Possible mobo/cpu damage after 1st Dry Ice session

[Nicholas Peyton]

Some guys were saying they were only doing grease/Vaseline etc in ram slots... kind of wishing I had done it now...?

here's what I did do though: (if anyone can add or subtract to what I've done here I'm all ears) ;-)

after spraying LET

Adding Armaflex insulation

filling in gaps between armaflex and edge of CPU (is this where I went wrong/is there a better method)?

adding kitchen roll as extra precaution

======
bottom of pot before insulating it

insulating bottom with armaflex & sealing edges with eraser

=======


pushing kitchen roll off IHS (while trying to dig it into the eraser at edges - to get a clean mount) << hope that makes sense

How it looked after removing the pot & tearing some kitchen roll away (edges of socket still looked pretty well sealed)?

Regarding the bottom of the board; I covered everything within 6 inches of CPU socket with waterproof duct tape then sat the whole board, on top of a layer of Armaflex insulation.

Got a good 6 hours first day, went to bed, then 4 hours the next day. Never removed pot in between. System still ran flawlessly up until about 12hrs after the dice session, after 2nd day. Then started getting black screens/odd restarts then CPU died.

Board is perfect, still going strong on my old AMD Phenom X6 35T (ripped out an old broken DELL machine).
4 days & counting. So aye definitely the CPU.

Surprised we're still seeing 100 bux for a 'used' 8350'. I'm obviously scouting for a new one

Alternatively I could chuck a 9590 in my board (although not officially supported).
My boards max rating is 140W
9590 requires a 220W board.

Yet my daily O/C on my 8350, caused those wattages anyway.
Stock 8350: 4.0GHZ (4.2 boost)
My overclock: 4.8GHZ

Stock 9590: 4.7GHZ (5gz boost)

If anything I'd be consuming less power as the chip would be running at 4.8GHZ at a lower voltage.

ASRock disabled support in future BIOS updates, but I could flash (downgrade to an earlier BIOS) and it should take the 9590. Used prices are the same, if not sometimes even cheaper. And I'd be 3 bins higher. ;-)

 

[jaggerwild]

I use the clay to build up around the CPU(evenly) as the POT overhangs on the edges. BUT, before I clay around the socket(you'll notice)when the CPU is SEATED in place there are tinny gaps around it(between socket edge and CPU edges) tape with masking tape there(over hang the tape off the socket sides and ontop the CPU edge). It will stop any ozzing of the clay into the CPU socket when the POT is seated. Then I clay over the masking tape, then I cover it with armaflex (rubber matting) when the POT is SEATED on the CPU it all works perfectly. I thought for sure that board was dead, so I assume it was a dead CPU?

 

[Nicholas Peyton]

I use the clay to build up around the CPU(evenly) as the POT overhangs on the edges. BUT, before I clay around the socket(you'll notice)when the CPU is SEATED in place there are tinny gaps around it(between socket edge and CPU edges) tape with masking tape there(over hang the tape off the socket sides and ontop the CPU edge). It will stop any ozzing of the clay into the CPU socket when the POT is seated. Then I clay over the masking tape, then I cover it with armaflex (rubber matting) when the POT is SEATED on the CPU it all works perfectly. I thought for sure that board was dead, so I assume it was a dead CPU?

Yes, I'm speaking to you now via the board

Definitely the CPU. Since I put the old Phenom in I've ran flawlessly for days (including prime95's).

Next time I'll tape from edges then, as suggested. Thanks

 

[Nicholas Peyton]

Anyone ever have an RIP chip come back from the dead?

 

[MrGenius]

Seriously? No. Can't say that I have. Why? You got a Jesus chip?

 

[FireFox]

You got a Jesus chip?

Nice one.

You made my day.

 

[EarthDog]

Anyone ever have an RIP chip come back from the dead?

if it came back from the dead... it was never dead.

On the third day, chips do not rise.

 

[Nicholas Peyton]

Story about my "Jesus" chip!

System would NOT boot with, what I thought was my *fried* FX-8350 after a Dry Ice session.

Swapping another chip in system would boot, FX-8350 back - no boot... several times (over several days). I even got deflated over it!

I chucked the 8350 away in an old mobo box somewhere (never even took any care, I literally *threw* it in (thinking it was definitely dead).

So was using my old PHENOM chip temporarily. (as some of you already know).

However I made a silly mistake, was messing about and stupidly forgot I'd left my Water Chiller running (which I've modified to bypass Thermostat so it will continue to go subzero if left alone).

Anyway I came back to the machine and noticed the water temp indicator on Chiller at 1.2c

Jumped for the 'power off button'.

My old PHENOM X6 35T (I was using temporarily in board, I hadn't bothered insulating) as I had no plans to try to overclock it....

Anyway I quickly removed the PHENOM and sat it on a hot radiator. (just in case)

...got bored waiting and for some reason (for life of my I don't know why but through my boredom I thought "hmm.. the 8350's been sitting for 4/5 days now.. lets give it one last go. I knew it wasn't going to work!

But it did. ?

System booted.

Then whats even more amazing is this:

The day after it came back to life (the 26th).

I'd taken the heatsink off to do the insulation around socket (for my water chiller - as I had it set up before).

I fell asleep.

Woke up and switched the PC on, went downstairs for a drink.

Came back upstairs and realised I hadn't put the heatsink back on, before I fell asleep.

So chip had been running for a few minutes (with NO heatsink/block). Bare chip.

Powered it off. (in panick mode) and removed it from socket to assess the damage. (Nothing visible).

Put chip back in, wouldn't boot.... put old phenom back in & it would boot. Done this several times. I was convinced I'd fried it.

In my "frustration" and "panic" I bent CPU pins. (I mean FFS -- like I'd not been through enough already)!

So spent last 2 nights straightening all the CPU pins out (on a chip that I had no idea was fried or not). (and the first night I attempted this, pre-research, I made it worse)! Bent even more!

Finally got them straightened out enough and only JUST enough as I had to apply a little pressure from a certain angle to get the CPU into the socket. <-- literally took reading glasses AND a magnifying glass. lol (For those who only know intel, the pins are on the chip with AMD, not the mobo).

Anyway it just booted. (literally 10 mins ago). I am still so surprised.

This chip is indestructible. lol

 

[dorsetknob]

Scottish Voodoo Baron Samedi wears a Kilt

 

[Nicholas Peyton]

Lol.

Well with the money I was going to use to buy a new CPU.

I've decided to treat myself to this instead: (was only 100 bux off Ebay).


ASRock 990FX Extreme9 (ASROCK's biggest flagship AM3+ board / 12+2 phase / support for 220w CPU's) <-- when it was first released; 12 phases was almost unheard of on a mobo.

Not sure about those 12+2 phases though? I've read they are using doublers on this board..

• But how much of an efficiency hit do you take, using doublers vs true phases? (I.E Is there some benefit to using doublers and having 12 phases (with doublers) as opposed to 8 normal phases)?
• My current mobo (is 8 phase & only rated for CPU's up to 140w
• Is comparing 'sizes' of chokes a good way of making a decent guess?

My current board (pictured top of this page): ASROCK 990FX Killer (8+2 phase). But I can't seem to find info if thats a true 8 phase or if they're also using doublers. If they are. Then this is a massive upgrade in power delivery?

I should expect higher/more stable overclocks?

Can't wait to see if claims about voltage/overclock stability on a bigger board is true.

I need 1.55v (1.5v +50mv offset) for 4.8GHZ prime95 stable (water). on my ASROCK 990FX Killer (8+2).

I'm very eager to see if the bigger board helps achieve lower voltage for the same clock.

Not denying what anyone's said, but having the opportunity to test myself 1st hand, is exciting.


I've decided to leave ZEN alone, for a few more months.

 

[EarthDog]

There was a thread talking about just this issue at ocf....

Let me see if I can dig it up..


As far as the bigger board lowering voltage, that depends. There will be some variance, but it's doubtful it will be significant...Unless your previous board was truly a dud. The board to get for ln2 fx was CHVZ, and the Sabertooth IIRC.

 

[Mr.Scott]

The Asrock board is junk. I killed 2 of them on my chiller with an 8320 @ 5.6
Their A/D hybrid VRM is overrated.

 

[infrared]

Nicholas, I know I'm not your favorite person after that thread... but I think the misunderstanding went both ways. Nobody was against your LN2 and extreme overclocking runs... (in fact I'll clear that up right now. I LOVE to see people doing this stuff, and look forward to trying it one day myself). It was the unfeasible warranty scheme you'd suggested that most didn't agree with, and you've totally brushed that under the rug, even in this thread you carefully hide that when you talk about it. I'm all for your idea of having more easily mod-able graphics cards and stuff, that would make things a lot easier. Anyway, what's in the past is in the past. I'm glad your motherboard is ok, gutted about your 8350, and hope to see you breaking more records soon!

On a side note, you say it's harder to overclock cards each year... back when I joined the forum the only way to up the voltage on x800 series ATI cards was with physical voltmods, and I loved doing that! We went though a nice period where you could do a lot with a bios mod, like on my old titans, I can go up to 1.3v/350w instead of the 1.16v/250w limits. I don't have any experience with the really new cards but if they are more locked down then that's not good :/

 

[fullinfusion]

There was a guy on here years ago that did hard core cooling and always insisted on using a magical liquid to insulate the board to dice and ln2. It's what the guys in the movie buiz use for creating cosmetic looks and such. Don't shoot me but I believe it was liquid latex that you'd pour around the socket and other sensitive areas. Once dry it protect the board and after the run was done it peel off pretty darn easy.

I'm sure someone here remembered the guy and what it's actually called.

 

[EarthDog]

Conformal coating, or LET (Liquid Electrical Tape). Both common methods of protecting the board from moisture.

 

[infrared]

When's your next dice/ln2 session EarthDog? I missed the live stream of your last one

 

[basco]

nick you might not see better or lower voltage with more phases but when you go extrem then there could be more headroom.

 

[sneekypeet]

There was a guy on here years ago that did hard core cooling and always insisted on using a magical liquid to insulate the board to dice and ln2. It's what the guys in the movie buiz use for creating cosmetic looks and such. Don't shoot me but I believe it was liquid latex that you'd pour around the socket and other sensitive areas. Once dry it protect the board and after the run was done it peel off pretty darn easy.

I'm sure someone here remembered the guy and what it's actually called.

Dragon skin?

 

[Nicholas Peyton]

Nicholas, I know I'm not your favorite person after that thread... but I think the misunderstanding went both ways. Nobody was against your LN2 and extreme overclocking runs... (in fact I'll clear that up right now. I LOVE to see people doing this stuff, and look forward to trying it one day myself). It was the unfeasible warranty scheme you'd suggested that most didn't agree with, and you've totally brushed that under the rug, even in this thread you carefully hide that when you talk about it. I'm all for your idea of having more easily mod-able graphics cards and stuff, that would make things a lot easier. Anyway, what's in the past is in the past. I'm glad your motherboard is ok, gutted about your 8350, and hope to see you breaking more records soon!

On a side note, you say it's harder to overclock cards each year... back when I joined the forum the only way to up the voltage on x800 series ATI cards was with physical voltmods, and I loved doing that! We went though a nice period where you could do a lot with a bios mod, like on my old titans, I can go up to 1.3v/350w instead of the 1.16v/250w limits. I don't have any experience with the really new cards but if they are more locked down then that's not good :/

Thats okay

My idea kind of came out all wrong. I was never expecting manufacturers to pay for us to do what the hell we wanted. I was rather trying to figure out a way to put more money in their pockets upfront to help them give us a bit more freedom in future releases. (But also without costing users who don't overclock more). The same way we enjoy pairing an enthusiast mobo + chip without lockdowns.

I never thought about the options I provided in the POLL properly. Which never helped me either.

And I admit i never thought it all through properly. Before posting.

Anyway as you say, I'm glad thats behind us. lol.

And lastly, I'd love to know electronics to do hardware mods (on older cards).



----------------

nick you might not see better or lower voltage with more phases but when you go extrem then there could be more headroom.

Aye hopefully.

Most I got on the FX Killer mobo is 5900 mhz

The killer also has no 'load line calibration' setting and doesn't allow voltage over 1.7. (and to get 1.7 you have to set a 1.55v (maximum on board). Then use their maximum "voltage offset" setting" of 150mv.

Whats the max voltage you guys would go to on DICE? (AMD FX 32nm)

My new extreme9 obviously allows up to 2.0v and has the LLC setting. Which I've never had a chance to play with before.

Thanks!



.

 

[basco]

llc will help you in setting 1.50volt and getting 1,50 volt under load where a motherboard without it could fall from 1,50 as low as 1,40+- under load
and max volt depends on heat output from chip to chip and you know there are different limits for a max cpu frequenzy pic or a 3dmark or wprime run where all cores are pushed but amd is known for taking the volts so yeah we needed 1,9 volt to get a max mhz shot at 8070mhz.
you will notice too much volts when you run a mutlithreaded bench and temp drops to much like 20°- shotly after bench started

 

[EarthDog]

If you are seeing .1 vdroop, your board has a problem.

 

[Nicholas Peyton]

If you are seeing .1 vdroop, your board has a problem.

You mean like this? (pictured below) (on my ASROCK 990FX Killer at stock.)
-V.core is:
1.288v
&
+50mv (asrock has a setting in BIOS called "voltage offset" with a "drop-down-list" of 0, +50, +100 or +150). Default is +50mv
= 1.338v. But prime95 & I see as low as 1.152v (although it never crashes/errors/warns at stock).
.

-My mobo is still on my desk (out the case) with a 3000RPM fan sitting on top of the VRM heatsink. (which is actually "cool to the touch".
-I replaced the thermal pads on it, two weeks ago. After noticing ASrock had missed two mosfets out by cutting thermal pad too short.
-Had the board for about 2 years now.



P.S.

This board has no LLC setting.
My new board will ;-)

 

[basco]

exactly that.
so with llc the guessing to how low it will drop is gone.

 

[Nicholas Peyton]

exactly that.
so with llc the guessing to how low it will drop is gone.

Excellent. thx

 

[EarthDog]

Just to confirm cpuz shows voltage is the same as the other program?

But yeah, if your board sags .1V like that...it's probably time to get something better... CHV or sabertooth are solid.