Dead 570's (Have you killed yours?)

[bassplayer]

As there seem to have been more than a few dead 570's, I'd like to collect some information.

Core Voltage:
Clocks:
OCP/Power limit disabled: (Y/N)
Load Temperature:
VRM Cooling:
What you were doing: (Gaming, benching, stress testing)

Here's mine:

Core Voltage: 1.212V (1.223 with DMM)
Clocks: 1100/1200
OCP/Power limit disabled: No.
Load Temperature: -40C
VRM Cooling: Stock unisink / cold from phase / fan
What you were doing: Running Vantage Performance





Here was my original post:

Well guys, got a 570 yesterday to play around with. I had one of my teammates do all the modding for me (I'm shameful with a soldering iron).

I was playing around with it yesterday and running some benches on air. I noticed that I was extremely limited in that there was absolutely no scaling above 1.15V. So tonight, I prepped the card for some subzero phase fun. Insulated it very well and booted in, idling at -50C.

So the first thing I did was kick the voltage up to 1.212V (max in software) to see how high I could get without adjusting the resistor for GPU voltage. I ran a quick stability test and made it up to 1100/1200. I decided to do a quick Vantage run and made it through both GPU tests and then crashed in the CPU test (). So I booted back up after adjusting my CPU in the BIOS and reset clocks.

Right after Miss Jane Nash showed her assests and dove onto the boatplanething, there was a bit of a pop, some smoke, and my computer restarted.

I turned it off and quickly dismantled my card. It had only been cold for about 10 minutes so there was absolutely no moisture at all.

Then I found my VRM's





Don't go over ~1.15V unless you add some phases!



Also, this isn't isolated to just me, check these links on OCN:

http://www.overclock.net/nvidia/9282...fried-gtx.html
http://www.overclock.net/nvidia/9141...-570-dead.html

[CHiLLiN1369]

thanks for the post and info Bass, looks like a few issues with 570's.
hope mine survives...

[bassplayer]

Yeah... was really hoping to snag a few golds as there haven't been any frozen 570's on the bot yet.

[sin0822]

I disabled OCP through software from TPU with the command line, i saw a 100 watt increase at 1.1v, totally stock card, no modded BIOS or anything. And i was keeping the whole card cold, just cold air, and it tossed out some ok numbers. i was going to do some modding, but now I am scared. i have the same reference design as you have, and its missing 2 extra phases for the 580GTX, seems like they should've only removed 1 not both from the reference cards

[bassplayer]

I absolutely agree. Pizzaman, the team's card doctor, is going to try to save it via a donor board, I believe. Any suggestions?

[Hondacity]

no heatsink?

[jeremyshaw]

Pizzaman? Of OCN, iirc?

EDIT: oh, right! Fail on my part.

[sin0822]

is there a link to the mods he did? i am interested

[miahallen]

One of my 570s died yesterday with 1.2V on air.....zero mods, cold temps, and very little use. Never even removed the heatsink

[sin0822]

One of my 570s died yesterday with 1.2V on air.....zero mods, cold temps, and very little use. Never even removed the heatsink

Scary, so what is with this NVidia max vcc of 1.3v I heard?

[The Nemesis]

Hmm been testing my cards at 1.213v on air. I think I'll back that off a bit now.

[PatRaceTin]

does Disable with GPUZ Command line help 570/580 overclock more?????

[STEvil]

wow, remind me not to buy a 570.

[bassplayer]

no heatsink?

While it was on phase, I left the unisink/stabilizer on. That got a bit cold from being so close to the phase and it also had a fan on it.

does Disable with GPUZ Command line help 570/580 overclock more?????

If you can make it through the benchmark without your VRM's dying, possibly. However, I wouldn't try it lol.




With regards to the mods done, they were as follows:

vGPU, vMem, vPLL, CB, and lots of caps. There were also a few read points.

Big image:

[mikecdm]

One of my 570s died yesterday with 1.2V on air.....zero mods, cold temps, and very little use. Never even removed the heatsink

This isn't looking good. I've been running mine for over a month now. Only time I use it is to bench it and is still holding up. It has an mcw60 on it, but stock sink for memory and vrms.

Hmm been testing my cards at 1.213v on air. I think I'll back that off a bit now.

I've noticed that temps help more than volts at a certain point. This unpredictable texas weather sure doesn't help. Some days I get better clocks than others just based on what comes in through the window. I haven't gained much by using 1.213 over 1.20v.

We'll just have to see if it survives once I freeze it

[bassplayer]

This isn't looking good. I've been running mine for over a month now. Only time I use it is to bench it and is still holding up. It has an mcw60 on it, but stock sink for memory and vrms.



I've noticed that temps help more than volts at a certain point. This unpredictable texas weather sure doesn't help. Some days I get better clocks than others just based on what comes in through the window. I haven't gained much by using 1.213 over 1.20v.

We'll just have to see if it survives once I freeze it

I agree. On standard air, I saw no gains over 1.15V. Getting it cold, I went from 960 max to 1100 at 1.212V, which was probably not the maximum. Didn't get to find out .

[fl0w3n]

Aww man that's a shame, but good to hear pizzaman can possibly save it.

[NoGuru]

Let's hope these cards get a revision. I think I may run mine a little slower for now.

[W1zzard]

i blew a gtx 570 just by overvolting and stress testing with ovp,ocp,nvidia power limit still in place

[zalbard]

Bad news about 570s...

Let's hope these cards get a revision. I think I may run mine a little slower for now.

Agreed.
At least people should be aware of the fact now, and take it easier or reconsider the purchases.

[nolonger]

Wonder if the 580 has the same potential problem.

[K404]

Wonder if the 580 has the same potential problem.

Biggest problem for GTX 580 reference is increasing the voltage to the GDDR. The PWM isn't really strong enough and is prone to burning out.

Given the reduced power consumption of the 570, I'm surprised the reduced PWM is running on the edge... the PWM on my 580 ran so so cool

[Defoler]

Wonder if the 580 has the same potential problem.

Hehe yeah me too.
I was considering doing a session this weekend on my cards without the power limiter from scratch, starting at 1.1v and see how they handle it.

I think I will skip that now

[bassplayer]

Hey guys, I updated the first post to gather some information. Please fill out the details requested if possible!

[El Gappo]

You really couldn't wait 1 day until the competition starts?

Mine wont even bloody post