killed a cpu

[OhYou_]

after 5ish years, I think I finally killed one of the two cpus in my pc from running a 40% overclock
rest in peace x5680 i guess, 1.55v vcore till it dies kek.
I think its the IMC that died though, I though tit was ram a week ago but today it got worse and wouldnt even boot with stock clocks and good ram.
Disabled one cpu and no boot, enabled it and disabled the other, boot and work just fine.

What are my chances its the cpu, or its the vrm? I'm thinking cpu since swapping ram fixed it a week ago

[da na]

5 minutes ago, OhYou_ said:

after 5ish years, I think I finally killed one of the two cpus in my pc from running a 40% overclock
rest in peace x5680 i guess, 1.55v vcore till it dies kek.
I think its the IMC that died though, I though tit was ram a week ago but today it got worse and wouldnt even boot with stock clocks and good ram.
Disabled one cpu and no boot, enabled it and disabled the other, boot and work just fine.

What are my chances its the cpu, or its the vrm? I'm thinking cpu since swapping ram fixed it a week ago

if you think it's the VRM, why not swap the chips' sockets? Seeming like CPU though.

[RONOTHAN##]

4 minutes ago, OhYou_ said:

What are my chances its the cpu, or its the vrm

Considering how when VRMs fail, they tend to cause the power supply to trip short circuit protection and the board won't stay on for more than a second, it's probably the CPU. 

[OhYou_]

3 minutes ago, da na said:

if you think it's the VRM, why not swap the chips' sockets? Seeming like CPU though.

no paste
I may do that tomorrow, I order another cpu too because $12

Just now, RONOTHAN## said:

Considering how when VRMs fail, they tend to cause the power supply to trip short circuit protection and the board won't stay on for more than a second, it's probably the CPU. 

thats what I would think too, but I wonder if instead a cap or two have failed and the voltage is now very noisy and causing problems, since the caps smoothe that out.
idk replacing the cpu or swapping sockets is easy test first

[RONOTHAN##]

1 minute ago, OhYou_ said:

thats what I would think too, but I wonder if instead a cap or two have failed and the voltage is now very noisy and causing problems, since the caps smoothe that out.

That's still relatively unlikely, as Intel does bake a lot of voltage tolerance into the spec of the CPUs that it would be incredibly unlikely for it to get to the point where the CPU doesn't even boot assuming the board isn't bottom of the barrel and had OK-ish voltage regulation when it was new. Now, that sort of thing reducing the maximum clock speed by 200-300MHz would be totally possible, but it causing the system to not boot is pretty unlikely. 

[OhYou_]

13 minutes ago, RONOTHAN## said:

the point where the CPU doesn't even boot

so like, it posts perfectly fine, no issues posting and getting into the bios and stuff, but as soon as it begins loading an os, immediate crash. even a usb stick boot disk, anything just immediately reboot.
its a pretty high end board, a evga sr-2, has like a billion vrm power stages and whatever.
sounds like I shouldnt be too worried about it then

just kinda worried because its all fun but the value of these boards keeps going up and stuff, like one sold recently for 1g lol

[Somerandomtechyboi]

3 hours ago, OhYou_ said:

so like, it posts perfectly fine, no issues posting and getting into the bios and stuff, but as soon as it begins loading an os, immediate crash. even a usb stick boot disk, anything just immediately reboot.
its a pretty high end board, a evga sr-2, has like a billion vrm power stages and whatever.
sounds like I shouldnt be too worried about it then

just kinda worried because its all fun but the value of these boards keeps going up and stuff, like one sold recently for 1g lol

god damn thats not just high end that a top end collector board cause overclockable dual socket 1366

 

try a diff cpu and see if its the chip that has failed, better hope its the chip that failed cause those sr2 are rare af and itd be pretty sad if it was the board and not some next to worthless xeon

 

1.55v is about what id personally daily for 32nm westmere/gulftown, maybe more like 1.6v, just that i dont have a xeon w or i7 since the x e l xeons have locked imc multi (stuck at 10) and cpu oc is not really that fun when you compare it to 3000+ ddr3 shenanigans

Spoiler

heres the best i can do on bloomfield for now atleast on aircooling but a w36xx or i7 970/980(x)/990x should be able to do 3600+ on air so im just saving some money to buy one since im not just gonna spend a large portion of money on one of these instead of multiplying my existing cash, itd be fun to do an unlimate x58 build with a binned gulftown and 3000+ (ecc) ddr3

 

1.55v is still a little on the high side atleast for 32nm so i guess 4 years isnt too unexpected but i havent done any real long term test on 32nm, highest ive ever gone is only around the 1.7v range for vcore and vtt just for high freq bios boot shenanigans and the only time i tested proper high voltage is when i ran 2.46v through my garbage 1gb f die stick with vtt at 1.48v on prime95 largeffts for 8 hours just to see if the 0.5v ∆ rule is bullshit (it is), degradation is temp and voltage combined so high temp maybe youll see degradation at 1.5v but lower temp might be fine all the way to 1.7v

 

maybe ill buy an e5620/40 or another x5650 and just torture it see what kinda cpu freq i can hit maybe even max bclk for ram freq alongside some degradation testing cause id rather not degrade a good sample if im not trying to do a max freq run yet

[freeagent]

1.55v is ok if you can cool it.

[OhYou_]

7 hours ago, Somerandomtechyboi said:

god damn thats not just high end that a top end collector board cause overclockable dual socket 1366

 

try a diff cpu and see if its the chip that has failed, better hope its the chip that failed cause those sr2 are rare af and itd be pretty sad if it was the board and not some next to worthless xeon

 

1.55v is about what id personally daily for 32nm westmere/gulftown, maybe more like 1.6v, just that i dont have a xeon w or i7 since the x e l xeons have locked imc multi (stuck at 10) and cpu oc is not really that fun when you compare it to 3000+ ddr3 shenanigans

  Reveal hidden contents

heres the best i can do on bloomfield for now atleast on aircooling but a w36xx or i7 970/980(x)/990x should be able to do 3600+ on air so im just saving some money to buy one since im not just gonna spend a large portion of money on one of these instead of multiplying my existing cash, itd be fun to do an unlimate x58 build with a binned gulftown and 3000+ (ecc) ddr3

 

1.55v is still a little on the high side atleast for 32nm so i guess 4 years isnt too unexpected but i havent done any real long term test on 32nm, highest ive ever gone is only around the 1.7v range for vcore and vtt just for high freq bios boot shenanigans and the only time i tested proper high voltage is when i ran 2.46v through my garbage 1gb f die stick with vtt at 1.48v on prime95 largeffts for 8 hours just to see if the 0.5v ∆ rule is bullshit (it is), degradation is temp and voltage combined so high temp maybe youll see degradation at 1.5v but lower temp might be fine all the way to 1.7v

 

maybe ill buy an e5620/40 or another x5650 and just torture it see what kinda cpu freq i can hit maybe even max bclk for ram freq alongside some degradation testing cause id rather not degrade a good sample if im not trying to do a max freq run yet

well if you are getting 1700mhz on ram per stick,how does that work with triple channel?
my stable oc was 187-188mhz bclk, 24 cpu multi for 4.5ghz and retaining the 10x multi for ram. 1.55 and 1.58 vcore, but booted it is an even 1.53 on each so idk. I threw 1.58v at each vdimm just because, not sure what my imc voltage was or what its even called, I think I just kinda bumped everything up to 1.4 so prob that

[freeagent]

I ran QPI/VTT at 1.4v for years, it was fine. 1.55-1.58 is getting a bit steep, should be just for benching. 

[Somerandomtechyboi]

8 hours ago, OhYou_ said:

well if you are getting 1700mhz on ram per stick,how does that work with triple channel?
my stable oc was 187-188mhz bclk, 24 cpu multi for 4.5ghz and retaining the 10x multi for ram. 1.55 and 1.58 vcore, but booted it is an even 1.53 on each so idk. I threw 1.58v at each vdimm just because, not sure what my imc voltage was or what its even called, I think I just kinda bumped everything up to 1.4 so prob that

nah thats just a frequency run with a single stick, this particular chip has a meh 2nd channel so doesnt post as high freq in the 2nd channel when compared to my 2nd best which can bios boot probably windows boot 3232 or maybe slightly over, vtt at 1.94v btw and yea seems like they degrade somewhat quick at this vtt which i find a little odd as my i7 930 has no degradation issues at 1.86v and still hits 3300 even after quite a few runs

 

if i really wanted to run stable then maybe around 3000-3100 on that 2nd best xeon chip but depends on where it starts flaking out vtt wise cause that i7 ive only run it at 2900-2950 stable with 2 sticks due to 1.7v vtt limit and it would suck if im limited to only 1.7v, probably should get some sticks that are lighter on the imc but id rather something that doesnt need cold like psc cause its more funny to run absurd freq purely on air

 

as for triple channel sadly x58a ud3rs have garbage third channel that flakes out above 2800 but even if that werent an issue i cant run 1:1 uncore with 3 sticks for whatever reason, uncore needs to be atleast 3 multis above ram multi otherwise no post and stuck on c3/c5 code, and no windows boot either regardless of uncore ratio if it isnt 2:1 =(  even if it did boot it wouldnt be stable as anything between 1:1 and 2:1 is not stable whatsoever and will instacrash prime95, so probably just gonna have to buy a gulftown if i wanna toy around with high freq triple channel alongside a better board but x58a oc are unobtainium and the common rampage boards are trash due to forced uncore ratio so gonna look for alternatives and so far theres the evga ftw3/classified, foxconn flaming blade, and the gigabyte g1 assassin

 

1.58v is really low volt for ddr3 but i guess thats to be expected if you arent even running 2000, my 3300+ freq runs need 2.3v and stable at ~3000 needs somewhere around 2.2v which i can finally test cause i got a better chip, but vcore at 1.55v seems excessive for only 4.5ghz so yea maybe you should look for x5675 instead cause afaik theyre usually better bin on average doesnt mean that you cant get a golden x5680 or x5650 or something but youll get better chips on average, vtt 1.4v is also really low so id just default to running vtt at 1.5-1.6v depending on how much uncore scales for 32nm which is what i did with my old x5660 but cant say for sure degradation wise as i saw none cause short oc sessions with no high vtt (>1.6v) though for bloomfield my vtt ranges anywhere from 1.7v - 1.98v depending on how far that particular chip can scale and i havent seen any degradation at <1.9v vtt on my i7 930 and i have done some pretty long oc sessions on that thing

 

only thing id suspect here is vcore and depends on what kinda temps you were running cause high temp and degradation will be quite abit quicker at this kinda vcore

[OhYou_]

finally put in the new cpu and yeah it was just that.
havent tested 4.5ghz yet because I didnt have enough thermal paste so I just tried to reuse it and it hasnt gone well. Does 4.3 all core which is better than not booting stock