HD4870 Problems

[Fafara]

In May 2009 i bought a brand-new the HD4870, it worked fine until last month when my problems started. The said problem (supposedly) is overheating, i've heard things about the HD4870 having issues with heat, being it's minimum "idle" temperature 70°C, some say it can handle up to 120°C without breaking, but i don't believe that.

Anyway, when i am playing a game (e.g. Counter Strike Source) the game crashes (together with the sound that freezes as well making it unbearable to hear), but it doesn't crash to the desktop, it actually (in a way) turns off the computer but not completely, the monitor fades out and my HD4870 becomes REALLY noisy, it looks like the fan is roughly spinning at 100%.

After that i started to run the game with Everest open to watch the temps, my VGA's temp easily go up to 80°C while playing, which by default (as everyone says) shouldn't be a problem... but it is.

So, basically, the computer fades out, the VGA starts to spin really really fast and becomes really noisy, the sound freezes (like repeatedly playing the last fraction of second of the last sound played) and i have to shut the computer off by unplugging it.

Do any of you know what's causing it? do i have to buy another one?[

Thanks.

 

[kalrith]

I'd start by taking some canned air and blowing out your heatsink and fan. If there's a bunch of dust in there, that could be causing the gpu to overheat. Are your temps at 80° when the system shuts down? If you can't play the game and see your temps at the same time, then I recommend running Furmark with Everest open to see how high your temps go, especially what temp it's at when it locks up.

BTW, is it still under warranty?

 

[f4phantom2500]

is your gpu overclocked? if so put it back at stock and see what happens. if anything else is overclocked try running the whole system at stock.

 

[Fafara]

I'd start by taking some canned air and blowing out your heatsink and fan. If there's a bunch of dust in there, that could be causing the gpu to overheat. Are your temps at 80° when the system shuts down? If you can't play the game and see your temps at the same time, then I recommend running Furmark with Everest open to see how high your temps go, especially what temp it's at when it locks up.

BTW, is it still under warranty?

No it isn't
Can the dust be a real problem? Because yes it's pretty dusty (i don't clean very often).
About the temp being 80°C when it shuts down, i guess so, since (as Everest says) it's 81°C as we speak, so yeah it is this hot when i shut it down.
I'll go after this Funmark and tell you what i get from it.
Also thanks for the quick answer

is your gpu overclocked? if so put it back at stock and see what happens. if anything else is overclocked try running the whole system at stock.

No, i don't overclock.

 

[DaveSimmons]

I set the fan speed manually using profiles. The default fan speed on mine (Powercolor 512MB model from 2008) is quiet but hot. Setting it higher is louder but keeps the temp below 60C at idle and under 70C under load.

When not gaming you can use a second profile to set the fan back to low speed.

 

[kalrith]

Dust can make a HUGE difference! However, if your card is only in the 80s, then I think something else is wrong (it shouldn't shut down because of heat until 100° at least). I think Furmark tests for artifacts as well, so that should be helpful.

Also, be sure to blow any dust out of the RAM sinks on the card. It's possible that the gpu isn't getting too hot, but the RAM is.

If blowing out the heat sink and fan doesn't help, then I'd complete unhook the card and reseat it. This will involve removing both of the PCI-e power connectors, removing the DVI cables, and removing the card from the motherboard. Then plug everything back together.

I've read about cards all of a sudden "going bad," and sometimes reseating them does the trick. If a PCI-e power connector came loose, then the card might not be getting enough power. Unplugging everything and plugging it back it will address both issues.

 

[Fafara]

I'd start by taking some canned air and blowing out your heatsink and fan. If there's a bunch of dust in there, that could be causing the gpu to overheat. Are your temps at 80° when the system shuts down? If you can't play the game and see your temps at the same time, then I recommend running Furmark with Everest open to see how high your temps go, especially what temp it's at when it locks up.

BTW, is it still under warranty?

HOLY CRAP DUDE
I just tested Furmark, ONE second after pressing "Go!' in the software my computer faded out and after that my VGA was screaming the wind out of it.
I am infinitely glad that for some reason, either the VGA itselt or the Mobo turned it off, because i believe the temps went from 80°C to 90°C or more in half of a second, i could have lost it lol.

Well, i think it's official, it's a matter of temperature.
Aside from blowing it (which i'll do now) what else can i do to prevent this thing from happening?

Also, i just noticed two LEDs lighted up on the back of the VGA, i googled it and i found this:

D1601 - Red LED On, shows critical temperature fault

D1602 - Red LED On, shows External power connector A was removed

D1603 - Red LED On, shows External power connector B was removed

D601 - Red LED On, shows critical Core power fault

I dont know which led is which, but if it was either D1602, D1603 or D601 i wouldn't be here typing this, the only option is D1601.

I set the fan speed manually using profiles. The default fan speed on mine (Powercolor 512MB model from 2008) is quiet but hot. Setting it higher is louder but keeps the temp below 60C at idle and under 70C under load.

When not gaming you can use a second profile to set the fan back to low speed.

My CCC doesn't open, but when i could open it, i couldn't play with the fan speed since there's no fan speed slider.
All you can see is Activity and another one.
It lacks of Fan speed.

 

[kalrith]

Here are some things to try:

1. Do this first: Blow out dust before doing anything else
2. Do this second: Unplug everything from the card. Remove the card from the mobo. Reinstall the card, and plug everything in again
3. The rest are optional: Remove heatsink and fan, clean off all thermal paste, and reapply better thermal paste (I use Arctic Silver 5)
4. If fan has two-pin cable, use a molex-to-3-pin adapter and plug the fan into the 3-pin side so that it connects with the two cables going to the 3-pin plug. This will make the fan run at 100% and should allow you to stress the card safely and see if the fan at 100% can keep the card cool at full load
5. If the fan at 100% keeps the card cool and doesn't crash, then find some way to mod the fan settings. If CCC doesn't do it, then you might try rivatuner or even flashing the card's bios if you don't mind the risk

It could actually be D1602 or D1603. A card will function just fine in 2d mode with an external connector disconnected. I experienced this with a 6600GT. I did some work in the case and then didn't play games for a few weeks. I tried playing a game, and the screen would go completely black after 1-3 minutes of playing. I couldn't figure out what was wrong or what I had changed. Then I noticed that I had unplugged the external power connector 2-3 weeks prior and never plugged it back in. The card functioned perfectly in 2d mode.

The above is why I recommend completely unhooking and rehooking your card and all the cables associated with it.

 

[superccs]

Here are some things to try:

1. Do this first: Blow out dust before doing anything else
2. Do this second: Unplug everything from the card. Remove the card from the mobo. Reinstall the card, and plug everything in again
3. The rest are optional: Remove heatsink and fan, clean off all thermal paste, and reapply better thermal paste (I use Arctic Silver 5)
4. If fan has two-pin cable, use a molex-to-3-pin adapter and plug the fan into the 3-pin side so that it connects with the two cables going to the 3-pin plug. This will make the fan run at 100% and should allow you to stress the card safely and see if the fan at 100% can keep the card cool at full load
5. If the fan at 100% keeps the card cool and doesn't crash, then find some way to mod the fan settings. If CCC doesn't do it, then you might try rivatuner or even flashing the card's bios if you don't mind the risk

It could actually be D1602 or D1603. A card will function just fine in 2d mode with an external connector disconnected. I experienced this with a 6600GT. I did some work in the case and then didn't play games for a few weeks. I tried playing a game, and the screen would go completely black after 1-3 minutes of playing. I couldn't figure out what was wrong or what I had changed. Then I noticed that I had unplugged the external power connector 2-3 weeks prior and never plugged it back in. The card functioned perfectly in 2d mode.

The above is why I recommend completely unhooking and rehooking your card and all the cables associated with it.

^ agreed ^

Yeah the same thing was happening to me with my 4850. The temps moved really fast, too fast to account for the thermal inertia of the Arctic Cooling heat sink I had installed (poorly). I took it off got rid of the stock grey paste and Arctic Silvered that mother on there and temps dropped through the floor. No more driver crashing, no more overheating. All is good. Started OC'ing the little beast and got well into 4870 territory (750/1050) with no change in temps (40's idle 50s gaming). You might want to look into it The Passive Arctic Cooling Rev 2 GPU cooler is pretty whicked if your case has some airflow. http://www.newegg.com/Product/Produc...82E16835186016

Hope this helps.

 

[Fafara]

Here are some things to try:

1. Do this first: Blow out dust before doing anything else
2. Do this second: Unplug everything from the card. Remove the card from the mobo. Reinstall the card, and plug everything in again
3. The rest are optional: Remove heatsink and fan, clean off all thermal paste, and reapply better thermal paste (I use Arctic Silver 5)
4. If fan has two-pin cable, use a molex-to-3-pin adapter and plug the fan into the 3-pin side so that it connects with the two cables going to the 3-pin plug. This will make the fan run at 100% and should allow you to stress the card safely and see if the fan at 100% can keep the card cool at full load
5. If the fan at 100% keeps the card cool and doesn't crash, then find some way to mod the fan settings. If CCC doesn't do it, then you might try rivatuner or even flashing the card's bios if you don't mind the risk

It could actually be D1602 or D1603. A card will function just fine in 2d mode with an external connector disconnected. I experienced this with a 6600GT. I did some work in the case and then didn't play games for a few weeks. I tried playing a game, and the screen would go completely black after 1-3 minutes of playing. I couldn't figure out what was wrong or what I had changed. Then I noticed that I had unplugged the external power connector 2-3 weeks prior and never plugged it back in. The card functioned perfectly in 2d mode.

The above is why I recommend completely unhooking and rehooking your card and all the cables associated with it.

So here is the current situation:
Since i didn't make this computer, i just bought all the hardware and paid a store to mount it, i don't know how i am supposed to remove the card from the motherboard, i didn't want to force anything fearing to break some plastic piece.

So i couldn't blow it.
But... when i was going to plug all the cords back i had some trouble with the VGA, when i turned on for the first time it made a weird sound, not the sound i always hear when i turn it on, but a different one, like if the fan was really really weak and then the computer turned off.
then i noticed the leds, i tried to unplug and plug the card's power supply again and it kinda worked, one of the leds is on right now.
I dont know why it is on or for what it stands for.
I should have stated in the beginning that i don't really know how to do this hardware stuff.
I am considering sending the computer back to the store but it's kinda expensive.

So... can you explain this two-pin cable thing?

By the way the cables i use for the card are these

Also after googling i found that the led lit is D601
Yeah, now i have a bigger problem.

 

[blanketyblank]

So here is the current situation:
Since i didn't make this computer, i just bought all the hardware and paid a store to mount it, i don't know how i am supposed to remove the card from the motherboard, i didn't want to force anything fearing to break some plastic piece.

So i couldn't blow it.
But... when i was going to plug all the cords back i had some trouble with the VGA, when i turned on for the first time it made a weird sound, not the sound i always hear when i turn it on, but a different one, like if the fan was really really weak and then the computer turned off.
then i noticed the leds, i tried to unplug and plug the card's power supply again and it kinda worked, one of the leds is on right now.
I dont know why it is on or for what it stands for.
I should have stated in the beginning that i don't really know how to do this hardware stuff.
I am considering sending the computer back to the store but it's kinda expensive.

So... can you explain this two-pin cable thing?

By the way the cables i use for the card are these

Also after googling i found that the led lit is D601
Yeah, now i have a bigger problem.

How much is kinda expensive?

 

[Schmide]

DO NOT TEST FURMARK until you get your temps under control!!!

 

[Fafara]

How much is kinda expensive?

27ish dollars.
Just letting you know i am Brazilian and that price converted is pretty expensive here.

DO NOT TEST FURMARK until you get your temps under control!!!

Well, what can i say, too late.
I think the led lit when the card shut down when i was using Furmark and i didn't notice.
Anyway, it sucks to be me right now.
D:

 

[blanketyblank]

27 isn't really enough for a new card comparable to a 4870 so I guess it's still the cheapest option. Also from the sounds of it, you probably don't have much choice as you can't even take the card out yourself.
You don't know anyone locally you trust with computers?

 

[Fafara]

27 isn't really enough for a new card comparable to a 4870 so I guess it's still the cheapest option. Also from the sounds of it, you probably don't have much choice as you can't even take the card out yourself.
You don't know anyone locally you trust with computers?

Actually 27 dollars is what i pay for the guy to stay with the computer to take the diagnosis.
And i just got told in another forum that D601 stands for overheating and not core power fault.

Confusing.

 

[f4phantom2500]

they make mesh filters that you can install on your case fans to keep a lot of that dust out of your computer. of course, you should clean them off every now and then...but it beats blowing huge dust bunnies out of your computer.

 

[NoQuarter]

Just to warn you removing the heatsink to replace the thermal paste can void the warranty on the video card so you should only do that if the warranty covers that or if it's out of warranty.

You should be able to blast the heatsink on the 4870 with a can of air without removing it though, not as well but enough to get it clean, and might as well hit the CPU and inside the PSU while you're in there. If you've never done it before, 1 year is a long time for dust to build up.

If you do need to remove the video card it's easy, only thing you're missing is the tab on the back of the slot it's in that's locking the card in. The tab either needs to slide back or be pushed sideways and it'll let the card out.

 

[Fafara]

Just to warn you removing the heatsink to replace the thermal paste can void the warranty on the video card so you should only do that if the warranty covers that or if it's out of warranty.

You should be able to blast the heatsink on the 4870 with a can of air without removing it though, not as well but enough to get it clean, and might as well hit the CPU and inside the PSU while you're in there. If you've never done it before, 1 year is a long time for dust to build up.

If you do need to remove the video card it's easy, only thing you're missing is the tab on the back of the slot it's in that's locking the card in. The tab either needs to slide back or be pushed sideways and it'll let the card out.

Okay, i understand, but i still am worried about doing that because last time i didn't even remove the card and when i turned it back on the led was lit.
Even though Furmark might have caused it.

I also downloaded Rivatuner (something i should have done long time ago) and GPU-Z, right now over here where i live it currently is 13°C (57°F) so it's pretty cold and GPU-Z displays the temps as (computer is idle):

If i increase the fan to 45% via RivaTuner i get the following minor change of (still idle):

Now i open Trackmania with all the graphics maxed out, everything i can ask for and i play for 10 minutes and i get the following:

Seriously, which point am i missing?
Previously Trackmania HAS crashed but if it did at such temperatures (even though it might be too hot, but nothing compared to the wrong info i was getting from Everest) it wasn't supposed to crash.

Can the problem be something else other than the VGA?

Thanks.

Edit: those VDDC temps are all high, i did some google and found out they have something to do with the PSU.
Might it not be related with my led problem? why is it so hot?

 

[VirtualLarry]

Edit: those VDDC temps are all high, i did some google and found out they have something to do with the PSU.
Might it not be related with my led problem? why is it so hot?

I don't know about the PSU, but those temps on the VRMs seem a tad high to me too.

Post a picture of your exact card model if you can, I'm curious about what kind of heatsinks or cooling they have for the VRMs.

 

[Fafara]

exactly this one

 

[VirtualLarry]

Well, it looks like it has a VRM heatsink, but a rather puny one.

 

[Fafara]

I suppose the VDDC temps will be something i'll be forced to live with until i buy another card, if you google "VDDC temps" you'll see that 90% if the results are about the 4970 with this issue, it's a feature (more like a failture) of the card.

Well i think that in order to prevent my computer from restarting everytime i play something is by temporarily increasing the fan speed.

If anyone has anything to say, say it (i'll be glad to answer/discuss) because i think the thread is coming to it's end soon.

 

[Schmide]

Edit: those VDDC temps are all high, i did some google and found out they have something to do with the PSU.
Might it not be related with my led problem? why is it so hot?

Nahh those VDDC temps are quite normal, they're designed to run to near 120c (when throttling starts)

If you're still able to run Trackmania and such, I think your card is power starved. What PSU do you have and how old is it?

BTW I put the same card into my brother's system.

 

[Fafara]

Nahh those VDDC temps are quite normal, they're designed to run to near 120c (when throttling starts)

If you're still able to run Trackmania and such, I think your card is power starved. What PSU do you have and how old is it?

BTW I put the same card into my brother's system.

My PSU is Zalman 600W
I never had problems with it, i dont see a reason for my card to be starving of power since all the cables are plugged.

I also saw someone's GPUZ displaying VDDC temperatures that were colder than the core itself.

the PSU is as old as the card.

 

[Schmide]

Have you checked your ram (Memtest86)? Maybe try one stick or the other?

What about other components? (list?)