Did I kill my G3258 delidding? (No, but I JUST KILLED MY 4670k!! - Need replacement resistor!)

seanzylol · Dec 16, 2015

Quote:

Originally Posted by Just a nickname

I wouldn't run 1.5v on any CPU air or water. 1.35v is the max I would do with haswell on water. 1.3v on air.
1.5v can kill cpu (and will kill cpu).

lol ive ran 1.7vcore and 1.6 vtt on my i7 920 on water for a hour for some 3d mark/super pi runs.... ITs running just fine. But on a 22 nm cpu i would stay around 1.3-1.4

Just a nickname · Dec 16, 2015

Quote:

Originally Posted by seanzylol

lol ive ran 1.7vcore and 1.6 vtt on my i7 920 on water for a hour for some 3d mark/super pi runs.... ITs running just fine. But on a 22 nm cpu i would stay around 1.3-1.4

I've read several post of novice overclocker that tried too high voltage with improper cooling resulting in the death of their cpu at just 1.5v. 1.38v on my 4790k is toasty, no way it could even run at more than 1.45v on air. Even on watercooling... unless delidded.
I've run my 750/870 at 1.55v on water and they both survived. I wasn't expecting them to live tho. High voltage will damage the gate and eventually cause the cpu to clock like crap.
Like you said, Nehalem is really not the same. 1.45v is dangerous on Haswell. I haven't tried it but it seems to be the general consensus.

seanzylol · Dec 16, 2015

Quote:

Originally Posted by Just a nickname

I've read several post of novice overclocker that tried too high voltage with improper cooling resulting in the death of their cpu at just 1.5v. 1.38v on my 4790k is toasty, no way it could even run at more than 1.45v on air. Even on watercooling... unless delidded.
I've run my 750/870 at 1.55v on water and they both survived. I wasn't expecting them to live tho. High voltage will damage the gate and eventually cause the cpu to clock like crap.
Like you said, Nehalem is really not the same. 1.45v is dangerous on Haswell. I haven't tried it but it seems to be the general consensus.

exactly why i said stick to below 1.4 on a 4790.

cookieboyeli · Dec 16, 2015

Alright, it's not surviving Battlefront (might be just the game though...Died as Han Solo on a massive kill streak!

) So I'm going to gradually step down the voltage and try to keep it as cool as possible. 90+% of the time It's near idle and I've got all power saving enabled aside from EPU mode (none of them seemed to make a difference with the overclock whatsoever).
Perhaps I can keep 4.7GHz with less voltage if I keep the chip cooler...

white owl · Dec 16, 2015

Did you get the black stuff off?
Your finger nail works. The edge of a CD works better. Any gap is a bad gap.

To spare me from reading the entirety of the thread, can you post a HWiNFO shot?
So I can see voltage and such.
Haswell clocks worse than DC but there's a plethora of things you can change to get more clock with less vcore. Like more current and using all of your boards phases.

http://rog.asus.com/254052013/maximus-motherboards/recommended-settings-for-overclocking-maximus-vi-motherboards/

cookieboyeli · Dec 16, 2015

Quote:

Originally Posted by white owl

Did you get the black stuff off?
Your finger nail works. The edge of a CD works better. Any gap is a bad gap.

To spare me from reading the entirety of the thread, can you post a HWiNFO shot?
So I can see voltage and such.
Haswell clocks worse than DC but there's a plethora of things you can change to get more clock with less vcore. Like more current and using all of your boards phases.

http://rog.asus.com/254052013/maximus-motherboards/recommended-settings-for-overclocking-maximus-vi-motherboards/

Yea I know all about the phases. I'm thinking of getting the Maximus VII Gene since it has better VRMs and 8 true phases whereas my motherboard does not. It also has a totally awesome amount of settings! I wish my board had them all! I'm probably going to remove the VRMs and add some copper shims to see if that helps any.

Here's a screenshot from the other day when I was using 1.55v @ 4.7GHz
I'm now back to 4.6GHz on around 1.44v

I didn't get the black stuff off yet or do anything. No time. I'll give another go at it. I just really wish someone could advise me where to get the exact resistor I need to replace the one I killed. I'd also like to know what it does.

Lastly, what's the best method to spread the liquid metal? Phobya LM has a needle to dispense with, but it balls up and is really hard to spread. How much should I be using? If it looks smooth like a thin puddle on some parts is that too much? What is the best thing to protect the transistors with? I know liquid electrical tape was mentioned, but there was something else too. I'm wondering if I have it here rather than needing to buy something else.

Eventually I will be getting a 4790k because all the DICE games like the battlefield franchise and Star Wars Battlefront benefit greatly from hyperthreading and faster clocks. I'm up at 85-90% CPU usage in Battlefront and I'm just shy of my minimum FPS target dipping into the low 70's. Ideally I'd get between 110-130fps on all maps 98-99% of the time so I could run my monitor at 108Hz and stay above the refreshrate and not get too much tearing. I find it tears if I cap fps at 100 with 96Hz. Still experimenting with triple buffering and whatnot.

EDIT: Hey thanks for that guide. I was already using this one: rog.asus.com/242142013/labels/rog-exclusive/maximus-vi-series-uefi-guide-for-overclocking/ But that has some critical info that I didn't know about and might actually help!

+Rep (Don't worry I'll go back and rep everyone that helped!)

white owl · Dec 17, 2015

I bet you won't hit it with the p95 small...

j/p
I think it's doing fine.
What do you mean about your board's phases?

The best I could spread CLU was how you described. I brushed the puddles on and removed some. I tried using more and less and it made no difference. None.

seanzylol · Dec 18, 2015

Quote:

Originally Posted by white owl

I bet you won't hit it with the p95 small...

j/p
I think it's doing fine.
What do you mean about your board's phases?

The best I could spread CLU was how you described. I brushed the puddles on and removed some. I tried using more and less and it made no difference. None.

voltage regulator phases.. The more the better

I got 3 920s laying around waiting on this one to die

Bench and push them till they die. Rinse and repeat

white owl · Dec 18, 2015

I know what the phases on a board do. More points to fire from. I just don't know what is different on his board as opposed to mine.
At least it's not a krait.

cookieboyeli · Dec 19, 2015

Quote:

Originally Posted by white owl

I bet you won't hit it with the p95 small...

j/p
I think it's doing fine.
What do you mean about your board's phases?

The best I could spread CLU was how you described. I brushed the puddles on and removed some. I tried using more and less and it made no difference. None.

I really like how you can suck a lot back in with the needle. Makes it much easier since getting the amount right the first time is nearly impossible. I'll use as little as possible that coats the die. I'm going to attempt to remove that black stuff tomorrow as well as use some copper shims for all heatsinks on my MB. I still haven't contacted that guy who repaired resistors yet, I'm far too busy ATM to to that.

My boards VRMs are good, but they could be better, hence my desire for the Gene!

By the way, 1.445v is required for this CPU to be stable at 4.6GHz. I've been messing with it and testing it probably 20/24 hours for the past 2 days and that's what it takes to keep it stable. Any less and there's errors in Prime95 27.9. I am switching back to that for most CPU testing since it is a lot more sensitive than Realbench. If I can't pass 2.5 hours P95 it I'll get BSODs in games and possibly even web browsing.

I also found out that I can increase CPU stability by keeping Analog IO BELOW +.220v but increase DRAM stability by keeping it ABOVE +.200 without affecting CPU stability in any measurable way. Also, using +.265v System agent voltage (around 1.225v total) gives DRAM stability gains VS only +.255v. Digital IO needs to be at least +.290v from what I can tell for DRAM to be stable at my extreme clocks. No testing on upper limits and effects on stability above +.310 yet.

Memtest86 6.2.0 by passmark is working properly for me now. I can use the parallel test on this computer without crashing. Apparently it's a known issue for many AMD motherboards and even some intel ones! Something to do with how the program interacts with the board's UEFI and calls multi threaded processing?? IDK go read up on it on their forums. I only skimmed a few posts.

More importantly, Memtest86 6.2.0 is detected 2 errors after 4 passes (2 hours 50 minutes) whereas Memtest 4.2.0 by HCI did not after 1000% (7+ hours)! That's a MUCH better turnaround! ALthough I can't use my computer during it AT ALL! (Not that I really could before, but at least I could open text documents and a single tab if need be). Reducing the DRAM Refresh by 128 in combination with increasing system agent voltage got rid of the errors and allowed me to tighten other timings significantly!

A big reason for doing all this testing was to be absolutely 100% positive if Battlefront was crashing because of my computer, or just because it's a newly released Dice game.

Well, I just passed the 2.5 hour mark on prime. Time to find out.

Anyway, this delid is a dream come true. Before I couldn't really get it stable at 4.6GHz, so I guess you could say that I DID gain MHz! Only 100, but hey it's something!

Quantum Reality · Dec 19, 2015

Quote:

Originally Posted by cookieboyeli

Memtest86 6.2.0 by passmark is working properly for me now. I can use the parallel test on this computer without crashing. Apparently it's a known issue for many AMD motherboards and even some intel ones! Something to do with how the program interacts with the board's UEFI and calls multi threaded processing?? IDK go read up on it on their forums. I only skimmed a few posts.

I had this exact issue on a stock G3258. SMP would lock up or throw spurious errors in memtext86+ but single core mode worked fine.

benjamen50 · Dec 19, 2015

So is the 4670K working right now or RIP?

cookieboyeli · Dec 20, 2015

Quote:

Originally Posted by Quantum Reality

Quote:

Originally Posted by cookieboyeli

Memtest86 6.2.0 by passmark is working properly for me now. I can use the parallel test on this computer without crashing. Apparently it's a known issue for many AMD motherboards and even some intel ones! Something to do with how the program interacts with the board's UEFI and calls multi threaded processing?? IDK go read up on it on their forums. I only skimmed a few posts.

Click to expand...

I had this exact issue on a stock G3258. SMP would lock up or throw spurious errors in memtext86+ but single core mode worked fine.

I've had the issue on an AMD Asrock, Gigabyte and Asus, but not Intel ASUS, though I only tested one intel board with it so far... It's such a common issue that I didn't even bother trying memtest after like the 3rd time. If it can't be resolved then the program is pretty much useless. It may be the MB's fault, but they need to work around that as no amount of bios updates is going to get the latest version working on all boards (since people just don't update).

The 4670k is somehow working. At first I thought it was busted because battlefront crashes every 2 or 3 rounds, but I think that's just Battlefront. New dice game and all. (Trust me I tried EVERYTHING, no seriously, EVERYTHING except trying 356.04 and 347.88. After that I'm contacting support and telling them they've got a hardware incompatibility issue).

cookieboyeli · Mar 20, 2016

UPDATE

So I think I should update this thread. My 4670k has actually died for REAL. It happened about 3 weeks ago. I turned off my PC for the night, and in the morning it wouldn't turn back on again. CPU not detected, motherboard said code 00. I tried remounting it 3 times and tested the G3258, the G3258 worked first try, the 4670k won't work anymore at all though. I guess it's to be expected.

I had a 4790k about 2 weeks out of the factory to replace it with. I was able to go From needing 1.46v for 4.6GHz to 1.34v for 4.7GHz, and I was able to drop almost every other voltage significantly. Cache voltage was able to be dropped from 1.2v something to 1.060v!

That's at 3900 for both. It feels like this chip is way above average. It responds to much smaller jumps in voltage. I haven't delidded it. I'm not going to use the razor method for 100% of the delid as long as they put SMDs on the topside of the board under the cap where there's danger of hitting them.

Besides this, I do have a question that I can't seem to find a definate answer to.

How do you reseal a CPU? I don't want to use just anything either, I'd prefer it loos as close as possible to how Intel does it. I want to seal up my CPUs after getting the liquid metal perfectly applied and the resistors protected, so I don';t have to waste liquid metal any time I change anything. (Plus I want to prevent it from evaporating, as there was almost none left when I took the cap off the 4670k to inspect, though load or idle temperatures didn't indicate thermal contact failure at all!

I have liquid electrical tape (black) for sealing the CPU back up under the IHS, would that work? I'm also planning on getting decent quality clear nail polish (I'll google to find what's good) for the SMDs. I'm not a fan of the liquid electrical tape aesthetics or coverage. It seems to be really hard to apply, especially if your goal is coating an object on all sides. It dries out and can easily come off if disturbed too...

seanzylol · Mar 20, 2016

Quote:

Originally Posted by cookieboyeli

UPDATE

So I think I should update this thread. My 4670k has actually died for REAL. It happened about 3 weeks ago. I turned off my PC for the night, and in the morning it wouldn't turn back on again. CPU not detected, motherboard said code 00. I tried remounting it 3 times and tested the G3258, the G3258 worked first try, the 4670k won't work anymore at all though. I guess it's to be expected.

I had a 4790k about 2 weeks out of the factory to replace it with. I was able to go From needing 1.46v for 4.6GHz to 1.34v for 4.7GHz, and I was able to drop almost every other voltage significantly. Cache voltage was able to be dropped from 1.2v something to 1.060v!

That's at 3900 for both. It feels like this chip is way above average. It responds to much smaller jumps in voltage. I haven't delidded it. I'm not going to use the razor method for 100% of the delid as long as they put SMDs on the topside of the board under the cap where there's danger of hitting them.

Besides this, I do have a question that I can't seem to find a definate answer to.

How do you reseal a CPU? I don't want to use just anything either, I'd prefer it loos as close as possible to how Intel does it. I want to seal up my CPUs after getting the liquid metal perfectly applied and the resistors protected, so I don';t have to waste liquid metal any time I change anything. (Plus I want to prevent it from evaporating, as there was almost none left when I took the cap off the 4670k to inspect, though load or idle temperatures didn't indicate thermal contact failure at all!

I have liquid electrical tape (black) for sealing the CPU back up under the IHS, would that work? I'm also planning on getting decent quality clear nail polish (I'll google to find what's good) for the SMDs. I'm not a fan of the liquid electrical tape aesthetics or coverage. It seems to be really hard to apply, especially if your goal is coating an object on all sides. It dries out and can easily come off if disturbed too...

epoxy lol

cookieboyeli · Mar 21, 2016

Quote:

Originally Posted by seanzylol

Quote:

Originally Posted by cookieboyeli

UPDATE

So I think I should update this thread. My 4670k has actually died for REAL. It happened about 3 weeks ago. I turned off my PC for the night, and in the morning it wouldn't turn back on again. CPU not detected, motherboard said code 00. I tried remounting it 3 times and tested the G3258, the G3258 worked first try, the 4670k won't work anymore at all though. I guess it's to be expected.

I had a 4790k about 2 weeks out of the factory to replace it with. I was able to go From needing 1.46v for 4.6GHz to 1.34v for 4.7GHz, and I was able to drop almost every other voltage significantly. Cache voltage was able to be dropped from 1.2v something to 1.060v!

That's at 3900 for both. It feels like this chip is way above average. It responds to much smaller jumps in voltage. I haven't delidded it. I'm not going to use the razor method for 100% of the delid as long as they put SMDs on the topside of the board under the cap where there's danger of hitting them.

Besides this, I do have a question that I can't seem to find a definate answer to.

How do you reseal a CPU? I don't want to use just anything either, I'd prefer it loos as close as possible to how Intel does it. I want to seal up my CPUs after getting the liquid metal perfectly applied and the resistors protected, so I don';t have to waste liquid metal any time I change anything. (Plus I want to prevent it from evaporating, as there was almost none left when I took the cap off the 4670k to inspect, though load or idle temperatures didn't indicate thermal contact failure at all!

I have liquid electrical tape (black) for sealing the CPU back up under the IHS, would that work? I'm also planning on getting decent quality clear nail polish (I'll google to find what's good) for the SMDs. I'm not a fan of the liquid electrical tape aesthetics or coverage. It seems to be really hard to apply, especially if your goal is coating an object on all sides. It dries out and can easily come off if disturbed too...

Click to expand...

epoxy lol

Haha very funny mate... I don't even think I should be allowed to use epoxy after that last mishap...

...Does it come in black? (mine thermal epoxy is getting old and hard to use...)

Did I kill my G3258 delidding? (No, but I JUST KILLED MY 4670k!! - Need replacement resistor!)

seanzylol

Just a nickname

seanzylol

cookieboyeli

white owl

cookieboyeli

white owl

seanzylol

white owl

cookieboyeli

Quantum Reality

benjamen50

cookieboyeli

cookieboyeli

seanzylol

cookieboyeli

Top Contributors this Month

Recommended Communities