How to make a card last..

[roland12321]

Hello. I've two questions.

In my life I've had:
GF 6600 - broke after 6 months
GF 6600 - broke after 2 years
GF GT7200 - broke after 4 years (best I've had)
GF GTX260 - broke after 2 and a half years

The GTX260 died soon after its warranty finished .. I monitored its temperature since day 1, and it never exceeded 85°C when gaming...

I found out its best to keep a GPU below 80°C, but I doubt those 5°C were the real killer.. So...

Question 1: Why do my cards keep dying so quickly?! Is it just bad luck? Or is it that NVIDIA just doesn't like me..?

So I decided I'm going to buy an AMD card next, either the HD6850 or HD7850, depending on the price.

Now I heard somewhere some time ago that if you underclock a card its temperature will drop and it will last you much longer. Is this true? Will it last longer? So...

Question 2: If I buy a factory-overclocked graphics card (so it has nice cooling), and then I underclock it to its original speed or slightly less, do you think it will last much longer than if I buy the standard non-overclocked version?

Any help will be greatly appreciated!!

Roland

 

[TheAdvocate]

I still have:

BFG 6800 GT OC
XFX 7950 GT
XFX 9800 GT

Before I had an MSI GF 4400Ti that I sold and it was used for another 5 years AFAIK.

I have run all of those in crammed Shuttle SFF systems on PSUs that were straining at times to supply enough juice. I have never killed a card. Worst I ever did was wear out the fan on my 9800GT.

I dont think Ive met anyone that killed a GPU unless they OC'd the hell out of it, or did something really stupid.

 

[roland12321]

I never overclocked any of them.. Like all other components, they went in, stayed there, until it just spontaneously broke -.-

 

[lavaheadache]

Hello. I've two questions.

In my life I've had:
GF 6600 - broke after 6 months
GF 6600 - broke after 2 years
GF GT7200 - broke after 4 years (best I've had)
GF GTX260 - broke after 2 and a half years

The GTX260 died soon after its warranty finished .. I monitored its temperature since day 1, and it never exceeded 85°C when gaming...

I found out its best to keep a GPU below 80°C, but I doubt those 5°C were the real killer.. So...

Question 1: Why do my cards keep dying so quickly?! Is it just bad luck? Or is it that NVIDIA just doesn't like me..?

So I decided I'm going to buy an AMD card next, either the HD6850 or HD7850, depending on the price.

Now I heard somewhere some time ago that if you underclock a card its temperature will drop and it will last you much longer. Is this true? Will it last longer? So...

Question 2: If I buy a factory-overclocked graphics card (so it has nice cooling), and then I underclock it to its original speed or slightly less, do you think it will last much longer than if I buy the standard non-overclocked version?

Any help will be greatly appreciated!!

Roland

Who's to say that original clocks are conservative or aggressive? It's not like the silicon tells the manufacturer what it's clocks are "supposed" to be.

I think there are many factors that play a role in a components life.

I think one of the biggest factors can be heat cycles and # of hours.

Somebody who never games vs. a 24/7 WOW guy or somebody who never shuts their computer off.

Even properly cooling your system down does wonders instead of running furmark and prime for 12 hours and then just turning the system off.

I say 7850 would be better since the 6850's performance is going to drop like a rock on newer games. /speculation

 

[tviceman]

Keep your computer systems free of dust (clean regularly), and don't run it 24/7 (turn it off at least at night). Anything will decay over time, but if you always have power running through it, then it will degrade faster.

 

[roland12321]

Keep your computer systems free of dust (clean regularly), and don't run it 24/7 (turn it off at least at night). Anything will decay over time, but if you always have power running through it, then it will degrade faster.

I cleaned it once a year or once every two years.. but still 2 and a half years, with maybe 1 or 2 cleans, could dust really be the problem?

I guess you do have a point as regards power. I often had my PC on overnight (maybe 2 times per week). I'll keep that in mind!

Who's to say that original clocks are conservative or aggressive? It's not like the silicon tells the manufacturer what it's clocks are "supposed" to be.
...
I say 7850 would be better since the 6850's performance is going to drop like a rock on newer games. /speculation

So you think underclocking it won't make a difference?

And yeah, the 7850 is quite a bit better, but I think it's going to be a little pricey.. I think I can get a 6850 for less than half the price of a 7850, so yeah..

 

[SHAQ]

Get a lifetime warranty. It's rare for them to last beyond 4 years if they are used every day. You could try baking them as well.

 

[roland12321]

Get a lifetime warranty. It's rare for them to last beyond 4 years if they are used every day. You could try baking them as well.

I looked for lifetime warranty, but they're hard to come by (here in Europe).

Tried baking, didn't work.. stank the place up though!

What I don't get is I've never had a broken CPU, I had a broken motherboard twice, once on a PC of 8 years old, and another on a PC of 7 years old. I've had a broken RAM chip once.. how come all these components work perfectly fine, but graphic cards cannot?? It just doesn't add up imo...

 

[Fallengod]

Because graphic cards have are composed of more components, use much more power and produce much more heat. I think the technology is just a little more mature for CPU's. Its rare for a CPU to fail, though I have had quite a few video cards fail.

It is also possible that you are helping your cards fail. Low quality or poorly made power supplies can do this. Older process cards that run hotter and use more power probably hurt its longevity. More heat and power always contribute to reducing somethings lifespan.

 

[T_Yamamoto]

What I don't get is I've never had a broken CPU, I had a broken motherboard twice, once on a PC of 8 years old, and another on a PC of 7 years old. I've had a broken RAM chip once.. how come all these components work perfectly fine, but graphic cards cannot?? It just doesn't add up imo...

CPUs dont simply break lol
no unless you oc the living mess out of it or run at a high voltage

 

[roland12321]

I do lose power every now and then.. My house has regular power cuts, like I'd say 5-10 times a year. Could this contribute to the problem?

The power supplies I use are of good quality. The one my GTX died in was a 700W gamer PSU and cost around €75 if I remember correctly.

 

[roland12321]

CPUs dont simply break lol

Yeah but graphic cards do simply break, which I just don't understand.. CPUs can get pretty hot too, and they can consume quite a bit of power. The feeling I get is that they want you to buy a new graphics card so they get more money.. Old VGAs never used to break. I had a GF 4 MX, GF 4 TI, TNT2, of which all of them lasted years and years. You'd think that with today's technology they could make one that runs for decades..

Anyway I guess I have one of two choices: 1. Prevent my new card from going higher than 70°, try to give the PC some rest every now and then, and then hopefully it will last me 5 years or so.

or choice number 2. Keep buying cards that give highest performance for lowest price every time.. =.=

 

[Fallengod]

I do lose power every now and then.. My house has regular power cuts, like I'd say 5-10 times a year. Could this contribute to the problem?

The power supplies I use are of good quality. The one my GTX died in was a 700W gamer PSU and cost around €75 if I remember correctly.

Being 700watt, a "gamer" PSU and cost do not necessarily equate to quality just fyi. I dont know what brand that is, but that is for another discussion.

And yes, power surges can definitely kill computer parts. Do you have your computer plugged directly into the wall? If so, I would not do that. Surge protector man. I use an APS Battery backup with mine. If yours is plugged directly into the wall, that is a very bad idea.

 

[DarkRogue]

Keep your computer systems free of dust (clean regularly), and don't run it 24/7 (turn it off at least at night). Anything will decay over time, but if you always have power running through it, then it will degrade faster.

Anectodally, I can see why people would recommend this, but no one I know has actually had a GPU failure by running their systems 24/7.

My rig is in my sig, still the same 8800GT purchased exactly 4 years ago, run 24/7 in my system, with the exception of maybe 10-15 days a year I'm not at home on out-of-state trips. Granted, it does have an Accelero S1, and I do de-dust my PC every 2-3 months, but it's still running good as new.

Friend of mine has an ATI HD4850, so about 3.5 years old now, 24/7 but his PC does go to sleep each night.

I'd be more concerned about those "regular power cuts" ...
Is your PC turned on during these cuts?

 

[T_Yamamoto]

loss of power will probably be the biggest factor at the moment

 

[roland12321]

Granted, it does have an Accelero S1, and I do de-dust my PC every 2-3 months, but it's still running good as new.
...
I'd be more concerned about those "regular power cuts" ...
Is your PC turned on during these cuts?

Yes, 3 out of 4 times my PC would be on. I'll get one of these surge protectors, would this one be okay?

DarkRogue, if you don't mind me asking, how hot does your card get during gaming?

 

[busydude]

Get a better PSU.

 

[DarkRogue]

If your PC is ON during the power cuts, it could cause all kinds of problems.
I would recommend a UPS of some sort, so that it would at least give you a few minutes to shut down properly.

As far as temps of my card, I have not measured it very much lately because I haven't had a need to. However, back when it was newly installed, I benchmarked it and measured temps of my card then.

The Accelero is simply a beast. Case in point, my overclocked load temps are the same as my stock idle temps. With the stock card, it was measuring 58C idle and 79C load. After I mounted the Accelero, idle was 39C with load of 58C. This was measured via ATITool while it was running the fuzzy cube to heat up the card and scan for artifacts with the overclock.

Not quite Furmark, but it should be close.
*edit* The 'stock card" I mention in the quote is referring to the stock eVGA heatsink/fan thing before I swapped it out. */edit*

I would say 60C is a good estimate for my GPU's load temps. Keep in mind though, that this GPU is super, super slow by today's standards.

 

[T_Yamamoto]

60C is a good estimate for my GPU's load temps. Keep in mind though, that this GPU is super, super slow by today's standards.

not to threadjack, but i finally got a discrete graphics card (it was free)


a x1300 lol

 

[VirtualLarry]

In other news, I moved my quad-GPU cruncher over to a relative's place, and it stopped working.

 

[bononos]

........ Question 1: Why do my cards keep dying so quickly?! Is it just bad luck? Or is it that NVIDIA just doesn't like me..?

So I decided I'm going to buy an AMD card next, either the HD6850 or HD7850, depending on the price.

Now I heard somewhere some time ago that if you underclock a card its temperature will drop and it will last you much longer. Is this true? Will it last longer? So...

Question 2: If I buy a factory-overclocked graphics card (so it has nice cooling), and then I underclock it to its original speed or slightly less, do you think it will last much longer than if I buy the standard non-overclocked version?

Any help will be greatly appreciated!!

Roland

I had some bad luck with video cards as well.
I learnt that the gpu temp is not the only thats important since gpus for the last few generations at least should be able to tolerate temps in the 80s since the core shutdown threshold is typically >120c. The problem is the high power consumption and resulting heat at the vrms which is not fully addressed by the gpu heatsink designs.

The 6 series era coolers had small fans (designs tended to suck air in and push air through the shroud) which could not push enough air through to reach those vrms sufficiently. And worse it was hot air that was being blown over the vrms.

I might've shortened the life of 1 card when I changed the tim (white goop to AS5) on the gpu and the memory chips (cooler also made contact and cooled memory chips) which helped the gpu but the design of the heatsink mountings was such that the contact with the memory chips was possibly not as tight as it should be with the very thin layer of AS5 - so thick white goop was possibly the better choice there.

Nowadays the heatsink design on gpus tend to have larger fans and fins with larger gaps in between to allow better airflow and also to make dust less of a problem - since small fans+'tighter' heatsink fin design(in the ubiquitous radial type) is perfect for getting clogged up. And the newer fans also tend to be pullers so cool air comes from outside the shroud over the vrms first and then out through the fan.

Getting a overclocked variety *might* get you a binned gpu which should tolerate higher temps but it wouldn't have any effect on the vrm cooling so you'll have to decide for yourself if the cooling is sufficient using educated guesswork. You don't have to underclock it because new cards have an low/idle speed feature that underclocks the gpu when you are not gaming.

Getting back to the 6600 which you had - iirc they had no idle speeds so they ran at full speed all the time. So if you left your pc on to download all night then they just ran hot the whole night.

Best thing to do is install dust filters to cut down on the dust or vacuum often, ensure good airflow with an extra fan or 2, cut down on clutter by tying up cables etc in addition to making sure your new card has a good cooling design.

 

[T_Yamamoto]

In other news, I moved my quad-GPU cruncher over to a relative's place, and it stopped working.

ill take the parts off your hand

 

[aaksheytalwar]

There can only be 2 reasons:

1. Power Supply, buy a top notch brand and something far more powerful than you need. A 75 euro and 700 watts for that gotta be crap.

2. Use aftermarket cooling.

The other reasons mentioned here are myths, not valid reasons. Using a computer 24x7 is not a problem, a few hours or rest to a computer every weeks is good enough.

As long as you only game a few hours a day, and with intervals, that can't be the reason either.

De dusting might not hurt, but that can't be the reason nevertheless.

Power cuts 5-10 times a year are unlikely to mean much. Just get a good UPS and that can't be the reason either after that.

I am 95% sure it is your PSU, the remaining 4% is over heating, 1% it could be something else

 

[imaheadcase]

The only thing i've ever had stop working on a computer is headphone jacks/onboard NIC and power supplies.

 

[T_Yamamoto]

you dont need to buy a big psu....