ATI HD4870 Alarm - I'm stumped...

TechLarry

RIP [H] Brother - June 1, 2022
Joined
Aug 9, 2005
Messages
30,481
I posted about this yesterday, but I can't seem to find it for some reason...

Bought a 4870 and installed. Whenever I drive away from the pits in iRacing, I start to get what I think is an overtemp alarm from the 4870.

I am not seeing any graphic anomolies.

Took the card back today and got a replacement new card.

Just to be sure, I also wrapped the system up in a new Antec Nine Hundred case.

4 120mm fans and the top 200mm fan. To start I put them all on high. You could chill beer inside this thing.

Installed a fresh copy of Windows Vista (it was previously running XP)

Same damned problem. I know it's a hardware alarm from something inside the case. I am only assuming it's the video card.

Motherboard and cpu temps are fine as far as I can tell.

I did the fan trick and have tried upwards of 80%, which is ridiculous, but the alarm still goes off.

It's interesting that the alarm silences on the straights, but when in corners (which causes considerably more graphic updates), the alarm starts to go off. When I get to the next straight, it goes away.

A8n32-SLI Deluxe, 4800x2, Antec 650w supply, 2GB standard ram, 300GB seagate drive, XtremeGamer X-FI sound card. HD4800.

If I remove the ATI and put the nVidia 7800GTX back in the problem goes away.

BIOS and all that stuff is up to date.

I tried removing the sound card to make sure it wasn't a PCI conflict or something silly like that. No joy.

I am totally baffled.
 
He's got an antec 650W, which really should be overkill for his system.
 
There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.
 
Is the alarm coming from the motherboard? If so, have you checked alarm codes in your mobo manual?
 
There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.

hmmm, I don know about that, minimum depends on what parts you have in the system, there is no need to get a 800 watt if your system runs 300 watt under load.
 
There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.

Might want to try a killawatt before you buy another PSU. I have a "feeling" that your feeling is off :p


Anywho, The only high temp alarm you would get form the card would be the fan spinning at 100% sounding like a blow dryer. Most likely it is the high temp alarm on your motherboard. It's adjustable in the bios.
 
There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.

Yes there is. Have you ever measured the power draw of a system? You'll probably find it is way less than you think. My UPS (which measures power) says that my system uses in the realm of 300 watts when I'm playing a 3D game. Now that's the total draw of the tower and the display (which accounts for about 50 of that). Also, that's power coming in, not power being provided to the components. PSUs aren't 100% efficient (mine is about 80%) so it is only actually outputting about 80% of that.

That is with a 26" LCD, a GTX280, 4 harddrives, a soundcard, and a good number of USB devices.

If you have a quality 1000 watt PSU, as in one that can actually output 1000 watts for long periods of time, I don't think you can build a desktop system that is more than it can handle.

Computers really don't use as much power as some people seem to think. There's three reasons why bigger PSUs are recommended:

1) Safety. You don't want to run something at it's maximum. Also, just because your computer load is overall a given level, it doesn't mean it can't peak above that. So if my computer seems to top out at 300 watts, I still want capacity over that to ensure overshoot and good operation.

2) Efficiency. Most PSUs are the most efficient at right around half their max load. Also, if you start to increase load too much, their fans have to kick up. Thus you probably want a PSU that's double your maximum sustained draw for best efficiency and lowest noise.

3) Cheap PSU tolerance. This si why graphics card companies tend to be so conservative. There are lots of crap PSUs out there that basically flat out lie about their wattage. When they say "500 watts" they may "This thing may output 500 watts, but only as extremely cool temperatures and then only for a bit."

If you have a good PSU that is serious about it's rating, you'll find that you are hard pressed to find a system that needs a very large one. That isn't to say you shouldn't over buy. Like I said, my recommendation is twice your maximum load, however let's not pretend a PSU is strained to do 50%. In fact, that is where it works the very best.
 
The PSU looks to be powerful enough to hold it's own against the HD4870, try pulling out the motherboard's cell which should reset all the configurations and see if you still get the alarm.
 
3) Cheap PSU tolerance. This si why graphics card companies tend to be so conservative. There are lots of crap PSUs out there that basically flat out lie about their wattage. When they say "500 watts" they may "This thing may output 500 watts, but only as extremely cool temperatures and then only for a bit."

This is my experience as well, I have an antec 850w Quattro powering my 4870 crossfire setup with a boat load of other devices and accessories as well. It wont need more than about 500w but it's nice to be safe. A crap brand of PSU rated for 850w may not power the same system for very long before dying.

The difference between a bad PSU and a good one can be quite significant.
 
Maybe get GPU-Z and see if the temps really are too high on the card? (make sure you check "Continue refreshing this screen while GPU-Z is in the background")

Also, have you tried fiddling around with which power connectors are plugged into the card and the motherboard? One of your 12V rails could be overloaded.
 
Bought a 4870 and installed. Whenever I drive away from the pits in iRacing, I start to get what I think is an overtemp alarm from the 4870.

I have to ask...simply because i've heard of worse...

Is the alarm an in game sound, like exceeding the pit lane speed limit (some racing games have)?

Does it happen in any other game? What happens if you leave the pit slower?

I once had a comptuer to fix which when you clicked on certain things would cause an awful noise and i was there for days replacing sound card drivers and testing for incompatability, when eventually found out it was simpl a windows sound theme and somoene had put a test sound of a helicopters rotor blades as a windows event, soooo stupid.
 
I'm pretty sure it's coming from the card. The alarm silences on the straights where there is little graphical load. As soon as you get to the corners, where the grapical load is the greatest, the alarm sounds.

I can't find ANY error information in the A8N32-SLI Deluxe manual at all. If there is an alarm on the board, it's not documented there.

I'm wondering if the close proximity of this card, which is known to run hot anyway, to the south-bridge is triggering something. I've heard the south bridges on this board run a bit warm anyway. Is there anyway to turn off the MB alarms to test this? I tried to turn off the CPU and MB temp sensor in the BIOS but I get the feeling that is for display purposes only.

Is the alarm coming from the motherboard? If so, have you checked alarm codes in your mobo manual?
 
That's one of the interesting things. The fan isn't going into overdrive, and I am not seeing any graphic anomolies.

Might want to try a killawatt before you buy another PSU. I have a "feeling" that your feeling is off :p


Anywho, The only high temp alarm you would get form the card would be the fan spinning at 100% sounding like a blow dryer. Most likely it is the high temp alarm on your motherboard. It's adjustable in the bios.
 
I have GPU-Z. Will that stay on-screen when the card goes into graphic mode?

There are two dedicated PCIe connectors for Graphic Cards on this supply.

Although it's not supposed to be necessary with this supply, I thought of adding in the optional 4-pin connector to the motherboard. This is only supposed to be for SLI setups but then again this board was made before we had cards drawing this kind of power.

Would it hurt anything?

Maybe get GPU-Z and see if the temps really are too high on the card? (make sure you check "Continue refreshing this screen while GPU-Z is in the background")

Also, have you tried fiddling around with which power connectors are plugged into the card and the motherboard? One of your 12V rails could be overloaded.
 
No, it is definitely a hardware alarm eminating from inside the case. It doesn't occur on the straights, only in the corners where the graphical load is the worse. It will beep on exit of the pits too, but well after the pit line and it's becaue there is a lot of left-right shift in the video, like in the corners.

I have to ask...simply because i've heard of worse...

Is the alarm an in game sound, like exceeding the pit lane speed limit (some racing games have)?

Does it happen in any other game? What happens if you leave the pit slower?

I once had a comptuer to fix which when you clicked on certain things would cause an awful noise and i was there for days replacing sound card drivers and testing for incompatability, when eventually found out it was simpl a windows sound theme and somoene had put a test sound of a helicopters rotor blades as a windows event, soooo stupid.
 
Ok, here's something else weird.

The "in use" GPU setting shows 500. If you are running defaults, should it not be 750?

Also, as an experiment, I tried underclocking the card all the way down (500/810) and retested. Not only did it not resolve the problem, but I saw ZERO change in performance.

I'm beginning to wonder if there isn't a compatibility issue between the 4870 and the A8N32SLI-Deluxe causing miscommunications between the card and the system.
 
Maybe not enough power?

Thats the issue. Its a Voltage alarm.

There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.

Jeez the PSU manufacturers brainwashed you good huh...

This is my experience as well, I have an antec 850w Quattro powering my 4870 crossfire setup with a boat load of other devices and accessories as well. It wont need more than about 500w but it's nice to be safe. A crap brand of PSU rated for 850w may not power the same system for very long before dying.

The difference between a bad PSU and a good one can be quite significant.

Look I hate to be a jackass, but the Quattro is enhance built loaded with teapo caps. Nothing wrong with that enharently, and the unit tested really good, so much so Jonny recomended it, but if it gets hit by something un expected, those little caps will hit the eject button.

Although it's not supposed to be necessary with this supply, I thought of adding in the optional 4-pin connector to the motherboard. This is only supposed to be for SLI setups but then again this board was made before we had cards drawing this kind of power.

Would it hurt anything?

if its a 4 pin molex (which it sounds like it is), thats the pcie power supply, and thats your problem. Plug that guy in and I'd wager a good deal your problems gone. There's no harm in doing so.
 
There is no such thing as overkill for a PSU today. I feel like I'm taxing a 1000 watt PSU with everything Im running. 650wat certainly isnt overkill, 600 - 800 should be the minimum nowadays.

Someone needs to strip you of your [H]ard title. From your sig specs, you are probably using 300-400 watts.
 
Try running Furmark stress test, whenever I run it it loads my video card more than any game I try. It also will show GPU temp during the stress test. Curious if you get the alarm in that also.
 
Unplug your speakers and then try the game again. This will rule out ingame noise or temp alarm.
 
I have GPU-Z. Will that stay on-screen when the card goes into graphic mode?

If you check the "Continue refreshing this screen while GPU-Z is in the background" checkbox it will continue collecting sensor data in the background, and once you get the alarm you can alt-tab back and look at the last few seconds of temperature data.
 
My vote is for power issues. Either the mobo 4-pin that's been described or the Antec simply isn't up to the task. We've been over this a million times in this forum it seems, but just because the (claimed) total is 650W doesn't mean that the current ratings PER 12V RAIL are what you need.

If your PSU is multi-rail, be careful where you spread mobo, CPU and GPU. Check the ratings of the rails to confirm you have what you need for your 4870.

Perhaps try to get a benchmarking utility to duplicate the alarm instead of a racing game, as has already been suggested.
 
I'm thinking it's probably capcitor squeal, or not enough on a single rail of the PSU.

Capacitor squeal can be very loud, I've had two Antecs that did it (one loud, my new one very quiet).

Any of the [H]ard guys around here that can explain what capacitor squeal ACTUALLY is for me / OP?
 
Think real hard here guys... have you ever seen a speaker on a video card? Motherboards do have a speaker but have you ever heard of one doing a low power alarm?

Now that I have you thinking logically, and you realize that there is just one speaker inside the case... what can set it off? oh! the bios! when you go over the max temp (some on by default, usually you have to enable it) or when its issuing a beep code! Some programs can control it too but you would know if you had one running. The video card driver is not included (if it can ive never heard of it!)

My best guess is that the video card is bottlenecked by his cpu and working it really hard.


How to figure the problem out on your own!

Run Furmark and watch your video cards temp. It may very well get hot enough to crash but I bet you that you won't hear any alarm. The screen will just black out and the driver crash.

Now run some orthos small FTT or two instances of prime95 at the same time on small FTT and see if you hear the alarm. (my guess is that this will do it which would point to a cpu overheat alarm).

If that doesn't do it then run both at the same time! that would simulate full CPU load + full GPU load at the same time. More than likely MUCH more abusive than the game. If your PSU were the issue in that game, your computer would probably restart or shutdown if you tried this.
 
It's definitely from inside the case.

I've never had to use the Molex before, and the current config doesn't supposidly warrant it, but I'm going to try it. Results forthcoming...



Unplug your speakers and then try the game again. This will rule out ingame noise or temp alarm.
 
There are a bunch of futuremark products. Which one am I after here?
 
Thats funny, after isntalling my 4870 I started getting noise I thought was the card. Turned out to be my 520W Corsair that "should" have been plenty of PSU for my system. I swapped in a TX650 and problem solved. Try to track down the component that is making the noise and RUN CPUID Hardware monitor before you go into a game so you can come out and see your lowest and highest temps/voltages. Put your ear to the back of your PSU and see if there is any noise from there.
 
There are a bunch of futuremark products. Which one am I after here?

For the record, I get strange noises with 3dmark and Vantage mark that will happen right as it goes into a test. Ive noticed this on many systems that are Nvidia or ATI. I talked to EVGA about this and basically said its normal with todays graphics cards pulling such a large amount of power. I dont hear it in games tho. Furmark has issues with the 4800 series from what I have read. Mine could not run it stable but I can run Vantage mark 8 times in a row @ 865/1100 and game COD4 for hours. Hope you find your problem!
 
Think real hard here guys... have you ever seen a speaker on a video card? Motherboards do have a speaker but have you ever heard of one doing a low power alarm?

Now that I have you thinking logically, and you realize that there is just one speaker inside the case... what can set it off? oh! the bios! when you go over the max temp (some on by default, usually you have to enable it) or when its issuing a beep code! Some programs can control it too but you would know if you had one running. The video card driver is not included (if it can ive never heard of it!)

My best guess is that the video card is bottlenecked by his cpu and working it really hard.

My video card makes a high pitch alarm sound when one of the PCIe plugs isnt plugged in. And yes, ive heard of, and seen video cards that have a low power alarm, not nessecarily a speaker. Please dont butcher the word logic, because logically speaking, speakers arent the only things that can make a high pitch noise inside a computer case.
 
I think this is capacitor squeal. My card does it too at high loads, and I guess you could mistake it for an "alarm".
 
I'm going to throw this out there on the off chance...

Control panel > sound/audio devices > sounds > select 'no sounds' and apply.
 
Thats funny, after isntalling my 4870 I started getting noise I thought was the card. Turned out to be my 520W Corsair that "should" have been plenty of PSU for my system. I swapped in a TX650 and problem solved. Try to track down the component that is making the noise and RUN CPUID Hardware monitor before you go into a game so you can come out and see your lowest and highest temps/voltages. Put your ear to the back of your PSU and see if there is any noise from there.

So, this implies that a PSU going WEEEEEEEEE is a bad sign? On my old rig I've had an Antec SmartPower 2.0 450 (known exploder, I know) thats been squealing for 2 years now on a 3.5 year old rig. It hasn't died yet!
 
So, this implies that a PSU going WEEEEEEEEE is a bad sign? On my old rig I've had an Antec SmartPower 2.0 450 (known exploder, I know) thats been squealing for 2 years now on a 3.5 year old rig. It hasn't died yet!

I dunno.. but when you moved the mouse the pitch would change:D. More then anything it was annoying. I would not want a PSU that made noise and would not expect one to make noise when its within its load rating. Mine could have been defective. The TX650 has no problems.
 
Ok, first off, the Molex connection didn't help. I didn't think it would but it was worth a shot.

Here is a snapshot of GPUZ and Hardware monitor after a few laps with the alarm on:

Other


I find it funny that it only reports a 500Mhz GPU. In fact, in CCC, it also shows a 500MHZ GPU in use and If I try to change it it jumps back to 500 after it saves.

Also, those GPU temps look WAY high to me. Thing is, I can guarantee you there is no lacking for airflow in this machine, and the card is well clear of obstructions.

I also did a short movie so you guys can hear this annoying-ass alarm :)

http://www.youtube.com/watch?v=nEuVm9VfxTY

Notice at the end when I spun and all the smoke was generated, the tone actually went up a level in loudness.

Is it possible there is an underlying compatibility issue between the A8NSLI32-Deluxe and this card, and the card simply isn't being read correctly?
 
Thats funny, after isntalling my 4870 I started getting noise I thought was the card. Turned out to be my 520W Corsair that "should" have been plenty of PSU for my system. I swapped in a TX650 and problem solved. Try to track down the component that is making the noise and RUN CPUID Hardware monitor before you go into a game so you can come out and see your lowest and highest temps/voltages. Put your ear to the back of your PSU and see if there is any noise from there.

I remember that problem, do you think there might be something wrong with your PSU???
 
Back
Top