Question Hot ATI 4850

groo

Distinguished
Feb 3, 2008
1,046
6
19,295
I've been running OBG for awhile with an inherited i5-4400 machine. (yes, I am not the gamer I used to be)
recently I decided I wanted to play WarThunder and the OBG were lacking in a big way.
I dug out the old ATI 4850 I had laying around and an Earth Watts 430 PSU for the 6 pin, and WT was definitely playable, until the machine shut off. I suspect is is shutting down to protect the 4850 from heat damage (or at least thinks it is).
It shuts off with no warning other than that that the fan definitely starts screaming during the loading screen. It goes several minutes before shutting down.
I don't get graphical glitches or the like, as I have in the past when OCing the card. I turned down both the CPU and memory speeds to the minimum with no noticeable change. I tried testing the new settings and they always fail, whatever that means.
It idles at an indicated 90s C range, GPU and memory, but I'm not sure I trust those temps as getting drivers for the 4850 are about impossible, so I might have a driver giving bad temp reading I thought.

The machine will idle for hours without a problem.

Do I pick up some thermal paste and give new paste a shot? I suspect the card is a bit over a decade old (I first put it on an Abit mobo come to think of it). What I don't get is if it was that, wouldn't the memory be at a different temperature than the GPU by more than a degree or two? Would I get visual indicators of a heat problem when in game?

I know the real answer is to pick up a new card, but with the wife not working thanks to the kids school being very questionable along with potential day-care and needing a new vehicle myself, I'd rather not spend the cash on it.

ps. I was playing World of Tanks with the OBG, before trying to step up to War Thunder, and that shut down the system as well.

pps, temps are at 93 - 94 C after typing this up.
 
Is the fan working on the 4850?
The shutting down part might be the PSU as well. It only gives you 207W over the +12v bus. It is far too low.
Try running a burn test with OCCT and loading just the video card. If it passes the test, try adding CPU+GPU load (that tests the PSU for stability)...
 

groo

Distinguished
Feb 3, 2008
1,046
6
19,295
Yes the fan is working. It was driving me insane looking for drivers because it was going full blast 100% of the time. The system is definately noisier now, and the fand does ramp up when loading the game as I mentioned. the temp monitoring also shows a fan speed feedback. idling it runs almost 60%

My first though about the shutdown was that it was the PSU, until I saw the reported temperatures before I game. This is the PSU I used with this card for years. I'm not saying I know for sure it isn't the PSU, but if it is, I would have to buy something anyways as this is my only PSU with a video card power plug. If I had to buy something, I'd just buy a 1650 video card or the like without a 6 pin power plug.

Maybe it is the thermal compound. After letting it sit overnight, is was much closer to room temperature at start-up.
 

groo

Distinguished
Feb 3, 2008
1,046
6
19,295
Update
I tried toothpaste and vasaline for a thermal compound, and it was a definite improvement. it was still getting quite hot, but didn't shut down.

i'm not sure where that leaves me though.

I'm thinking WT is pushing the antique card to the bleeding edge, and may not last even if I do get some actual thermal paste.
 
Last edited:

groo

Distinguished
Feb 3, 2008
1,046
6
19,295
I think its the graphics card, it just cant handle the games anymore.
The gtx 1650 super would be a decent card on a budget (similar to a gtx 1060), and would fit your system.
probably going that route, but need to make sure my mobo has enough PCIe wattage to handle a 1650 first.

eta: seeing nothing to suggest the TC-605 produces less than adequate pcie wattage. I've also had no problems playing maybe an hr straight suggesting that my 6 pin PSU is fine.

peak temp I am seeing on the monitor is 111C/231F. I can't see this graphic card lasting long, even with real thermal paste. It dropped down to the 80s pretty quick after stopping the game, so I think the cooler is working, just that a natorious hot card is being over-worked is what I think is going on.
 
Last edited:

groo

Distinguished
Feb 3, 2008
1,046
6
19,295
It's still chugging along, as hot as ever. I did do the chrysalys (or whatever AMD software is called) testing. It failed at the default OC levels, but passes at min settings, so I'm leaving it there.
Plan is to replace it with a 1650super. It was going the be the vanilla, but I'm not 100% sure the mobo pcie slot is up to it. I know the psu is up to it now.
I was going to just buy one, but Mom as always wants me to tell her something to get for my birthday (way too old for that, but whatever), so I'll wait a month for it. Let's see if the Colgate card can hold out that long. Worst case, back to OBG for a bit.