Oh god what the hell is this?

Unstable video card, underpowered video card, too low of voltage, too high of temps.

What are your exact system specs, including the make and model of your power supply?

thideras said:

Unstable video card, underpowered video card, too low of voltage, too high of temps.

What are your exact system specs, including the make and model of your power supply?

Click to expand...

pic of the inside of my computer, the specs of the PSU are written on it

Motherboard: Gigabyte GA-P67A-D3-B3
Processor: Intel i5 2400 3.6ghz
Ram: Kingston CL8 8-9-9-18 2x2gb 1600mhz CL9 KHX1600C9D3K2/4GX
Graphics Card: MSI R7850 Twin Frozr III OC 1.1ghz core 1.375ghz memory
HDD: 1tb 7200rpm 32MB WD10EALX
PSU: Superb E620 620w
Monitor: Samsung 23" S23B300B
Case: Antec One

Unstable OC.

thideras said:

Unstable video card, underpowered video card, too low of voltage, too high of temps.

What are your exact system specs, including the make and model of your power supply?

Click to expand...

manu2b said:

Unstable OC.

Click to expand...

Yeah but it seems weird... I had 5-8 more fps in bf3, from 55 in one scene to 60+. If It was unstable it would've had extremely low fps from the beginning right? Even did a benchmark a while ago on 3dmark (http://3dmark.com/3dm11/3984803 and 3d mark takes like 10 mins so it was stable then, why crash in desktop where the gpu's downclocked to like 300mhz? makes no sense...) but of course I stopped doing it since I had very high temps and gpu shut down because of this (bad case airflow). Should I increase voltage or lower core a bit?

an00bis said:

Yeah but it seems weird... I had 5-8 more fps in bf3, from 55 in one scene to 60+. If It was unstable it would've had extremely low fps from the beginning right? Even did a benchmark a while ago on 3dmark but of course I stopped doing it since I had very high temps and gpu shut down because of this (bad case airflow). Should I increase voltage or lower core a bit?

Click to expand...

No, it's not weird: with an unstable OC, you get higher FPS until your system crashes.

You can either increase vCore or lower frequency.
Gray lines usually mean too high vram clock.

manu2b said:

No, it's not weird: with an unstable OC, you get higher FPS until your system crashes.

You can either increase vCore or lower frequency.
Gray lines usually mean too high vram clock.

Click to expand...

Hmm I think It maybe was that, I was tweaking the gpu to see if there were differences between 1375 and 1450mhz. You sure this was it? This thing isn't damaging my gpu right? just a more severe crash sort of like what happens in nvidia cards sometimes when the driver stops responding except here it's worse right? today was the only day I tried the memory overclock to 1450mhz maybe it was too much... I'll see if it happens again, then I'll downclock a bit more.

Gpu downclocks because it's getting too much voltage or it's overheating. This is why the OVP/OHP kicks in and downclocks the card to prevent damage to it.

Not all video cards are created equal. Just because Joe's card clocks x amount, does not mean yours will clock the same even tho both cards are identical.

A component on your card is overheating or may already been damaged because of the overclock, overvolt or overheat.

Test the card in another machine.

Nebulous said:

Gpu downclocks because it's getting too much voltage or it's overheating. This is why the OVP/OHP kicks in and downclocks the card to prevent damage to it.

Not all video cards are created equal. Just because Joe's card clocks x amount, does not mean yours will clock the same even tho both cards are identical.

A component on your card is overheating or may already been damaged because of the overclock, overvolt or overheat.

Test the card in another machine.

Click to expand...

Well the card works fine I guess no damage yet (not that I know of, do you seriously think something got damaged?), and yeah I know that's a bad example but c'mon, I've seen on the internet some 7850 who come with higher stock voltage like almost 1.2v and they're fine, I'm not going to fry anything at 1.17 and at 60 celsius. And I disabled OVP, well sort of since I increased the power limit thing to +20 so there should be no throttling but the problem I explained in the OP appeared just suddenly out of nowhere, wasn't even playing games. Seems like my card does all kinds of funky stuff outside games but in games it's fine.

Maybe the vram thing is plausible since then I overclocked the ram a bit more than usual to see the difference it brought (it didn't do anything) and my gpu has no thermal pads on the memory chips so maybe they overheated too much.

If it does that out of the blue, yet in games it works fine, then I suggest returning the card and get a replacement. Don't hold on to it if it acts out of whack while you're on desktop.

Sometimes overclocking the card to get 2-4fps more isn't worth it imho. I used to clock my cards to absolute max and all it got me was maybe 5-7fps more, but I didn't notice the difference visually. I only knew because of fraps. Only difference I know was the heat coming off my cards. Now I run my cards basically at stock speeds and run my games.

Nebulous said:

If it does that out of the blue, yet in games it works fine, then I suggest returning the card and get a replacement. Don't hold on to it if it acts out of whack while you're on desktop.

Click to expand...

You sure? Maybe it was only a driver problem or something. I had a bad experience with the guys that I got my gpu from (1 month to repair a keyboard then they just replaced it... with another model) and I don't want to do anything like this if my GPU works fine except for a minor crash.

Well see if it continues to act that way. Up to you, but if it was me I'd sure get it replaced.

Nebulous said:

Well see if it continues to act that way. Up to you, but if it was me I'd sure get it replaced.

Click to expand...

So this is a serious problem or just a minor crash from a bad OC? I just don't want any physical damage on the card wanna know if it was just a normal driver crash or something. I mean the card's fine I overreacted there, it's normal in windows I have a problem that I had with my older card gts 450, sometime the desktop refreshes on itself and stuff but I think it's windows-related not from the hardware. The only problem was this thing I never saw this... crash screen with thin black/gray lines moving like static everywhere.

I would run the card at much lower clocks for a while on stock volts (1050 / 1300) and see if the problem happens again.

These cards have a power saving feature, they switch between low and height clocks and volts, if the card is highly over clocked it could upset that balance when your on the Desktop, while in gaming the cards continuously runs at its highest level, while on the Desktop is when it switches up and down.

Frakk said:

I would run the card at much lower clocks for a while on stock volts (1050 / 1300) and see if the problem happens again.

These cards have a power saving feature, they switch between low and height clocks and volts, if the card is highly over clocked it could upset that balance when your on the Desktop, while in gaming the cards continuously runs at its highest level, while on the Desktop is when it switches up and down.

Click to expand...

Well looks like I have another problem now that you mention it.
The only program that lets me overclock past 1050mhz is TRIXX and when I do it there the volts are constant, so If I set it at 1.170 it's at 1.170 non stop, instead of lowering to about 0.9 it stays at a constant voltage. Maybe they'll fix it or someone will find another program that doesn't do this **** and lets us overclock normally but until then I have to use this crap. (I tried asus, msi, trixx, amd's overclocking in CCC and only this worked)
Now I've been running the card at 1200mhz but I raised the voltage to 1.180 from 1.170 and so far (about 1 hour or more) there have been no problems, runs very good, maybe all it needed was an extra 0.01v.

1050? Thats not low, thats the max CCC does allow.

1200? Thats kinda like a truck on a 7850. Its initial clock is 870...

Nebulous said:

. I used to clock my cards to absolute max and all it got me was maybe 5-7fps more, but I didn't notice the difference visually. I only knew because of fraps. Only difference I know was the heat coming off my cards. Now I run my cards basically at stock speeds and run my games.

Click to expand...

5-7 FPS at 60+ is no value, but its of value at 30+ FPS. 15 vs. 20 FPS is feelable difference, although that is already a OC+ of 25%, while most GPUs may only be capable of 15%. Of course thats just a raw estimation of its gain, actual gain will vary very hard dependable on GPU type (usualy a core with high amount of ALUs will have higher scaling) and game itself.

Another point which may speak for OC is the fact that people are willing to spend a massive bonus for a GPU only 15-20% stronger than another (and they even dare to call the GPU which is only 15-20% weaker "inferior performance, AMD did lose to Nvidia so hard that AMD have to vanish from the planet and things like that". But with OCing its still possible to reach the performance of a expensive GPU at a way lower cost.

So whats the point of paying so much if a OC could do the same? Well... we have granted OC, granted performance and non granted OC, non granted performance. But finally, the difference is, either you pay a lot, or you pay lower and will have to bear the risk in order to compensate for the lower cost through OC. The expensive GPU is a "pay to win" and the cheaper GPU is a "dice to win"... guess thats the nature of OCing and its gain.

1200Mhz on the GPU is not to be sniffed at, that's what i'm running and its as high as i can get it because of my own problems with the GPU, Mine is a 7870 and came out of the box running 1100Mhz.

My problem is i can't get higher volts to stick, but you know what, 1300Mhz and 60+ FPS on BF3 Ultra preset 1920 x 1080 is exactly the same experience as -60 FPS on the same settings.

So why push it that hard in the first place?

for me its just about knowing it can do it, and posting a nice high bench. but that's it, there is nothing more to it than that.

I have not played with TRIKK on this card yet, not really, so i might

My own desire why i may OC and run a benchmark is simply in order to figure out the quality of a certain processor. When it goes high i have a proof of its "inner value", but usualy when im done with, i go either back to stock clock (usualy the case with CPU) or i leave it at a gentle OC of only 10-15% (dependable on the achieved results). When a processor can go high, i give it a stamp of quality and will have a proof that the "daily clock" is a walk in the park with close to no possibility of failure. When it will crash at only 5-10% OC, then it will not look very confident at its initial stability (if possible i trade it in for another card). Usualy when it goes 15%+ (current series) i say its a quality chip.

Usualy i only may OC GPU because my CPU is not limiting, definitely not. Usualy i just want to know what works and what not. But i wont increase volt, so i should drive safe testing out some capability. Nothing is foolproof but even a stock clock is not fool proof, thats a fact. Hardware can die on stock too (ask my boards how fast they can die for nothing but its own pathetic work), but the chance is decreased. When a GPU will even bust up at 15% to the point where it is getting damage (and that without any volt increase) then it was never truly safe to use in the first place.. I want my hardware to be overperforming and with a lot of hidden power left in order to tackle "aging" (electro migration) and all the other bad factors which is appearing over time.

Besides, OCing of RAM is fun but kinda outdated for real world performance. Guess almost any enthusiast is using the sweet spot of 1600 Mhz with some good latency values (at least CL9) thats all whats needed. May even be fine running 1333 (stock) but 1T instead of 2T. More than that is kinda a waste of energy... and stability.


Besides, who said that a OC is barely worth it? or "only" a bit of OC FPS? Well... a maxed out 7870 Hawk (still on air, could be used for daily) 3DMARK11 score = equal to a 7970. At Alien vs. Predator it got near equal FPS such as a stock GTX680, 20% above 580 GTX. Even the 580 GTX Lightning Extreme... outpaced by that mid-range card with incredible hidden power.

Oh lol It seems I forgot to di the most basic thing everyone should do while overclocking... running a test in furmark.
It seems that when I overclocked to 1200mhz the card was throttling itself at 63% constantly, must've been why it was heating so much and why I had that crash. Now I set it to 1175 at 1.125v and It never goes past 60 degrees except in furmark where it goes to 62-64 and gets 99% usage in it. I still get performance = or better than a gtx580 so It's perfect. Now If only the stupid TRIXX could let me have a lower voltage in idle... it always stays at 1.125 no matter what I do, but the clocks do get reduced.

So the crash I got was nothing serious? Wouldn't like to find out that I damaged my card a little during that time. Just the card shutting itself down because of the unstable OC right?

If TRIXX is keeping the volts to what you set them at under load and idle its probably not a bad thing, trust me.

I'm installing MSI Afterburner and installing TRIXX right now to see if it does the same for me, because i want it to

PS: don't use Furmark for any length of time, like some people. its ok as a short stability test but any more than that and there is a chance it could damage your card.