfbs allows you to change the fsb at will so why would you leave it overclocked while doing nothing? Simply down clock it while idle and problem solved no more bsod while doing nothing....

 

For the same reason you would buy a higher clocked CPU to have while "doing nothing" ? some people(read: the majority of people) don't want to have to continually open and close the program just because they want to do something.

 

>.> two clicks seems rather inconvenient to me compared to a total system crash and having to restart the computer. To each their own I suppose.

 

Each BSOD is a hardware error from over extension of the hardwares capabilities. You are putting your hardware at risk and your data as well. You have to look at it as your camel will only carry so many straws, which one will finally break the camels back?

 

Exactly.

One BSOD a week may seem like nothing, until that one BSOD causes you to lose 5 hours of work.

 

would not trade 20% OC for one BSOD a week....

 

I've tried this, blend and stress work fine except when I go up beyond 2.6GHz, then it gives a rounding error, but the system does not crash.

Well, I got another BSOD. And I figured it was time to stop being lazy and learn how to use debug. The culprit file is ataport.sys. I can't seem to find any information about this file except that it may be involved in RAID crash issue, but I don't use RAID. Does anyone have any insight/

 

Yeah this bugs me. But why does would it BSOD at only a 50MHz overclock? I know every chip is different and that this one may have barely passed for a P7350, but then I would expect it be far less stable with a 450MHz overclock, which is not the case. I even ran it for a few hours with a 600MHz overclock with no BSODs except for the Orthos rounding errors.

The good thing is that there is no data to lose. I use Google Docs exclusively, don't even bother with localized office suites anymore. Let's face it, in 5 years, we'll all be using Chrome OS and losing data due to system crashes will be a thing of the past.

 

Yeah this is probably the right solution and what I'm going to try, but I have a gut feeling it is SetFSB that is causing the BSOD, not the overclocking itself. I wonder if running SetFSB.exe -s200 will be the same as shutting down and booting back up having never run the program. Time for more experiments I guess.

 

(ahh the multi-quote button, I'll use that next time).

Yeah it is inconvenient, but what you gotta love about win7 & only 3 startup programs is that a reboot takes 12 seconds.

 

It isn't the program but it sets variables for the PLL. apparently one of these variables for the bus is causing an issue. It could be just a timing setting or even a latency setting. You may even find using SetFSB to set to stock clock you still get the BSOD, you can do this by first overclocking the use SetFSB to go back to stock speed.

This could also be as you are using a PLL that allows you to overclock but it is not an exact match. Unless you are absolutely sure the PLL is correct you could try going through the list and see if another works as well or better.........

 

I use setfsb to overclock my p7450 to 2.6ghz and have had zero bsod so far. I think your luck with fsb overclocking is really just based on the quality of ram.

 

I agree that setfsb is causing the problem.

For example when I was trying to get setfsb working on my notebook (after I did the pin mod) it would cause the system to lock up, then BSOD.

I found that it was because setfsb was setting the source for SATA clock to PLL2. This would cause my HDD to go nuts.

Anyway, a quick email to Abo (creator of setfsb) and he gave me a test version which did not change the source for the sata clock, it works for me now.
I think this might be your problem.

Another user also had a problem because setfsb was changing the source for SRC Main, which again caused a BSOD.

So here is my advice:
Open setfsb.
Click on the diagnosis tab.
click on getfsb.
Now change the source for the sata clock.
Hit update and apply.
Your system should BSOD eventually. If it does, then you can tell abo and he will give you a test version.

If you dont know which bit is your source for the sata clock, check your datasheet for the PLL. I can help you with better instructions if you tell me which PLL you have.

I'm sure we can troubleshoot this, and you will not BSOD anymore

 

My is PLL is ICS9LPR604AGLF. I'm trying to find a datasheet to tell me which bit is my sata clock source, but no luck so far. Would you know where to look for this?

The info at this link:
http://nbmod.blogspot.com/2009/07/overclocked-by-modifying-my-pll-clock.html
says that it is at bit 1 of byte 0, for the PLL ICS9LPR501SGLF.

Since both our PLL share the initial ICS9LPR, might it be the same for me too? (sorry if that's a stupid question).

Here is a capture of SetFSB after a boot from shut down and hitting Get FSB (without touching anything else):



It seems my bit 1 of byte 0 is already set at 1, but maybe this is irrelevant.

I do have a question:

1. What is the difference between the top slider and the bottom slider? So far I have only been using the top slider. However, I just noticed that this has been changing my PCI-E and PCI frequencies from stock. Should I be moving the bottom slider? If I change the bottom slider to 12/760, then the PCI-E freq changes to 26.8, which you can see is my "current" stock setting. PCI freq does not change with either slider and is different from my stock setting.

2. If I leave the bottom slider at 11/760 and bump the top slider to 200.2MHz, and click SetFSB, the following happens:



The current PCI freq goes down to 8.9?

Thanks for your help moral hazard it's appreciated, but of course don't feel obligated to continue helping if you don't want to

 

I'm looking for your datasheet now. I don't think our bits are the same.

Your problem might be that your PCI freq isn't locked (myne is).

I'll let you know if I find the datasheet for your PLL, give me 2 mins.

 

I can't find the datasheet.

Can you try changing the bits one by one until the PCI freq goes down?
Then when you find the bad bit, change it back.

 

OK will try that

 

Don't want to jinx it, but no BSODs for the past 100 hours! Currently running @ 2.45 so a 450MHz overclock. The only thing I have changed is adding -p27 to my SetFSB command line so that it does not change my stock PCI-E frequency (don't know why it was changing it by default).

Apparently this has made all the difference. Thanks for the help in this thread guys. If anyone else with a G51 has experienced BSODs with using SetFSB according to the G51 Optimization thread directions, try this.