GTX 660 whines and causes lag

[mrmonkeybiz231]

Hello,

I just for the first time built a desktop computer. I'm very happy with what I have, except I am having one issue. My Gigabyte GTX 660 has recently started making a whining sound, and now suddenly my games lag occasionally. I have been playing Crysis 2 at Ultra settings, and the first few days there was no lag and no whine. Now the card whines, and the game lags here and there. When the game lags, the whine changes pitch and makes a sort of "edgy" sound (similar to when a hard drive is giving up the ghost).

Also, I did a benchmark and found that the core clock was 549MHz and nVidia's website says it should be 980MHz. The memory clock came out at about 3000MHz and nVidia's website also says it should be roughly twice that. I'm not sure if this really means anything, but it seems very concerning.

My question is, what should I do? Should I see about doing an exchange for another? Any help would be super appreciated!

 

[srhnd]

There is no taking chances with Furmark here. Gtx 660 is not a 50 dollar graphics card to burn up in Furmark or anything. That was 3 years ago with less throttle systems on the cards or when people stupidly overvolt their cards with extreme overclocks that are most likely to fail. Furmark is a good tool to load up the power draw on your graphics card and has not caused any issues neither on my system nor on the systems of my friends who have been using it similarly.

If you have experience in soldering electronics better not to talk about overvoltages. This switching PSU can cause harmonic overvoltages that might ... estimate the peak values and durations of switching overvoltages. However, their application can lead to harmonic and transient problems.

If your GPU or PSU fails under Furmark it only means one thing: It was a faulty piece of hardware to begin with and it was probably going to fail anyhow. So the faster you carry out this test like right after you set up your PC the faster you can replace the card or other parts within first 30 days. Or you might simply miss on this and have your card die in a later time only to be more pain in the butt.

This is actually a risk minimization and elimination method as opposed to what you consider it to be. I actually think just testing it in 2-3 games will leave the risk paths open. You need to test it under both extreme and casual scenarios to know how your card performs, how hot it gets in which situations etc.

I have learned a great deal about my 2 660s during my testing processes so I know what to expect from my card and what not to expect.

Furmark is a good way to test the maximum power draw on graphics card and hence the peak temperatures. Running it along with Prime95 for 10-15 minutes is NOT dangerous, it is totally safe for a NORMAL and HEALTHY system and it will prove 100% that if you pass it your PSU is just doing perfect and you will know how hot your cards get so you can create a fan profile etc. If it is not a healthy and normal system then this is your chance to find out about it.

Just because there is a 1 in a 1,000,000 chance of failure you cannot tell people it breaks your card or PSU or tell them not to test their systems. That kind of thinking is nonsense. It will be better for a person to test out his new PC so he can replace the parts in the beginning before companies act like "Oh no, it is too late".

We can talk about everything about airplanes too, how they are wired up, manufactured, how parts work individually etc. But airplanes still crash, and it is still the safest way to fly, people will still fly. Those who don't use it because it can crash are probably not getting along well with statistics and simple logic.

 

[vikilaba]

There is no taking chances with Furmark here. Gtx 660 is not a 50 dollar graphics card to burn up in Furmark or anything. That was 3 years ago with less throttle systems on the cards or when people stupidly overvolt their cards with extreme overclocks that are most likely to fail. Furmark is a good tool to load up the power draw on your graphics card and has not caused any issues neither on my system nor on the systems of my friends who have been using it similarly.

If you have experience in soldering electronics better not to talk about overvoltages. This switching PSU can cause harmonic overvoltages that might ... estimate the peak values and durations of switching overvoltages. However, their application can lead to harmonic and transient problems.

If your GPU or PSU fails under Furmark it only means one thing: It was a faulty piece of hardware to begin with and it was probably going to fail anyhow. So the faster you carry out this test like right after you set up your PC the faster you can replace the card or other parts within first 30 days. Or you might simply miss on this and have your card die in a later time only to be more pain in the butt.

This is actually a risk minimization and elimination method as opposed to what you consider it to be. I actually think just testing it in 2-3 games will leave the risk paths open. You need to test it under both extreme and casual scenarios to know how your card performs, how hot it gets in which situations etc.

I have learned a great deal about my 2 660s during my testing processes so I know what to expect from my card and what not to expect.

Furmark is a good way to test the maximum power draw on graphics card and hence the peak temperatures. Running it along with Prime95 for 10-15 minutes is NOT dangerous, it is totally safe for a NORMAL and HEALTHY system and it will prove 100% that if you pass it your PSU is just doing perfect and you will know how hot your cards get so you can create a fan profile etc. If it is not a healthy and normal system then this is your chance to find out about it.

Just because there is a 1 in a 1,000,000 chance of failure you cannot tell people it breaks your card or PSU or tell them not to test their systems. That kind of thinking is nonsense. It will be better for a person to test out his new PC so he can replace the parts in the beginning before companies act like "Oh no, it is too late".

We can talk about everything about airplanes too, how they are wired up, manufactured, how parts work individually etc. But airplanes still crash, and it is still the safest way to fly, people will still fly. Those who don't use it because it can crash are probably not getting along well with statistics and simple logic.

Okey your GPU is designs using TSMC's 28nm but my GTX470 is 40nm tehnology. This allows for a substantial reduction in temperature, thus lowering overall power consumption and allowing for higher clock speeds. Games like Crysis and Metro fully exploiting all the resources so that there is no need for a dedicated tests. I'm crying for my dead second GTX 470 from SLI configuration

 

[srhnd]

On Crysis 2 or 3 my temps don't go above 67 max. In Furmark I see 73-74 max. This indicates to a higher power draw and core utilization and hence more comprehensive testing. OP's card is a GTX 660 so your 470 is totally irrelevant when we are talking about Furmark.

 

[mrmonkeybiz231]

Well I changed the power supply to a more reliable one and I am still having the issue with the card. I'm thinking of either trying the nvidia inspector thing or possibly swapping out my card for a Radeon 7870, or maybe an r9 270x. Any suggestions?

 

[iam2thecrowe]

Well I changed the power supply to a more reliable one and I am still having the issue with the card. I'm thinking of either trying the nvidia inspector thing or possibly swapping out my card for a Radeon 7870, or maybe an r9 270x. Any suggestions?

just do the nvidia inspector thing, install inspector, copy the template in the link provided, put in your clock speeds and voltages of your cards default clocks. it takes 10 minutes and is a surefire fix for the clock drops, and possible fix for the crash to desktop. probably wont fix any other issues though. if you can return it for a 7870 easily, that wouldnt be a bad idea, they are a slight bit faster.

 

[srhnd]

Disable any overclocks you have applied and if it does not work I suggest r9 270x. It is performing better than 660.

 

[vikilaba]

Well I changed the power supply to a more reliable one and I am still having the issue with the card. I'm thinking of either trying the nvidia inspector thing or possibly swapping out my card for a Radeon 7870, or maybe an r9 270x. Any suggestions?

What I said was that the PSU is because my system worked with low power PSU but turns off when in the game(Metro 2033). Now I have 1000W:lol:

 

[vikilaba]

Yes AMD Radeon is better for ordinary users, Nvidia still advanced and demanding things... For me GTX 590 is very well option. Why is it not on the list from srhnd post?

 

[vikilaba]

On Crysis 2 or 3 my temps don't go above 67 max. In Furmark I see 73-74 max. This indicates to a higher power draw and core utilization and hence more comprehensive testing. OP's card is a GTX 660 so your 470 is totally irrelevant when we are talking about Furmark.

I do not agree with it. My experience is that high-temperature damage to all hi-end graphics card from any series they are. Furmark is a stress to the system and get it at your own risk.

 

[srhnd]

On Crysis 2 or 3 my temps don't go above 67 max. In Furmark I see 73-74 max. This indicates to a higher power draw and core utilization and hence more comprehensive testing. OP's card is a GTX 660 so your 470 is totally irrelevant when we are talking about Furmark.

I do not agree with it. My experience is that high-temperature damage to all hi-end graphics card from any series they are. Furmark is a stress to the system and get it at your own risk.

It is risky after 80 degrees with 97 C being the max for this card. On a 10-15 minute furmark run if you are reaching 80 you should simply either setup some cooling system, reposition the case or replace the card or go buy another model. A temperature up to 80 has no damage whatsoever unless it has been applied for a long long period of time which is probably measured in days.

You can believe in whatever you want to believe but if you say that 10-15 minute furmark run is risky for or can damage a normal and healthy gpu then you are wrong and misleading. You don't need to be superstitious with technology and one bad experience on your side does not indicate to anything that can be generalised. Furthermore a gtx 470 and gtx 660 are 2 generations apart, with many modifications and improvements in terms of power and temperature control on 660's side.

It makes sense to stress-test it at day 1 so you know whether or not you got good parts. Not testing it may be OK if it is good but if not maybe you will miss on it simply because it is not pushed enough in the games you play. Not everyone has to install and play Crysis or games like that for that matter. Furmark is a convenient tool and to a big extent it eliminates the risk of having a faulty GPU that you don't know about. You can insist on your superstitious tech beliefs, just don't impose it on others as it is useless.

 

[17seconds]

Hey guys, I cleaned up this thread. Let's keep focused on helping mrmonkeybiz231 with his problem.

 

[vikilaba]

Furmark is NOT dangerous on a proper system. I have let Furmark run for an hour on my SLI setup without any issues. It is not a strain over the limit it is same as a game utilizing a lot of the capacity of the graphics card except the VRAM usage, which stays low in Furmark.

Unless you are overvolting the card Furmark is safe and hey I have even been there, volted my cards to both 1.212v from 1.15-1.62 regular voltage and to 1202Mhz core and 6600 Mhz memory, furmark for 30 minutes, no problem whatsoever.

@viki the fact that it caused something to your computer or to 2-3 guys' computers does not mean it will to anyone else's PC.This is like saying don't use airlines because it crashes and you die, yeah true it happens but come on.

Even in a 10 minute run of Furmark you will see a temperature that you will not see in any game at all, that is how you know it is all in good shape.

You can also use MSI Kombustor or OCCT for that matter. Used them too never had any problem.

Mind you my SLI setup is packed on a mini ATX motherboard where there is not even 1/5 cm between my cards, if I push top card a little forward its cooling fan's blades start hitting the back of the second GPU, lol. Still no problem with any Furmark runs or intense benches.

Furthermore, both Nvidia and AMD throttles the cards usage on programs like Furmark so you are more than safe. When you get new PSU use Furmark OR Kombustor at the same time with Prime95 for at least 10 minutes to see how it goes.

It is like saying Prime95 kills your card, it does not kill your card but your overvolting and heat can, since OP is not doing that it is safe.

I take chances when chasing in my car Golf GTI but with PC just playing games

you take chances with your life and a more expensive car than you do playing games on your pc? you have interesting priorities.....

Hey guys, I cleaned up this thread. Let's keep focused on helping mrmonkeybiz231 with his problem.

Thx for cleaning this garbage. We focused on that global problem

 

[mrmonkeybiz231]

Well... Now that all that is cleaned up, I have an update. I got tired with dealing with the issue with the Gigabyte gtx 660, so I swapped it for a Radeon 7870. I was not pleased with that card at all, so I decided to throw a little more money at this and got an MSI Gtx 760. So far, so good! Played Assassin's Creed 4 for an hour straight with no issues, but I suppose only time will tell if the problem is really fixed.

 

[17seconds]

Well... Now that all that is cleaned up, I have an update. I got tired with dealing with the issue with the Gigabyte gtx 660, so I swapped it for a Radeon 7870. I was not pleased with that card at all, so I decided to throw a little more money at this and got an MSI Gtx 760. So far, so good! Played Assassin's Creed 4 for an hour straight with no issues, but I suppose only time will tell if the problem is really fixed.

Sounds good. If you're done, why don't you pick a Best Answer and we'll wrap this one up?

 

[vikilaba]

Is this card worked now or what (GTX 660)? Solution is new graphics card (GTX 760)! IS THIS BEST ANSWER FOR AS, SOUNDS BAD FOR ME 🙁

 

[hang-the-9]

Dont run any PC stress test, only Prime95 for system stability you can do it. Furmark and similar are dangerous for hardware! From my experience as an electronic specialist. Each strain over the limit is harmful and causes permanent harm on electronic circuits. Good test is Crysis(1,2,3) for examine how it really stable computer in gaming(process of playing games). The only way I could see how it helped me new thermal grease from Cooler Master. 10 degrees celsius colder temperatures... on GPU!

I've never ran into any issues with system tests, Futuremark does not stress the system any more than Prime95. Prime95 can peg the CPU and ram at 100% usage for hours, you think that is better than anything the other tests do? The only thing the other tests do different is that many of them also test the hard drive speeds and video card, but that again is no more than any type of enthusiast regular use would be. Converting an HD video, playing a high-end game at full settings, etc...

 

[vikilaba]

Yes, huuuuuuh god. Stressing out the PC is good homework!