Friendly reminder overclocking can be bad

M

Monkey_feces

Gawd
Joined
Jun 27, 2006
Messages
702
From reading these forums, I felt I was retarded for not overclocking my video card. Since I knew my 7900gt came with refreshed memory, I decided to overclock the shit out of it (without an overvolt). The GPU had a meagre overclock of 520 from 475 on the core and 1560 from 1320 on the ram. I especially felt safe since I adhered to overclocking guides that recommended I backed off 5-10 mhz from the test fail cap. I backed off 30mhz from the ram's cap and 20 on the core. 3 days of flawless functionality followed with an increasing number of color dot artifacts until they covered the screen.

I'm not saying I wouldn't recommend overclocking, I'm just saying that overclocking fail safes aren't as fool proof as they are propped up to be (at least in my case). From reading basic overclocking newbie guides, I got the impression that you have to be retarded to kill a card with vigorous clocks because the thousands of retard net fail safes would just kick in. They didn't and now I feel like a retard for shitting away $250(at the time).
 
Well, any overclocking has its risks. I think everyone should realize the possible consequences of what they are doing when OCing.

However, it kinda sounds like you had a faulty card. You did what I did, backed off 30 from cap and didn't push anything to hard.

Sorry to hear that man.
 
You went too high on the ram. From my experience ram is much more vulnerable to damage when OC'ing than the core. My cards would probably be fried right now as well if I oc'd the ram 250mhz. I backed down from 1870 to 1760 (only a 60mhz oc) when I noticed small white dots in one single game.
 
You probably did go a bit on the ram... but... I have to agree that it sounds like you had a "faulty" card, or one that was on it's last legs before you started your overclock...

And yeah.. anytime you o/c anything, you better understand that there is a chance your going to fry something, so if your not comfortable paying the price, don't do it:)
 
The way I see it, I'd rather pay $250 for something that will work, last a long time and be stable than something that costs $150 and can overclock like hell but last only 18 months or less. Thing is, I read you could overclock two similar CPUs for example and one will work a long time and the other won't.
 
sounds like a faulty card to boot. That said, ram is indeed more vulnerable since there's no thermal throttle.
 
yup as everyone says ram is not the best option to overclock unless you will keep it lower temps than it was at stock.. so what cooler did you use? ramsinks? what temps did you have? seems like lots of 78xx-7900 had defects with ram..

i havnt oc'd my gfx cards ram yet but i plan on testing if i put a 120mm side case fan and ramsinks...
 
I've had my last video card overclocked for years straight no problems. My new one isnt old enough to say anything, but I got it two days ago and it has been overclocked for one day now. ...had to make sure it was fine at stock first ;o)
 
Thats true actually. The 7900 cards had some major ram defects. I think thats why my 7800GTX 512 died in less than a year. It had similar problems (screen corruption but only in 3d games) so I think the ram chips died on it. It was one of the last 7800GTX 512s made so it probably used similar ram to the 7900 cards.
 
killster1 said:
yup as everyone says ram is not the best option to overclock unless you will keep it lower temps than it was at stock.. so what cooler did you use? ramsinks? what temps did you have? seems like lots of 78xx-7900 had defects with ram..

i havnt oc'd my gfx cards ram yet but i plan on testing if i put a 120mm side case fan and ramsinks...
Click to expand...

I had vf-900 ram sinks with the cooler blowing on them. I really don't think heat played a part in it because many overvolted cards on air run much hotter. I didn't catch the artifacts fast because I thought they were dead pixels since my monitor is of an unknown brand (tyris). Could I have saved my card if I downclocked soonerer, or would it have just died slower?
 
I've had my 7900 overclocked since day 1 too. hynix chips. I recall the first batch of gt's having issues, but later batches were remedied from what I've read.

My ram's been pushed at 800mhz since day one. Stock cooler and zalman ramsinks. I've even bumped vram to 2.2v and ran 880mhz for a while as well, but went back to stock 2.0v since I'm not ram limited and 2.2v was overloading the vreg, since I didn't cool it
 
My card shares all the characteristics of an OC'd card...overheating, random restarts, blue screens, etc

...without the actual performance gains.

Sweet deal.
 
It's obviously a cooling problem. If your video card's ram isn't attached to any heatsink, chronic heat from overclocking may damage it.
 
Electronically you CANNOT damage a vid card or other computer part if you don't increase the voltage.

If all you did was up the clock speed of the memory with no voltage increase your card just coincidentally went out at about the same time you were overclocking.

I've been around overclocking a long time and speak from both electrical theory and exp --- I have a BS degree in this stuff.

You can cause artifacts and glitches when you overclock to far --- but if you back off again (so long as there is no voltage increase while overclocking) you will not have damaged your chip.

The artifacts aren't because the chip is being damaged - it is because the chip/memory is working outside of capable specs...as soon as those specs are put in place again you'll be fine.

In your case you just got a bad card that was going to fail at this time anyway.

VOLTAGE increase and heat cause failure - not simply jumping the mhz a few notches.

The amount of heat you increase without increasing voltage is negligable at best.
 
No you can kill a video card by going to far.. It generates more heat and can kill it plus the 7x00 random death problem and 7900 memory issues.
That sucks though.. I've done it :p
 
well your pre- over clocking speeds are slower than my 7900GS ?


how is that , could it be your chip damaged ?
 
Monkey_feces said:
From reading these forums, I felt I was retarded for not overclocking my video card. Since I knew my 7900gt came with refreshed memory, I decided to overclock the shit out of it (without an overvolt). The GPU had a meagre overclock of 520 from 475 on the core and 1560 from 1320 on the ram. I especially felt safe since I adhered to overclocking guides that recommended I backed off 5-10 mhz from the test fail cap. I backed off 30mhz from the ram's cap and 20 on the core. 3 days of flawless functionality followed with an increasing number of color dot artifacts until they covered the screen.
Click to expand...

As soon as you see artifacting you STOP what you're doing and downclock the card, if you think you've hit a good overclock and you're not getting any artifacts, but then later go on to see artifacts appearing again then you need to downclock further.
 
Archaea said:
Electronically you CANNOT damage a vid card or other computer part if you don't increase the voltage.

If all you did was up the clock speed of the memory with no voltage increase your card just coincidentally went out at about the same time you were overclocking.

I've been around overclocking a long time and speak from both electrical theory and exp --- I have a BS degree in this stuff.

You can cause artifacts and glitches when you overclock to far --- but if you back off again (so long as there is no voltage increase while overclocking) you will not have damaged your chip.

The artifacts aren't because the chip is being damaged - it is because the chip/memory is working outside of capable specs...as soon as those specs are put in place again you'll be fine.

In your case you just got a bad card that was going to fail at this time anyway.

VOLTAGE increase and heat cause failure - not simply jumping the mhz a few notches.

The amount of heat you increase without increasing voltage is negligable at best.
Click to expand...

I'm inclined to disagree with this post, you cannot fry your chip electronically without volt modding it, but the faster you run the GPU/Memory the more heat this causes and heat can do permenant damage to your hardware. Not to mention that when the heat increases the resistance of the components starts to change which in turn alters the voltage and current through different parts of the card.

There are failsafes in place to stop the GPU from exceeding certain temperatures but I doubt that is there to stop damage to the card, the failsafe temps are very high (125 degrees C IIRC) and I would imagine that they are there to avoid any kind of situation where the card might become a danger to other components in the system (if you've seen the old videos at tomshardware concerning drastic cooling failure on old AMD chips, you will see that a microprocessor which heats uncontrolably will basically explode).

In the case of someone overclocking and getting no artifacts, then later killing their card due to overclocking can be down to one of many factors, ambient temperature changes is one of them, I always have overclocks lower in the summer when my house heats up uncontroably.
 
You can't generate much additional heat without increasing voltage and You can't generate enough heat on stock voltage in order to fry any of the electronics....

The reason hardware fails is time, heat, voltage irregularities and input voltage being higher than spec'd.

If anyone has any electrical engineers ask them for verification.

You can't trust what you THINK, it's what you know...
 
Monkey_feces said:
From reading these forums, I felt I was retarded for not overclocking my video card. Since I knew my 7900gt came with refreshed memory, I decided to overclock the shit out of it (without an overvolt). The GPU had a meagre overclock of 520 from 475 on the core and 1560 from 1320 on the ram. I especially felt safe since I adhered to overclocking guides that recommended I backed off 5-10 mhz from the test fail cap. I backed off 30mhz from the ram's cap and 20 on the core. 3 days of flawless functionality followed with an increasing number of color dot artifacts until they covered the screen.

I'm not saying I wouldn't recommend overclocking, I'm just saying that overclocking fail safes aren't as fool proof as they are propped up to be (at least in my case). From reading basic overclocking newbie guides, I got the impression that you have to be retarded to kill a card with vigorous clocks because the thousands of retard net fail safes would just kick in. They didn't and now I feel like a retard for shitting away $250(at the time).
Click to expand...

Is it me, or is "retard" in there a lot? lol. Don't sell yourself short dude...


Jk.
 
what fail safes? core shut down only applys to core not memory.. there is no fail safe on the memory.. what temps did your case sit at? I also would not consider the ramsinks that come with zalman heatsinks to be very good.. i have a ziplock baggy filled to the top with them all unused.
 
Since I have a budget board, I don't think it can output case temps. I do know my cooling is quite adequate since it's an antec 900 (a big ol fan that you can fit computer parts in).
Even though vga ram doesn't have a shut down feature, I followed the "simple overclocking guides" to the letter. That is, I 1)test whether your overclock works by increments of 5, 2) back off 5-10 mhz (i backed down by 20) when I hit the cap, and 3) I tested for artifacts with 2 straight runs of 3dmark07 and non came up.

Why aren't the zalman sinks any good? Even then, shouldn't the fan's design offset the sinks since it blows directly on the ram?
 
My second 7900gt has been hit with artifacting, although with a lesser degree. Could my artifacting be motherboard or psu related? It's kind of weird that both have begun to show artifacts after several months of flawless performance. My second 7900gt hasn't been pushed as hard as the first one, because I knew it wasn't part of the memory refresh serial.
 
some were probably way worse than others. personally, mine's been at 800+ ram since day one and 700+ core
 
zamardii said:
Is it me, or is "retard" in there a lot? lol. Don't sell yourself short dude...


Jk.
Click to expand...

I agree. Don't beat yourself up with these types of questions: Could I have saved it if...? More than likely, it was going to go bad anyway. There were *many* ram problems with those cards, and as for heat causing it, what would have happened over time? Unless you religiously blow out your case with canned air, the temperatures will go up a bit, then you would still be in the same boat. What if your A/C died while you weren't home - that would have raised temperatures and you'd still be in the same boat.

Overclocking too far can have consequences, certainly, but it sounds as though you weren't being insane about it (particularly with no voltage increase). Chalk it up to a bad card, and rejoice that you now have a good excuse to upgrade :D.
 
Caffeinated said:
I agree. Don't beat yourself up with these types of questions: Could I have saved it if...? More than likely, it was going to go bad anyway. There were *many* ram problems with those cards, and as for heat causing it, what would have happened over time? Unless you religiously blow out your case with canned air, the temperatures will go up a bit, then you would still be in the same boat. What if your A/C died while you weren't home - that would have raised temperatures and you'd still be in the same boat.

Overclocking too far can have consequences, certainly, but it sounds as though you weren't being insane about it (particularly with no voltage increase). Chalk it up to a bad card, and rejoice that you now have a good excuse to upgrade :D.
Click to expand...

Unfortunately having an excuse for an upgrade doesn't give me the resources for one :( .
Is there a chance that my artifacting can be the cause of faulty pci-e slots? It's weird that both of my 7900gts went out on me at about the same time. On both cards, the only artifacts I get are flickering dots. Are the dots a strong indicator of faulty ram, or is there a possibility it can be something else? Another noteworthy problem is that I get a BSOD whenever I open up the nvidia control panel when running SLI. The message states that it was an infinite loop found in my nvidia display driver.
 
I've been at this thing for hours, swapping motherboards, psu's, pci-e slots, cpu's, ram sticks,
mobo batteries, downclocking, pencil volting (someone told me this fixed their artifacts) and tinkering with jumpers, but it seems that swapping my dvi cable for my vga cable fixed the artifacting. What are the chances that my dvi cable is faulty, what about my monitor's dvi port? I have done research on other forums and everyone who replaced their dvi cables didn't get any better results. I don't think there's anything wrong with my input dvi slots since the chances that all 4 are faulty are pretty slim.

PS: I know many of you are probably tired of my self loathing rants, but I feel even more retarded that I went through a 2 page thread of panic and more than 5 hours of trouble shooting only to fix it by swapping a measly cable.
 
Archaea said:
Electronically you CANNOT damage a vid card or other computer part if you don't increase the voltage.
Click to expand...
I too disagree with this statement. Heat is one of the worst enemies here, and certain cards like the X1900XTX - which is hot as F*** to begin with, could easily be pushed towards their death by overclocking them. Chances are, it may have died for another reason, sure. But to say overclocking without extra voltage is absolutely harmless is a bold statement. My X1900xtx lasted about six months, clocked as high as the built in ATI utility would allow me to clock it. Six months of rock solid stability. Best video card ever I thought, except for the fact that it sounded like a hair dryer in my case. It might not fail immediately, but the extra speed and heat could easily assist in killing the hardware.

TO the OP - we feel your pain. Now suck it up and get back to overclocking. Chalk it up as a lesson learned. Overclocking means different things to all of us. Some of us have nothing better to do and will spend hours/days squeezing out 3 extra mhz, some of us just want to push the hardware a bit more. That's one reason I will always buy a card with a lifetime warranty...
 
Archaea said:
You can't generate much additional heat without increasing voltage and You can't generate enough heat on stock voltage in order to fry any of the electronics....

The reason hardware fails is time, heat, voltage irregularities and input voltage being higher than spec'd.

If anyone has any electrical engineers ask them for verification.

You can't trust what you THINK, it's what you know...
Click to expand...

You haven't been in overclocking long enough to figure it out. At first, I thought the same thing, but if you want to see it in real life, go get a Pentium D 950 and try it yourself. My Pentium D 950 @ 1.38V does around 4.5Ghz. At stock speeds at that voltage, I idle around 45C; at 4.5Ghz, I idle at 50-55C. This is with the same cooler, same setup, no change in thermal paste or anything. Didn't learn much in college?

Also, your temperature CAN change with-out voltage changes OR clock changes. There are a lot of variables out there which can cause temperatures increases, one of which being human error. A lot of people poorly apply thermal compounds to their heatsink and CPU//GPU, which can cause a massive flux in temperatures when comparing to many different users. Heatsink designs can also alter temperatures....

As for killing parts with heat, it's pretty hard to do that. It would require prolonged exposure to heat for the chip to deteriorate, and usually they slowly deteriorate, and you can see this via lowering clock speeds or increasing voltage to hold stability. If it was a heat problem, I think your 7900 was already going down. If not, my guess was that it wasn't stable and you should drop it back down more mhz. Btw, don't use the nvidia detect settings crap, it's told me that I can run my 7600GT at 780//1620.... I can only do 736ghz stable on core with that card minus a volt mod, lol.
 
theres a lot of noob speculation here.

might want to read up on the melting and boiling points of copper, aluminum, and silicone.

i don't think anyone is hitting temps those high.
 
All it has to do is destabilize and provide unequal power to something and its dead. Electronics are very very weak to heat. Burning out transistors is the main problem.Bulging capacitors, faulty power regulators. etc etc.
 
Any EE (including yours truely) who has completed his/her solid state physics (or equivalent) course will tell you that there's no way overclocking without increasing voltage will actually harm the physical silicon. Chips that die due to old age usually don't die because of prolonged heat damage. More likely an unstable power supply killed some component on the board.

That being said, I do believe that a lot of the 7900 GTs were designed out of spec to begin with (crappy power regulators and/or memory chips). I know because eVGA replaced mine for free.
 
AzN1337c0d3r said:
Any EE (including yours truly) who has completed his/her solid state physics (or equivalent) course will tell you that there's no way over clocking without increasing voltage will actually harm the physical silicon. Chips that die due to old age usually don't die because of prolonged heat damage. More likely an unstable power supply killed some component on the board.

That being said, I do believe that a lot of the 7900 GTs were designed out of spec to begin with (crappy power regulators and/or memory chips). I know because eVGA replaced mine for free.
Click to expand...

Take the heat sink off of your CPU and get back to me.

Heat death can and will occur in solid state electronics. Over clocking does shorten the lifespan of a processor down from the ~10 years they will receive. Anyone thats taken basic electronics understand the fundamental issues with temperatures relation to the circuitry such as wires resistance varying at multiple temperatures.
 
Vashypooh said:
Take the heat sink off of your CPU and get back to me.

Heat death can and will occur in solid state electronics. Over clocking does shorten the lifespan of a processor down from the ~10 years they will receive. Anyone thats taken basic electronics understand the fundamental issues with temperatures relation to the circuitry such as wires resistance varying at multiple temperatures.
Click to expand...

fortunately, it's not yesterday anymore. thermal throttling nowadays make it so it won't even really lock up without a heatsink on the cpu anymore :)
 
Vashypooh said:
All it has to do is destabilize and provide unequal power to something and its dead. Electronics are very very weak to heat. Burning out transistors is the main problem.Bulging capacitors, faulty power regulators. etc etc.
Click to expand...
correct, BUT

THIS IS OBVIOUSLY LIMITED BY THE MELTING AND BOILING POINTS OF

COPPER
ALUMINUM
SILICONE


you people blind?
 
Uberhitzung said:
theres a lot of noob speculation here.

might want to read up on the melting and boiling points of copper, aluminum, and silicone.

i don't think anyone is hitting temps those high.
Click to expand...

What does the melting and boiling points of breast implants have to do with this?

(hint: The word you're looking for is silicon, not silicone. ;))

Archaea said:
You can't generate much additional heat without increasing voltage and You can't generate enough heat on stock voltage in order to fry any of the electronics....

The reason hardware fails is time, heat, voltage irregularities and input voltage being higher than spec'd.

If anyone has any electrical engineers ask them for verification.

You can't trust what you THINK, it's what you know...
Click to expand...

I trust what I know. And I know that you're wrong. Too many hardware sites with far more experience than you or I have had hardware fail on them during overclocking tests, and too many users here have had the same happen, without increasing voltage.

As others have said, when you increase heat, you can increase electrical resistance. You also increase the rate of electromigration, which normally takes a long time to cause circuit failure, but can be drastically accelerated when overclocking a piece of computer hardware. Some hardware is more robust, so it may take a long time for the effects to show. Other IC's when overclocked, may fail far more rapidly.

You can overclock an IC to failure, just as surely as you can ruin a Socket A Athlon XP processor in 10 seconds by powering it up without a heatsink/fan unit, with no voltage changes at all.
 
LoneWolf said:
I trust what I know. And I know that you're wrong. Too many hardware sites with far more experience than you or I have had hardware fail on them during overclocking tests, and too many users here have had the same happen, without increasing voltage.
Click to expand...
I can't be wrong because I am simply making a statement of pure physics.

You, on the otherhand, are making a PURELY SPECULATIVE LEAP into the obviously UNKNOWN.

Plain and simple. You state.

overclocking, that is, increasing the speed at which a cricuit operate can potentially damage the circuit strictly based on the "it wasn't meant to go that fast" theory.

Too many holes unfortunately. I will stick with VERIFIED physics. If it was "clocked too fast" then it "going too fast" would have to cause excess heat internally.

On Planet earth, "fastness" does not alter the physical properties of metals, HEAT DOES.

Elements have melting/boiling points. Physically, they should remain unaltered unless those limits are reached. Over time prolonged heat will slowly alter a chip, however, "going fast to the extreme" will not have any affect outside of Earth Physics.

we aren't talking about the mars rovers are we?
 
and that article seems to prove you wrong as well.

A reduction of the structure (scaling) by a factor k increases the power density proportional to k and the current density increases by k2 whereby EM is clearly strengthened.

In modern consumer electronic devices, ICs rarely fail due to electromigration effects. This is because proper semiconductor design practices incorporate the effects of electromigration into the IC's layout. Nearly all IC design houses use automated EDA tools to check and correct electromigration problems at the transistor layout-level. When operated within the manufacturer's specified temperature and voltage range, a properly designed IC-device is more likely to fail from other (environmental) causes, such as cumulative damage from gamma-ray bombardment.



So apparently your processor is more likely to die of "cumulative damage from gamma-ray bombardment" than overclocking.

WAY TO GO:rolleyes:
 
You must log in or register to reply here.
Back
Top