How hot is too hot?

So I took eltee's advice and did a "touch it and see" test on my heat sink on my video card.

And uh, it's HOT.

I can't put my finger on it for more than a few seconds before I have to pull away.

Is this normal? And, what's the best way to fix it?

LF

Lordfly, no matter what anyone tells you, DO NOT TOUCH RUNNING EQUIPMENT.
If you had touched your CPU instead, you would now be wearing a bandage around a very burned finger.
Furthermore, seemingly innocent electronic equipment like a monitor, can accumulate and retain enough juice to KILL you in a second, even if its not currently turned on or even plugged in.
If you want to measure temperature, there's them newfangled techno gadgets out there called thermometers. You should give them a try.

What kind of video card is this anyway?

That seems a little extreme. All modern consumer equipment has bleeder resistors on the capacitors to avoid that problem. If the monitor is unplugged at the time you start unscrewing the case you aren't likely to kill yourself. That said, it's true people shouldn't screw around with electronic equipment if they don't understand the possible dangers.

You never know what them blasted third world fabs take out to keep costs down. Normal people dont need to open monitors anyway

Thanks Mr. Wizard

I did in fact turn off the computer (via the power button), turned off the power supply, unplugged it, and then held onto the computer case to ground myself before I touched the heat sink.

So, now answer my question

Oh get off it, man! I touch my monitors all the time and it never hurts me...look, I'll even lick the glass surface to prove to you that nothing will hap.....*bonk*



<At this point the intelligible portion of Lumiere Noir's message ends abruptly. I have removed twenty-seven lines of continuous "BUZZZZZZZZZZ..." for pity's sake. --Forum Moderator>

The human body's optimal operating temperature is 37 degrees C. For an idea of how much heat the body can safely withstand, the US government's Consumer Product Safety Commission recommends a maximum temperature of 40 degrees C in hot tubs.

By comparison, Motherboard Monitor says my CPU is currently operating at 50C. (Which is a little hot, but my room's hot in the late afternoon.) My video card doesn't have a thermal diode, but I'm told that it's probably within a few degrees of that.

So yes, you should find it hot to the touch; even hot enough to burn your skin. That's normal. Any time you want to poke body parts into active or recently active machinery or electronics, you should expect to come into contact with temperatures and voltages that exceed the human body's ideal exposure.

Basically, the test for video card overheating is this: Are you getting visual artifacts and crashing? Incidentally, this is also the standard "do you own an ATI card" test.

I'm getting artifacts only in Second Life. I ran SL for 15 minutes, then turned the computer off and touched the heat sink. Very hot to the touch.

Then I let it cool off, then played a less graphical game (Victorian Empire, a 2d strategy game), then did the same thing. It was much less hot, I'd even go so far as to say lukewarm.

I guess I'm just trying to gauge how hot is too hot... it worries me a bit that SL makes my video card hot enough to make it puke triangles on the screen... after all, if mine is doing it, what about Joe. Q. Newbie's card?

I guess I'll go find another fan of some sort.... dunno the best way to position it though

LF

Well, SL isn't a very video-intensive app, not when compared to a video-bound game like Quake or Unreal. Try running one of them. Do you still get video problems? If not, then it's probably not your video card.

What kind of video card and fan do you have?

I would wager that SL does as much or more work with video memory - weather it utilizes all the advanced features of a video card or not - than any quake or FPS game out there. So while it may not be pushing the envelopes on rendered pixels per second, I would bet it gives your cards video memory - and thus the rest of your card - a work out like it is definatly not used to seeing and won't see in Quake or Unreal.

TADA! http://www.thinkgeek.com/gadgets/electronic/5d95/

That's true; SL uses as much video memory as it can get its hands on, but it doesn't work your 3D hardware as much as something like Half-Life 2 would.

How do I know this? Well, if I try to run Quake, Halo, Unreal, Lineage 2... they all get video artifacts and crash in the late afternoon. SL doesn't.

Also, the Lindens have said as much.

Basically, it's not the RAM that heats up -- Quake or Halo aren't going to have anywhere nearas many textures as SL, but they are going to put a lot more strain on the GPU itself.

Ahh, but that only measures surface temperature, not core temperature, Darwin. (Yes, yes, I know. So does your finger. I'm just saying. )

Quake3 and Unreal run flawlessly. Like I said, I get absolutely no qualms from any other game I have on my system (20+), most of which are graphics-intensive.

I have a Radeon 9000... only one fan in the case, for the pentium 4. The radeon has a large heatsink on it.

These artifacts happen on any version of drivers that I can run (3.5 on up). For some reason, previous versions of ATI drivers make SL not recognize my video card prior to driver version 3.5.

LF

If the heatsink is too hot to touch, and you're getting artifacting, the card is probably overheating. See if you can duct a fan to blow directly on the heatsink.

Extra cooling certainly couldn't hurt, but it's most likely just SL and ATI's drivers not playing nice. There may be very little you can do.

Excess heat produces visual artifacting. Not even God's own ATI drivers could prevent that.

When my last card started crapping out, I got a single black pixel at the corner of every texture, so in big flat places I got a grid of dots, and in places where the texture was horribly stretched, I got a big black ball.
I had no idea that it was a sign of the card crapping out and so I failed to replace it within the warranty period.
The memory on it started failing and so My card dropped from 64 to 52 megs a couple of megs at a time.
So yeah, if you see visual artifacts, or if games start crashing on ya, or if some game tells you that your memory aint quite as big as it should...
IMHO the best way to find out if your card is getting enough cooling would be to get a thermometer and read the card's specs in the manual or the company's website.

For nVidia, it's desktop -> properties -> settings -> advanced -> (cardname) -> Temperature Settings. It will tell you the current temperature and the threshold temperature. ATI might have something similar.

My FX5900 runs around 60C. 60C is 140F, which is too hot to touch for more than a couple seconds, but still very far below the danger zone - which is listed as 140C, or 284F.

Just my tuppence here, but on a system where the bottleneck's the GPU (as in LF's case - a high-end P4 should be cranking out data at a rate of knots) the card could well be getting overworked, especially a Radeon 9000 which I'm given to understand isn't the newest card in the world...

Hot weather and limited passive cooling isn't going to help your GPU's cause much - it may seem like overkill but for a purely passive solution something like this may prove a worthwhile investment, as the design is such that it'll fit a wide variety of cards. There's also a quiet bolt-on fan if you upgrade to something more heat-producing.

More redundant advice:

If you're having heat problems, make sure that you've got a good amount of air flowing through that case. It's more than just having a fan blow on the card. For instance, have a pusher fan on the front of the case, then add to the PSU exhaust with either a case fan mounted in the back or a slot fan in the IDE slot space directly below your video card. Make sure the airflow is unobstructed, replacing ribbon cables with round cables and such. Given the right PSU/fans, this can even be a quiet solution.

No man knows when the time has come for his video card to pass on, but all hope that it happens within the warranty period.

as someone who grew up spending summers in phoenix.. i will say 50c isn't exactly going to kill you.. and in short burts can actually be sort of 'refreshing' (thats exactly 120f) in a sort of sauna kinda way... in fact most saunas are in the 50-60c temperature range hardly scalding... 75c is where i personally find electornics very hard to touch, they're too hot.. and by 80-85c 'scalding' would be an appropriate term, but again i wouldn't 'HOLD' them for a long time jus tap fingers on them...


as for touching electrically inert components in a pc (such as heatsinks) as long as yer hands aren't sweating, and you are in direct physical contact (with other hand or grounding strip) to the computers' case at all times, there is essentially nothing to worry about.



As to heat SL is actually far *FAR *FAR* FARFARFARFAR* more likely to induce heat in video cards than any other 3d app i've ever seen (save one or two i've seen and/or made myself for that explicit purpose)

a typical 'window' of second life is rendering something like 85,000 polygons or so... in the city or more crowded areas that can easily climb up over 150k... entire *LEVELS* in quake3 had less than that, and actively occluded and didn't render the vast majority of them at any one time ontop of that... ontop of that many of those polygons are unduly-expensive... transparency being the biggest contributor of that requiring an entire second rendering pass for however many transparent polygons are included in a given scene (often 20k or up)


as to whether you need a fan lordfly... essentially *ALL* fanless video cards assume a) they're running far less complicated games than SL, which is as i have shown, a fair assumption.. and b) assume two fans, an exhaust fan, usually part of the power supply, and an intake fane, speeding fresh air over the internal components.


My 9800 pro was easily topping 80c in SL originally and had no end of artifacts (disco electro flashy world) and crashes... it has behaved *MUCH* better since getting a MONSTEROUS 1 pound heatsink from zalman with large 80mm fan blowing directly on it... and replacing the cheapy builtin hsf.



*SERIOUSLY* at this point ATI's drivers, while not perfect, are *EASILY* the equal of nvidia's in stability and performance in both openGL and direct3D.

This is coming from someone who *IS* an openGL developer, so yes I am paid to actually test alot of this stuff. SL is abit of an extreme case and had opened up some cracks in the floorboards that other games had breezed over... but at this point any disparity left would have to be attributed more to the newview.exe code.

(actually *ALL* ati fixes other than the reflection mapping that 4.5 catalysts have patched up have been fixed by LL, within second life's rendering code)

--edit:

im not saying that the people having trouble aren't having trouble... simply that computers are insanely complex machines, electrically, thermally, and software wise its even worse... simply tossing any and all complaints at ati's doorstep is futile and self-defeating.

If you do have a problem with SL there is most likely a very specific reason for it, and its probably either the hardware, or the secondary support drivers/software such as xp issues, bad agp enablers for the motherboard chipset etc.... from talking with people i'd say the following generalizations seem to be about par of the course

60% of the problems seem to be thermal in nature
15% or so seem to be related to power, or lack thereof
15% or so stem from any number of BILLIONS of possible FUBAR issues with xp and secondary drivers/malware
10% seem genuine as far as some latent lasting issues inside of SL or ati's actual driver stack.. pulling the two apart without *ALOT* of study and consideration is a fool's recourse



and as to agp issues the reason its off is cause Linden Labs specifically TURNED it off whenever they detect an ati card..

not that it *EVER* helped me when enabled in the many many hours of performance testing i've done while it was still togglable, or held any tangible stability penalty either... it is essentially a useless setting... don't fret over it

As I was taught finger temperature testing...

You put your finger on the heatsink (or chip if you can), and if you can hold it for a count of 5, it's okay. If you can't hold it that long, it's probably too hot.

The zalman zm-80 (Or whatever that huge passive video card heatsink is called now) Is not enough to cool a radeon 9800 or modern geforce if you leave it passive. You MUST add the optional fan (or a fan of some sort), at which point it's probably one of the best video coolers on the market.

Anyone that tells you you'll need to get bandaged after touching a heatsink or chip is working on some very not-consumer grade computers, or thinks you're really stupid. If you can tell it's too hot to touch without touching it... you probably shouldn't touch it. (Maybe check to see if that pot of boiling water is hot enough to put your eggs in by sticking your hand in it next?)

Like everyone's said, texture artifcats are usually because of overheating. They can also be caused by bad drivers and bad programming. So if your video cards drivers are up to date, then update your motherboard's drivers, if available. Then if that hasn't solved the problem, update the bios, if available. Next, update anything you can think of! Then throw your computer into a lake and become a hermit.


You're going to have do quite a lot of idiot work to kill yourself with a computer case. On the otherhand, if you're not grounded, you probably can kill the computer pretty easily.

Monitors are definately deadly though, don't open those unless you've had quite a lot of training.

Full agreement, with practical experience. Second Life is a graphics card workout.

Playing Second Life causes fans in my graphics card and PSU to blast into high speed. I've never heard these fans on 3D games like FarCry and Unreal Tournament 2004.