Boxxen Down and it can't get up

J

jfb9301

[H]ard DCOTM x4
2FA
Joined
Jan 11, 2005
Messages
1,757
More woes for me and the recent [H]ardbox that I added to my farm.

I finally found the source of the "run away" process that I had. Seems that as the Voltage Regulator got hot, it would cause the board to throttle. This was indicated by the "run away" process. Adding a cooling fan to the VRM heatsink caused the problem to go away, and allowed me to get to 3.6 GHz on air. CPU is a D0 920 and I was using 180x20 to get there.

Today, the UPS man dropped of a shiny new Antec Kuhler 920, and 12G of Mushkin DDR-3 1600C8 ram.

I had hoped to be posting a happy thread with pics of the boxen folding.

The install was tight, but flawless. Booted it up and was folding just fine with the ram and cooler. CPU was running around 64C (10-13C cooler than the heatsink I won) I left it go and went back to playing some Skyrim.

I noted FHM.net showed the machine offline. That's not good. Machine, hard locked. Reset. No post. Post Code 88 (undefined)

Reset. Won't reset.
Hard off and on>
No post, no code.

MB manual here for the post codes http://www.dfi.com.tw/DFI-Lanparty/support/MBManualDownload.jsp?selectManualItemId=641

88 seems to indicate a failed or errored CMOS
blank seems to indicate failed MB or PSU ATX5V.

I'm open to suggestions.
Even if it was bad vid, bad ram or bad CPU it should still have post codes. Not bad vid, I used a bench card to eliminate that. I can use some bench ram, but I don't have a bench CPU.
Is there any way to test the PSU with a volt meter? All 5V connections get 4.9906v with the PSU jumpered to run.
I really don't want to disassemble my other rigs to test this one, I suppose I could take Gamma off-line for it's PSU, or get Delta's PSU and attempt to boot Gamma with it. Omega (while another socket 1366) is fairly off limits.

Wondering what Gamma, Delta and Omega are? They are in my sig.

 
April has not been nice to me :(

Kendrac, the new board is awesome... I managed to get the BCLK north of 180 :), but not all the way to 200, it's in my son's room, so I don't have as good access to it, and it does not get the best AC, so I'll settle for 190x20 for a solid 3.8 GHz. the hardfolder, Delta, is a solid performer for me.

Yesturday morning, my HTPC Gamma went down hard :(.... so that's a 2600K and a GTX460 not folding. It too had a post code reader installed on the MB which indicated that the motherboard was having an initialization problem. This came after CPU and Memory checks. Googling the error indicated that it should be a MB problem, but might also be CPU related. I've been having long term issues with Gamma getting slowly more and more unstable and having to drop the multiplier to get it folding again. Troubleshooting, I bought a brand new MSI Z77A-GD55 from the egg. Surprisingly, it showed up on my door step today (3 day shipping overnight? kudos to UPS on that one). Installed the MB and no post. Looks like my 2600K has given up it's ghost for FAH. I can't afford another CPU until after payday, so fixing this box is on hold for a bit.

I want to keep the new MB, so it looks like I am going to be selling my older Asus Maximus IV GENE-Z (as it seems to have not been the problem).

Can anyone loan me a 2600K or compatible CPU? I could even trade plus some cash the old MB.

p.s. once I get this all sorted out, I'll utilize my thread necromancy skills and update the "died in the line of duty" thread. And this one with some pics of the hard folder. Which is a beauty. Simple and just works.
 
Cant get it up? Dont they make a pill for that?

Sorry, had to. leaving now
 
Are you sure you killed the 2600K? What volts was it running at? Killing a CPU is actually pretty difficult, believe it or not.

Also if you try to post a board with CPU ONLY, no ram, and no video, you should get a post error code for having no ram (3 beeps) or no video (1 long 2 short, I believe). Then add parts one by one.
 
To keep the overclock, the multi has been creeping down and the volts creeping up. the last attempt was 4.6 GHz at 1.415 with loadline calibration set to 100% (no droop at all) with VRM OC protection disabled. Most of the CPUs life has been 4.6 @1.400v and steep load line calibration settings. I had crept the multi down to 40x and had been having issue there too. It no longer folded stable with "optimized defaults." Setting load line calibration to 100% got be back to 45x but after 2 days that became unstable. I really think I was pushing my luck with this one. Beer and OCing is a poor mix.

No ram gets 3 beeps. No video gets the same response as with video, not surprising as the new MB has on-board video via the CPUs GPU. I got the same basic responses out of the GENE-Z, no ram, got me a RAM fail post code. I didn't try no video, but I swapped video cards and still get same.
 
Yeah, can't really do no video with some of the newer stuff, but if you get 3 beeps with no ram and then add one stick of ram, I would think it should post.

On the i7 920 box I just put together last night, I had to have the ram in the second slot for each channel (farther from cpu), it wouldn't post with sticks in the first slot for each channel (closer to cpu), I thought that was a bit odd.
 
extide said:
Yeah, can't really do no video with some of the newer stuff, but if you get 3 beeps with no ram and then add one stick of ram, I would think it should post.

On the i7 920 box I just put together last night, I had to have the ram in the second slot for each channel (farther from cpu), it wouldn't post with sticks in the first slot for each channel (closer to cpu), I thought that was a bit odd.
Click to expand...

On the 920 I had the same with the ram on the first mb. Figguerd it out, got it folding, then it died. 920 is going at 18KPPD as I type.

Did the one stick trick, no post either. That was odd too, as instead of no post and no beep, either stick in dimm1 got me a non-post reboot. Two sticks, just halts with no reboot. I've been installing ram in dimm 1 and 3, per the manual, but I'll have to give 2 4 a shot and see what it does.
 
Glad to see that I was able to solve at least one of the problems.

:(
 
K,

You're awesome :D

another symptom that I thought of, right before it went down hard, I was trying to figgure out why it would not go beyond a 16x clock. I had been disabling anything "green" in the BIOS, but it would not engage any kind of high multi.
 
This is quite odd, been a long time since I seen a CPU killed lol, but for some reason it just seems odd
 
Loaner 1155 CPU anyone?

As of payday, I will be able to afford a 2600K to put in the machine. Though it would be nice to know if the $300 will actually fix the problem.

Come on Intel, start selling some IB cpus...... push that price down.
 
jfb9301 said:
Loaner 1155 CPU anyone?

As of payday, I will be able to afford a 2600K to put in the machine. Though it would be nice to know if the $300 will actually fix the problem.

Come on Intel, start selling some IB cpus...... push that price down.
Click to expand...

You could get a cheaper 1155 like a low end i3 to do the testing with - not so much to fork out then if it isn't the cpu
 
Oh man, load line calibration worries me. If it was that agressive, it was trying to keep your 1.4V UNDER LOAD, so when it drops off of load, the voltage spikes up. Same as the RPM of an engine on a dyno. We really cant tell just how high it could have shot up, and I dont think it would have lasted long hitting high voltages even with no load
 
But then I'd just end up with an extra CPU.

Is a thought though.
 
Untitledone said:
Oh man, load line calibration worries me. If it was that agressive, it was trying to keep your 1.4V UNDER LOAD, so when it drops off of load, the voltage spikes up. Same as the RPM of an engine on a dyno. We really cant tell just how high it could have shot up, and I dont think it would have lasted long hitting high voltages even with no load
Click to expand...

I was following ASUS's OCing guide for the 2600K that they posted here. Ended up running settings that should have reached 48x just to be stable at 46x. I figured if I kept temps in check I could avoid CPU damage
 
Yeah I would stick with ~50% LLC. With 100% you can actually RAISE the voltage under load and yes, WAY overshoot coming off load.
 
See what happens when you set your voltage in the BIOS without load line calibration, is it just sets the no load voltage to whatever you set it to (sometimes a bit less depending on the voltage regulation). So if you set say 3V in the BIOS it might be 3V at idle, under load in may be 2.8V. Load line calibration tries to make it always 3V, so idle you might be at 3V, and for load it would do its best to keep it at 3V. When the load drops, the voltage can spike up to say 3.1V or 3.2V, but then will quickly return to 3V. However if that 3.2V was high enough to damage some transistors, the CPU will begin to degrade.

I just picked some arbitrary numbers to try to show how it works.

Essentially it is better and more predictiable to just set the base voltage compensated for load rather than running load line calibration.
 
link to the guide

So the consensus is my 2600K probably died due to aggressive LLC (it was set at 75%) most of the CPU's life.

So now the shopping question..... 2600K @ $315 on the egg or 2700K @ $330?
 
Well if you buy a 2600K and it doesnt fix it then you better buy 2 new mobo's and build a box out of each cpu!
 
Intel is RMAing the CPU.

They did advise me that I had been OCing the memory controller by using DDR3-2133 RAM and that I shouldn't do that, I should use 1333 ram.

Never asked me about multi OCing.

CPU is in the hands of FEDEX now. We'll know more when I get my CPU back.

$15 is cheaper than $315.
 
jfb9301 said:
Intel is RMAing the CPU.

They did advise me that I had been OCing the memory controller by using DDR3-2133 RAM and that I shouldn't do that, I should use 1333 ram.

Never asked me about multi OCing.

CPU is in the hands of FEDEX now. We'll know more when I get my CPU back.

$15 is cheaper than $315.
Click to expand...

Wow, that is pretty nice! I wish you luck jfb9301! Hopefully that is your problem.
 
RMA cpu showed up today.

Win7 hated the swap from z68 Chipset to z77. I can't get any monitoring program that accesses the SM Bus to function. CPU-z also accesses the SM Bus, so no go on that too.

Real Temp accesses the TJ directly, and works just fine. Folding at 4.5 GHz, 1.385v with LLC set at 4 (on a scale of 0-8, so I hope around 50%)

I'll post some more after some stability testing.

Hopefully with some pr0n
 
Interesting, I moved my current windows 7 install on the 3930k from an X58 system, it came across just fine, X58 -> X79.
 
And..... some pr0n

This is the [H]ardbox that I won from the contest........

i7 920 on a MSI mainboard.... thanks to everyone who donated.

IMG_0751.jpg


For those with sharp eyes, yes that is an AMD CPU fan cooling the NB
 
Last edited:
extide said:
Interesting, I moved my current windows 7 install on the 3930k from an X58 system, it came across just fine, X58 -> X79.
Click to expand...

Yea, in my experience updating MBs is sometimes challenging. I haven't had much trouble recently, but this rig fought for two hours before I got windows functioning enough to try to fold.
 
jfb9301 said:
Yea, in my experience updating MBs is sometimes challenging. I haven't had much trouble recently, but this rig fought for two hours before I got windows functioning enough to try to fold.
Click to expand...

The one key to remember is that there is only one driver that windows actually REQUIRES in order to boot. That is the disk controller. If you are moving between 2 boards that use the same sata drivers, OR use a pcie card then you can always get it done.

There are plenty of crappy things that happen and windows really hates this, so I wouldn't really recommend it but it's definitely doable if needed.
 
Simplest way to switch motherboards:

1. Change the disk controllers in device manager back to the default microsoft ones.
2. Uninstall any propietary mobo drivers from Add-remove programs
3. Reboot, check to make sure you are running plain M$ drivers
4. Swap board
5. Boot up and let it install all the drivers it ca, cancelling out any that it can't find drivers for and rebooting where required.
6. Run whatever driver CD came with the board, reboot where asked
7. Probably have to re-register windows after.
 
One way that is guaranteed to work, even going from one totally different chipset to another (for example AMD Nforce to Intel P4)

1. Install a PCI/PCIe SATA controller, DO NOT HOOK YOUR DRIVES UP YET
2. Boot to windows and load the driver for the card
3. Move your boot drive to the sata card
4. Boot up and verify al is well
5. Now move the hard drive (or an image of it) to another box and use that SAME SATA card, with the boot drive on it.
6. Boot from the sata card on new mobo.
7. Allow windows to install the SATA controller on the new mobo
8. Move boot drive to onboard SATA on new mobo
9. Remove pci SATA.
10. boot up and you are good to go, you may remove the pci card driver, along with the drivers for the old mobo.

To do this you need to add the environment variable DEVMGR_SHOW_NONPRESENT_DEVICES = 1 and then open up device manager, and select "Show Hidden Devices" -- this will now display hardware no longer in the system -- 'Ghost Drivers' -- you will see tons for the old mobo. All of them are safe to delete EXCEPT some of the ones in the Sound Card section, and also don't touch the stuff in the Non-Plug and Play Drivers section.
 
You must log in or register to reply here.
Back
Top