[Help]Video Card Temp = 63C/140F !!!

Some of you may know I've been having big problems in TF2 lately with choppiness and artifacts.
I installed Speedfan and found that my video card temp is high. 63C. Supposedly that's the core temp. Ambient isn't coming in.

CPU Cores are at 24C.
HD is at 41C.
ACPI(whatever that is) is 43C.

So, it seems my video card is way too hot.
The wife's laptop just ate it 2 weeks ago. Video card went kersplat. She still doesn't have an acceptable replacement, but at least she can boot up and see her monitor now.

So, what do I do? How do I make things cooler in this laptop? It seems to be the video card that's worst off. I think the other stuff isn't *too* hot, just the Video.

Inspiron 9400(the 17" laptop) with nVidia 7900GS if that helps.

63 is about normal. When you get close to 80, then worry.

That's not hot. That's not even warm. Some Google searching suggests that the chip's throttle temperature is 101-102 Celsius.

Here, we see people overclocking the 7900 GS in your very laptop (as well as the E1705), running in a normal operating range of anywhere from 60-80 Celsius just fine.

Anything below 80 is completely normal. Now, the GPU itself may be dying, which could lead to your artifacts, but it's not because it's running "too hot".

That said, if you're talking about 60 Celsius as a purely idle temperature, then maybe it's hitting some high temps at load. If so, first thing to do is crack that case open and blow all of the dust out of the fan vents and heat sinks. I did that with my Inspiron 6000 when my CPU temperature was getting too high, and it plummeted over 20 degrees Celsius after that.

First response: So?

Second thought: Legion already went farther than I would have ^^^

The artifacts could be bad ram, or a previously damaged/failing GPU. I've had quite a run of dying 7900 GT desktop cards lately. Things like covering an active video overlay with green equals signs, or another had a flickering hot pink line.

Strangely, I don't generally see the artifacts in HL2. I haven't fired up portal in forever, but I'm only seeing choppiness and artifacts in TF2.

I guess I'll see if I can find a good HowTo on opening this damn thing up.

duckilama wrote:

Strangely, I don't generally see the artifacts in HL2. I haven't fired up portal in forever, but I'm only seeing choppiness and artifacts in TF2.

I guess I'll see if I can find a good HowTo on opening this damn thing up.

Our video errors were in edit bays, and generally only showed up in the active window of our NLE. When you switched windows the problem moved, but only in the NLE, not in any other programs.

Nothing short of changing video cards fixed it.

63 is pretty good, I would say too good to be the upper limit of what the card hits under load, especially in a laptop. What was the machine doing when you got that reading?

Good cleaning never hurt nobody btw....here is a link to the service manual.

I'll echo the rest. I have a 7900GS in my laptop and it hits the high 80's under load. No artifacting or heat related issues that I've ever seen.

And actually I think we even have the same laptop, maybe a year seperated? I have the E1705, which is just a 9400 with a new name if I remember correctly.

My video card (8800GTX-KO) hovers anywhere between 59C and 80C (if I forget to set it to manual and turn up the fan after startup because nvidia sucks.)

You can adjust your fan settings in the Bios.

Dude, when I was playing TF2 on my thinkpad the 7900 in it went up to 90.

Im with the other sowhats. My card sits just under 60 24/7 and I haven't turned my machine off for more than a reboot in 6 months or so.

pol wrote:

63 is pretty good, I would say too good to be the upper limit of what the card hits under load, especially in a laptop. What was the machine doing when you got that reading?

Absolutely nothing.
I'll try to remember to tab out of TF2 next time I see an artifact and get a reading.

duckilama wrote:

I'll try to remember to tab out of TF2 next time I see an artifact and get a reading.

Use RivaTuner to get GPU temperature readings while in-game. Here's a guide, check out the Hardware Monitoring setup part.

RivaTuner is a bit involved to set up, but if you own an NVIDIA card, it's extremely useful.

rabbit wrote:

Im with the other sowhats. My card sits just under 60 24/7 and I haven't turned my machine off for more than a reboot in 6 months or so.

You are a crazy person rabbit. My power bill will not allow me to keep mine running in such a manner. Do you read by light of a 300W halogen flood-lamp?

KingGorilla wrote:
rabbit wrote:

Im with the other sowhats. My card sits just under 60 24/7 and I haven't turned my machine off for more than a reboot in 6 months or so.

You are a crazy person rabbit. My power bill will not allow me to keep mine running in such a manner. Do you read by light of a 300W halogen flood-lamp?

I run my lappy and my 750w PSU for my desktop continuously basically. Then again, I'm in college and have an all bills paid room.

Ok, so it seems to be limited to TF2, my performance and artifact issues mentioned above.

So while starting up net_graph 3 in the console tonight, I saw a bunch of red lines in the log.

Current Map is cp_junction_v2 NextMap is cp_desertfortress ConVarRef hud_quickinfo doesn't point to an existing ConVar
ConVarRef topcolor doesn't point to an existing ConVar
ConVarRef bottomcolor doesn't point to an existing ConVar
ConVarRef cl_himodels doesn't point to an existing ConVar
ConVarRef cl_playermodel doesn't point to an existing ConVar
ConVarRef cl_crosshairusealpha doesn't point to an existing ConVar
ConVarRef cl_crosshair_red doesn't point to an existing ConVar
ConVarRef cl_crosshair_green doesn't point to an existing ConVar
ConVarRef cl_crosshair_blue doesn't point to an existing ConVar
ConVarRef cl_crosshair_scale doesn't point to an existing ConVar
ConVarRef cl_crosshair_file doesn't point to an existing ConVar
--- Missing Vgui material vgui/crosshair_bg
ConVarRef doesn't point to an existing ConVar
ConVarRef cl_crosshairscale doesn't point to an existing ConVar
ConVarRef hud_quickinfo doesn't point to an existing ConVar
[GWJ] KeyboardNotPresent killed [GWJ]-WiredAsylum with tf_projectile_pipe.
[GWJ] KeyboardNotPresent killed [GWJ]-WiredAsylum with tf_projectile_pipe_remote.
[GWJ]-Tempest Blayze killed turbocopland with flamethrower.
tboon killed [GWJ]-WiredAsylum with flamethrower.
Nihilo killed tboon with minigun.
[GWJ]-MyBrainHz suicided.
[GWJ]-Rebel Blob connected
Faceless Joe killed [GWJ]-Serengeti with knife.
ConVarRef hud_quickinfo doesn't point to an existing ConVar
ConVarRef cl_himodels doesn't point to an existing ConVar
ConVarRef cl_crosshairusealpha doesn't point to an existing ConVar
exec: couldn't exec userconfig.cfg
Host_WriteConfiguration: Wrote cfg/config.cfg
Unknown command: cl_crosshairscale
ConVarRef hud_quickinfo doesn't point to an existing ConVar
ConVarRef cl_himodels doesn't point to an existing ConVar
ConVarRef cl_crosshairusealpha doesn't point to an existing ConVar
exec: couldn't exec userconfig.cfg
Host_WriteConfiguration: Wrote cfg/config.cfg

I had changed some video settings, going down to low/off on everything. I've tried changing from 1680x1050 to 1440x900, which doesn't make any difference.
At this point, I reloaded TF2.
More red in the log.

Both ConVars must be marked FCVAR_REPLICATED for linkage to work (r_VehicleViewDampen)
maxplayers set to 24
Steam config directory: c:\program files\steam\steamapps\duckilama\team fortress 2\platform\config
Both ConVars must be marked FCVAR_REPLICATED for linkage to work (r_VehicleViewDampen)
Unknown command "sv_backspeed"
Network: IP 192.168.0.100, mode MP, dedicated No, ports 27015 SV / 27005 CL
Connecting to 72.5.250.142:27020...
Connected to 72.5.250.142:27020

Team Fortress
Map: cp_desertfortress
Players: 11 / 32
Build: 3420
Server Number: 48

ConVarRef bm_nextmap doesn't point to an existing ConVar
SetConVar: No such cvar ( bm_nextmap set to ctf_turbine_v3), skipping
No pure server whitelist. sv_pure = 0
Failed, using default cubemap 'engine/defaultcubemap'
SOLID_VPHYSICS static prop with no vphysics model! (models/props_spytech/radio_tower001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_2fort/silo_cluster01a.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_2fort/silo_cluster01a.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace006.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace005.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_spytech/radio_tower001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/blast_furnace001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/window_skylight001.mdl)
SOLID_VPHYSICS static prop with no vphysics model! (models/props_trainyard/window_skylight001.mdl)
duckilama [GWJ] connected
Redownloading all lightmaps

Ok, no better. Still 12-16 FPS. Disconnected and went to an empty Control Point server running Granary - logic being, maybe it's the custom maps.
More red.

Connecting to 63.209.33.64:27015...
Connected to 63.209.33.64:27015

Team Fortress
Map: cp_granary
Players: 1 / 24
Build: 3434
Server Number: 89

ConVarRef bm_nextmap doesn't point to an existing ConVar
SetConVar: No such cvar ( bm_nextmap set to cp_well), skipping
Got pure server whitelist: sv_pure = 1.
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bmod
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bm07_lg
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bm07_med
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bm07
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bm_red
CMaterial::PrecacheVars: error loading vmt file for decals/custom/bm_blu
duckilama [GWJ] connected
Redownloading all lightmaps

And still bad low FPS.
Ugh.
If I "Delete Local Content", how hard/easy is it to redownload it from Steam?
Could that help?

Very frustrated that the one game I really like playing right now is being pissy.

Edit: HL2 works fine at autodetected settings, some high, some med, some low, either 1920x1200 or 1680x1050, 35+FPS. Not screaming, but buttery smooth nonetheless. Help me Obi-wan.

Its really easy to re-download.
I would suggest that, and driver updates at this point.

boogle wrote:

Its really easy to re-download.
I would suggest that, and driver updates at this point.

Driver updates? I'm running the latest nVidia drivers. And not seeing ***any*** issues anywhere but TF2. Still update/reinstall drivers?

Looking at some of the support results from Valve, I too would vote reinstall. It's easy, just takes a while.

You might try uninstalling the drivers and running something like Driver Cleaner. We've had a couple of folks lately with problems like yours that cleaning out old drivers took care of.

Any chance you have some examples (screenshots) of the artifacts?

Re-install sounds like a great idea, but looking at the output you pasted I'm wondering if our definition of "artifact" should be in question here. Wild ass guess, but I think results of the issues listed would look more like missing textures.

Artifact - some wacky graphical "thing" that shouldn't be there.
For example, sometimes there will be infinitely long "poles" sticking out of gibs, like a head, essentially like the bone/vector/whatever of the model suddenly becomes visible.
Or coming out of Granary CP doors between 2 and 3, when I get 3 steps out of the door, suddenly there are planes(think geometry/3d work) from various objects that make it so I can't see jack until I get across the bridge, then they are behind me and I guess they get culled correctly or whatever.

I'll get some screenies next time it happens, but lately it's just really bad lag/fps/whatever with the occasional artifact.

Hmmm... 167.51 is the latest driver they have for Go/M cards.
169.21 is the latest for other 7 series cards.

I wonder what would happen if I installed the non-Go/M drivers?

duckilama wrote:

Hmmm... 167.51 is the latest driver they have for Go/M cards.
169.21 is the latest for other 7 series cards.

I wonder what would happen if I installed the non-Go/M drivers?

A tear in the fabric of reality, which will expand and consume the universe over the course of the next hours.

Oddly, the older version has been updated more recently.

If I remember, it just won't take that driver. Bitches about not having appropriate hardware or something like that.

Ah, ok.
Don't cross the streams.

Ok, I watched the tech replace the wife's video card today and then took my laptop apart and blew everything out, even taking out the video card and the fan for its heatsink and blowing them out.
We'll see if it helps.

I wonder if upgrading from 2gigs of system RAM to 4 would help performance? WinXPPro, Core2Duo T5600 @1.83GHz, 7900 GS Go.
I think DDR2 SoDIMMS are pretty cheap these days, aren't they?

Doh, Kingston says my system can only handle 2 gigs. Damn.

Here's an example of an artifact I can play around. I'll have to find a Well game to repro the really bad stuff.
***I*** painted the red there, to highlight the pole/normal/vector/bone coming from a gib.
This particular one is coming from the helmet(soldier) and as the helmet was rolling on the ground, it was moving, so it is clearly part of the (supposed to be) invisible stuff underlying the models. It's not so bad, it's when I start getting underlying stuff from the map that kills it.

IMAGE(http://img187.imageshack.us/img187/9416/artifactctf2fort0001ux1.jpg)

I wont say that isn't an artifact (fits the definition afaik), but the ones Ive gotten from hot cards are no where near that....tame.

Did you get a chance to re-install yet?

My 3850 AGP is 45C idle, but I am sure it climbes to the 60s during 3D operation. In general I found that the best defense against an overheating peripheral lies in vacuuming the dust out of their fan and heatsinks. A small car vacuumer can provide very powerful suction and it works well for me.

shihonage wrote:

My 3850 AGP is 45C idle, but I am sure it climbes to the 60s during 3D operation. In general I found that the best defense against an overheating peripheral lies in vacuuming the dust out of their fan and heatsinks. A small car vacuumer can provide very powerful suction and it works well for me.

I'm more of a blow guy myself. Vacuums can generate significant static charges. (which is why you use a little car one, I know)

pol wrote:

I wont say that isn't an artifact (fits the definition afaik), but the ones Ive gotten from hot cards are no where near that....tame.

Did you get a chance to re-install yet?

Yeah, and I did say that was one of the most mild things i see. It was what I could generate alone on Stan's this afternoon after blowing out the whole system. The only things I didn't take apart were 1)CPU + Fan (I still blew out the fan/HS/CPU area) and 2)the LCD itself.
I did reinstall TF2.

LilCodger wrote:
shihonage wrote:

My 3850 AGP is 45C idle, but I am sure it climbes to the 60s during 3D operation. In general I found that the best defense against an overheating peripheral lies in vacuuming the dust out of their fan and heatsinks. A small car vacuumer can provide very powerful suction and it works well for me.

I'm more of a blow guy myself. Vacuums can generate significant static charges. (which is why you use a little car one, I know)

See above. Already done, in great detail.