RMA? No BIOS LEDs/POST |
Post Reply | Page 12> |
Author | |
30GHz
Newbie Joined: 20 Jul 2015 Status: Offline Points: 13 |
Post Options
Thanks(0)
Posted: 04 Aug 2015 at 5:05pm |
Well there was no load when it died.
All of the settings were manual with no offsets enabled. >what were the multipliers/ratios set to? Rather high, like the maximum that has been known to be successful. But that wouldn't have killed it, it must have been the voltage, specifically seems it could have been the Uncore voltage. Max (and perfectly safe on water) VCCIn is 2.1 according to many review sites, Intel is conservative. You can see in my linked post that the guy said he got to 4.7GHz uncore on water with an 8core no less! Some guys with the PentiumG Haswell(non-E) 4core/8Thread have it up to 5GHz with 5.6GHz core. My 5820K chip was tested good for 4.5GHz @1.27v by a friend, but he didn't try beyond that conservative Vcore: so 5GHz @1.437 wasn't an unreasonable request. That 1.437Vcore is cited by many as safe, but I only found that one guy(linked post) who said uncore over 1.2V was possible, and maybe I misinterpreted that, when he really meant Vcore instead of Vuncore in that cite. It is just hard to believe that setting it to that, even if wrong killed the CPU in one sec, it never rebooted again!
Edited by 30GHz - 04 Aug 2015 at 5:09pm |
|
parsec
Moderator Group Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
|
Those voltages are high! Plus it's not just setting the voltages, it depends on the CPU VCore Voltage Mode and CPU Cache Voltage Mode settings. If you use Auto/Adaptive, that will cause the true VCore and/or Cache voltage to increase above whatever your voltage settings are. Add +Offset (if you did) to this and who knows what those voltages actually were! You did not mention the Voltage Modes used. IMO, rather bold of you to just set both VCore and Cache voltage so high at the same time, what were the multipliers/ratios set to? From the Intel 2011-v3 Processor Datasheet, what they list as the "Absolute Maximum" for Vccin, CPU Input Voltage: 1.98V. That does not include any "transient overshoot" of the voltage, meaning any short term over voltage from the VRMs, but this value allows for the maximum transient overshoot voltage specification. The standard Vccin specs are 1.47V Minimum, 1.8V Nominal, and 1.85V Maximum. The Absolute Maximum spec for DDR4 memory is 1.35V, but I've read lately of DDR4 XMP profiles setting the DDR4 memory voltage to 1.5V. As you know, it is common for experienced over clockers to ignore the Intel specs, and most of the time they get away with it. Which is one reason why you believed what you were told about the maximum "safe voltages" for Haswell-E processors. Prior to Haswell, we knew what the maximum VID (VCore) was for the processor family, it was in the specs. That was the maximum voltage that the processor could have applied to it using the standard VID control method. For Ivy Bridge, it is 1.52V. That does NOT mean the processor could survive with that much voltage applied, but gave us some idea what the maximum voltage would be when using standard CPU cooling methods. Haswell processors have the VCore created inside the processor itself with its internal VRMs. The Haswell datasheet does not include the VID range for the internal VRMs. For that we need another document, the VR12.5 specification, which is still unreleased to the public. So we are going completely by guess work and experience with Haswell, IMO. Some Haswell OC guides I've seen claim the Cache/Ring cannot be OC'd beyond ~3.6GHz with a max VCore of 1.2V. But the various 2011-v3 "OC Sockets" with more pins in them used on some X99 boards are said to overcome that limitation. Your Fatal1ty X99M Killer/3.1 board, and my X99 Extreme6/3.1 board both have the "X Series" OC Socket. So you were able to set a high Cache OC and voltage. IMO, the stock clock and Turbo speeds of the Haswell-E processors give us a hint about how safe they are with high voltages. All of the Haswell-E processors have speed specs that are below those of the Ivy and Sandy Bridge HEDT processors. My 5820K has a base clock speed of 3.3GHz. If I OC it to 4.4GHz, that is a 1.1GHz OC. If I can do that at 1.2V or less, I'm happy and done. |
|
Xaltar
Moderator Group Joined: 16 May 2015 Location: Europe Status: Online Points: 24248 |
Post Options
Thanks(0)
|
I am sorry to hear you are still having problems even after a new board. I have not actually seen a burnt CPU in almost a decade with all the safety features implemented these days. Good luck with the CPU replacement.
You can try RMA the CPU, you may get lucky and get it replaced.
|
|
30GHz
Newbie Joined: 20 Jul 2015 Status: Offline Points: 13 |
Post Options
Thanks(0)
|
http://forum.hwbot.org/showpost.php?p=375986&postcount=162Here is where he said it was safe for 1.42 Vuncore on water. But other reviews said 1.2V max so maybe that was it, seems weird that one single reboot could kill the CPU, it never recovered from that setting.
I sure hope it's not my video card or something preventing the BIOS LED from turning on. (Though it worked fine before, so the CPU setting change must have killed the CPU, I guess.) I guess I shouldn't have trusted some random post, or maybe that number wasn't uncore even though it sure seemed like it.
|
|
30GHz
Newbie Joined: 20 Jul 2015 Status: Offline Points: 13 |
Post Options
Thanks(0)
|
I tried just removing the CPU and the RAM, but then it wouldn't turn on? The fans stayed off.
So based on your recommendation that there should be a "missing CPU" dr.debug code, I assumed it was the motherboard and ordered another identical one. Same problem with the new motherboard. Monitor recieves signal but the signal is a black screen, the Fans are on, but the red BIOS led is off. So I will try ordering a new CPU then.
|
|
Xaltar
Moderator Group Joined: 16 May 2015 Location: Europe Status: Online Points: 24248 |
Post Options
Thanks(0)
|
I suggested removing the CPU because if it the CPU took a hit and its VRMs were damaged it could cause the board to fail to reach the dr debug tests. As we suspect the problem could be the board or the CPU, it makes sense to try without the CPU in the picture. There is no Dr. Debug code for a bad CPU, only for an absent one.
|
|
30GHz
Newbie Joined: 20 Jul 2015 Status: Offline Points: 13 |
Post Options
Thanks(0)
|
There is only a Killer version of 3.1 I highly doubt rebooting with the uncore would kill the CPU as there was no load, i.e. it never worked with that setting. I tried taking out the RAM and there is no Dr.Debug codes, if the motherboard was fine then it should show in either case of missing cpu OR ram. I am pretty sure if there are no lights on the motherboard that means it is dead, so I reordered the same board, but I have to pay $69 restocking now. Seems the CMOS got corrupted and can't be cleared, seems separate problem from the BIOS, since that switch doesn't help. Edited by 30GHz - 22 Jul 2015 at 6:16am |
|
Xaltar
Moderator Group Joined: 16 May 2015 Location: Europe Status: Online Points: 24248 |
Post Options
Thanks(0)
|
You stated your motherboard as X99M/3.1 is it the Killer board or one of the extreme series? It is pretty hard to kill a CPU but 1.425v on the uncore is higher than I have seen on the many OC threads I have read on overclocking the 2011 v3 platform. The other voltages you listed are not dangerous however.
The only way I can think of to determine if the board is dead is to remove the CPU and power on the system, the Dr. Debug LED panel should display a "00" error if the board is working correctly. If it does not then the board has probably died. If it does display a 00 code then try checking for bent pins and dirt on the contact pads then reinserting the CPU. Clear CMOS again before you attempt another boot at this point. If there is still no life at this point then the CPU is the fault. Good luck, I hope it is just poor contact in the CPU socket.
|
|
30GHz
Newbie Joined: 20 Jul 2015 Status: Offline Points: 13 |
Post Options
Thanks(0)
|
It's a brand new mobo, booted to BIOS once, the 1.437VCore,1.425Vuncore,2.0VCCin,1.35Vdimm settings are max safe recommended.
If the fans that are connected to the motherboard spin, it seems it is alive but I think if there are no LEDs there then it must be dead, how would the CMOS clear fix that? I tried leaving the battery out before, the CMOS jumper wasn't in the clear position for that time, I guess I can try that. The Dr.Debug should show codes if CPU was bad. UPDATE: With No RAM it still doesn't beep with the PC-speaker or headphones. LAN lights up but no dr.Debug. I hope it's not the CPU that died because (I assume dr.Debug would show up in that case) I will try RMA'ing the mobo.
Edited by 30GHz - 20 Jul 2015 at 6:57pm |
|
Xaltar
Moderator Group Joined: 16 May 2015 Location: Europe Status: Online Points: 24248 |
Post Options
Thanks(0)
|
What voltages did you raise?
I would first try a battery cmos clear: Power off the system and unplug the power cord Press and hold the power button to clear residual charge from the PSU Set the clear CMOS jumper to the clear position and leave it there Remove the CMOS battery leave the system like this for 4 - 8 hours set the clear CMOS jumper back to normal replace battery connect power cord try to power on the system This is what I like to call a "failsafe" clear and will clear any and all cobwebs from the bios. If this fails then remove the CPU and check for charring/dirt on the contact pads as well as look for bent pins. Depending on the voltage you set the CPU could have taken damage. Bare in mind that intel CPUs have the VRMs built into them these days. You can additionally try to power on the system with no CPU installed and compare symptoms.
|
|
Post Reply | Page 12> |
Tweet
|
Forum Jump | Forum Permissions You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |