Weird issue overclocking my GTX570.

[Waco]

I'm betting it's the card itself being unstable with the high voltage (the VRMs may not be able to pump out that kind of current). Extreme GPU failures can often hard-lock or reboot the machine (especially if it's a power error). My 4870X2 does the same thing if I push the voltage/clocks too high. It runs "stable" but will crash without warning or display corruption.

[Nyt]

Remember Merc that GTX570's kept dieing from VRM's blowing up , see here - http://www.overclock.net/nvidia/929152-have-you-killed-570-a.html

 

What I found :

- Gigabyte 570 OC Windforce 3x and MSI 570 TwinFrozr II both use the reference PCB design with 6 phase VRM (4 for GPU and 2 for memory). They simply put a custom cooler on top. MSI cards have similar incidents reported

 

- Gainward 570 Phantom, Gainward 570 GS Goes Like Hell, Palit 570 Sonic Platinum all have 8 phase VRMs (6 GPU and 2 memory). Same amount with the GTX 580. All these cards have identical PCB since they belong to the same graphics card manufacturer group

 

- Asus GTX570 DirectCU II also has 8 phase VRMs (6 GPU and 2 memory)

 

So it seems you may not be safe Merc , dont push too hard . Even people with a WC'ed GTX570 have blown it up - IT IS NOT DUE TO TEMPS so dont let that fool you

[Nyt]

The MSI GTX570 Twin Frozr III has more VRM's (6+1 : Twin Frozr III) not the TF II

[El_Capitan]

I'm sure Merc knows the risks, but I think it's more about the batches of GTX 570's rather than it being a VRM issue. My first reference MSI GTX 570 was a dud at stock voltages and settings, remember? Even the 1 star reviews at Newegg were numerous. I'm running both my MSI and Palit Sonic Platinum at 1150mV some of the time, but mostly at 1050mV the rest of the time. No issues for me, yet. I just made sure the VRM's are properly cooled.

[MercuryDoun]

Remember Merc that GTX570's kept dieing from VRM's blowing up , see here - http://www.overclock...lled-570-a.html

 

What I found :

- Gigabyte 570 OC Windforce 3x and MSI 570 TwinFrozr II both use the reference PCB design with 6 phase VRM (4 for GPU and 2 for memory). They simply put a custom cooler on top. MSI cards have similar incidents reported

 

- Gainward 570 Phantom, Gainward 570 GS Goes Like Hell, Palit 570 Sonic Platinum all have 8 phase VRMs (6 GPU and 2 memory). Same amount with the GTX 580. All these cards have identical PCB since they belong to the same graphics card manufacturer group

 

- Asus GTX570 DirectCU II also has 8 phase VRMs (6 GPU and 2 memory)

 

So it seems you may not be safe Merc , dont push too hard . Even people with a WC'ed GTX570 have blown it up - IT IS NOT DUE TO TEMPS so dont let that fool you

 

 

The MSI GTX570 Twin Frozr III has more VRM's (6+1 : Twin Frozr III) not the TF II

 

All of the above i know. But thanks for confirming my research as well, My goal is 1Ghz core clock though, and im bound and determined to make that happen.

 

Im very interested in the Twin Frozr III card.... trust me... Want one so bad... lol.

 

 

I'm sure Merc knows the risks, but I think it's more about the batches of GTX 570's rather than it being a VRM issue. My first reference MSI GTX 570 was a dud at stock voltages and settings, remember? Even the 1 star reviews at Newegg were numerous. I'm running both my MSI and Palit Sonic Platinum at 1150mV some of the time, but mostly at 1050mV the rest of the time. No issues for me, yet. I just made sure the VRM's are properly cooled.

 

Ill fully aware of the risks involved, but whats the worse that happens? I blow up the VRM on my 570 and have to RMA the card, and im without a gaming computer for 3-4 weeks. Its probably a good thing as the weather is finally starting to get decent out and i could use a good reason to go outside for awhile lolol.

 

Im getting the feeling that the first round or 2 of 570's just had plain bad VRM's if you notice people arent having anywhere near the amount of issues with the VRM now as they were.

 

I didnt get a chance to do anything overclocking wise. So maybe tonight after i get home from work.

[Nyt]

Yeah the Twin Frozr III is nice ... If it came to South Africa before I wanted my DCII then I would've bought it . The TFIII cooler is pure awesomeness .

If your 570 dies then dont say we didnt warn you

[El_Capitan]

My goal is 1Ghz core clock though, and im bound and determined to make that happen.

It can be done, my reference MSI GTX 570 could do it at 1150mV! You can DO it!

[imm0rt41]

I hope you can sort out this issue! I just reached my first wall on my GPU OC for our duel, but am still at stock voltages. Let's just say a 1GHz core for me will not be an issue

 

Any hotspots on the PCB? Kinda old school/low tech but if the areas near the VRM are getting really hot you might need some heatsinks on it or something. Think ghetto haha

[MercuryDoun]

It can be done, my reference MSI GTX 570 could do it at 1150mV! You can DO it!

 

Yeah, i just cant get my card to run at 1150mv without tanking almost instantly lol.

 

Im pretty sure you nailed it at ICH/QPI/PCIE voltage Capi. I didnt get a chance to try anything last night, but i havent touched those voltages in along time. In fact, if i remember right, i manually set the ICH/PCIE voltages to stock settings so that they didnt get overvolted by the mobo when i was shooting for 4.5Ghz.

 

 

I hope you can sort out this issue! I just reached my first wall on my GPU OC for our duel, but am still at stock voltages. Let's just say a 1GHz core for me will not be an issue

 

Any hotspots on the PCB? Kinda old school/low tech but if the areas near the VRM are getting really hot you might need some heatsinks on it or something. Think ghetto haha

 

Lol, well our duel has been the motivation i needed to finally shoot for 1Ghz on the core, But ive wanted to do it for a long while now. The only part of my video card that is hot is the heatpipes themselves... i touched one on accident... oww! Other than that its all just warm, nothing that seems oddly hot compared from the stock settings to the overclocked settings. So i dont think i need to worry about any additional heatsinks or anything. I just need to find out whats making it unstable, and preferably have time to figure it out too lol.

 

Well i really hope you do get 1Ghz on your card, because your gonna need it to compete with mine whether i get 1ghz or not.

[imm0rt41]

Lol, well our duel has been the motivation i needed to finally shoot for 1Ghz on the core, But ive wanted to do it for a long while now. The only part of my video card that is hot is the heatpipes themselves... i touched one on accident... oww! Other than that its all just warm, nothing that seems oddly hot compared from the stock settings to the overclocked settings. So i dont think i need to worry about any additional heatsinks or anything. I just need to find out whats making it unstable, and preferably have time to figure it out too lol.

 

Well i really hope you do get 1Ghz on your card, because your gonna need it to compete with mine whether i get 1ghz or not.

Ok I'll let the cat out of the bag...core is at 970MHz on the stock voltage and can run Kombustor stress test for a good hour with no issues. Even played a little BC2 with the roommate and ran 3dmark vantage a couple of times lol

 

Just out of curiosity, are any of those heatpipes running near the voltage regulator? I wonder if heat soak could be an issue -- the heatpipes getting too hot and having some of that dissipated heat ending up in the area of the regulator? In my old rig, I had a thermalright heatsink with a tornado on it...it actually caused the capacitors on the mobo near the CPU to heat up because the hot air was getting pushed out of the heatsink but airflow was being blocked from the intakes. I ended up making some cheap ramsinks to control the heat coming off the caps then rearranged fans at the time and it worked out well. Probably not an option due to the restricted size of the GPU though.

[MercuryDoun]

Ok I'll let the cat out of the bag...core is at 970MHz on the stock voltage and can run Kombustor stress test for a good hour with no issues. Even played a little BC2 with the roommate and ran 3dmark vantage a couple of times lol

 

Just out of curiosity, are any of those heatpipes running near the voltage regulator? I wonder if heat soak could be an issue -- the heatpipes getting too hot and having some of that dissipated heat ending up in the area of the regulator? In my old rig, I had a thermalright heatsink with a tornado on it...it actually caused the capacitors on the mobo near the CPU to heat up because the hot air was getting pushed out of the heatsink but airflow was being blocked from the intakes. I ended up making some cheap ramsinks to control the heat coming off the caps then rearranged fans at the time and it worked out well. Probably not an option due to the restricted size of the GPU though.

 

Well the heat pipes actually come out and above the card. So i dont think its coming anywhere near the the VRM, but i could be wrong. The more i think about its behavior the more i think its a BIO's setting. It doesnt run long enough for heat to even pretend to be and issue on any voltage higher than 1125mv. But now im curious enough to check where exactly the heatpipes go... maybe your onto something here??

 

You know both AMD and nVidia installed circuits on their boards that prevent Kombustor and Furmark from fully utilizing a card? Kombuster and Furmark can no longer be used as stability testing tools for the newest generation of video cards. They sorta work... but not really because the card will not allow the program to put full stress on the card.

 

Well... 970mhz on stock voltage is pretty good id say!! How loud is the fan at max speed? Hopefully with stock voltage you could run that speed with auto control still and keep things pretty quiet! That would be awesome for a 24/7 core clock then. You had any luck with overclocking the memory?

[imm0rt41]

You know both AMD and nVidia installed circuits on their boards that prevent Kombustor and Furmark from fully utilizing a card? Kombuster and Furmark can no longer be used as stability testing tools for the newest generation of video cards. They sorta work... but not really because the card will not allow the program to put full stress on the card.

 

Well... 970mhz on stock voltage is pretty good id say!! How loud is the fan at max speed? Hopefully with stock voltage you could run that speed with auto control still and keep things pretty quiet! That would be awesome for a 24/7 core clock then. You had any luck with overclocking the memory?

haha that's actually why I tried a few different tests to load the GPU because I noticed Kombustor was only using 80% of the GPU. I wish it would still load it to max. If I try to go to 990 on the core it does lock up though, so it is fairly accurate in that it let me discover where the limit was on stock voltage.

 

I currently have the fan set to 75% speed and it's tolerable with the other PCs in the room, but any higher and if I lived alone I could see it being annoying. I have had similar luck with the memory. I started by increasing each clock by 10MHz and stressing it for a half hour to an hour. So far, I am at 970/1470, but I think the memory can go higher. I was just increasing them by the same increment because I'm not at the absolute limit and can increase voltages still. I will fine tune the clocks once I am done though lol

 

I'm just hoping I can figure out my CPU OC in time for benching this weekend. I have had a hell of a week this week

 

I saw someone else with an issue with an unstable OC that was previously stable, and I think they had the setting in afterburner to keep the voltage constant unchecked. Forgot what it's called though but might be worth looking at?