Please Help! Anti-aliasing causing crashes and freezes in some programs.

C

CherryonioN

n00b
Joined
Aug 16, 2010
Messages
29
SORRY for the long read... but I really need your help guys...

Okay, so I'm banging my head against a wall here. First and foremost - system is...

Asus p6x58d premium (running latest bios [august 2010 i believe])
intel i7 930 ( @ stock 2.8ghz)
12gb corsair dominator (now running at stock 1066mhz and stock timings)
radeon 5970 (with cat 10.10 drivers and stock clocks)

I have some strange issues with my computer that I didn't think I would experience, given the system power I am working with here. In some programs I can run max anti-aliasing just fine, such as left 4 dead.

but then some programs will crash or freeze after a certain amount of time, most notably: farcry2, call of duty black ops, and battlefield bad company 2.

In the case of farcry 2, I began the game running at max aa and af settings @ 1920x 1080. The game ran at a VERY smooth frame rate, but I noticed that it would crash after 25 to 40 minutes of gameplay. So out of curiosity I began lowering my video settings. Naturally I went for AA first, and completely turned it off (in game). This fixed my crashes.

In the case of bad company 2, I have been playing the game since the catalyst 10.6 drivers, and have been experiencing CTDs with that game since the beginning. I ran the game at max settings (1920x1080 8x AA 16xAF) and would get VERY random CTDs. Sometimes I could play for 20 mins, and sometimes I even played for about an hour before ctd. Sometimes it would CTD in 15 minutes, then I would reload, and it would CTD again in an additional 15 minutes.

Last night I got 3 CTDs in a row, and then finally set the in game AA to 1xAA (I guess technically thats off..). After removing AA, the game does not crash at all.

In the case of call of duty black ops, I once again began playing the game at my normal settings, 1920x1080 and max aa and af (in this case 16xAA and 16x AF). The game ran great, however a little choppy in some scenes. But after a random amount of time, the game would just FREEZE, with a rapid looping sound, which required a hard reboot to stop (this happened like 4 times before I began tinkering with settings). Once again, I tried lowering AA. And once again, Removing AA solved my freezes.

So my question is very broad. WTH is wrong with my computer? I strongly feel that I have enough system resources to run these programs at extreme detail settings. Has anyone else experienced this? Maybe I'm giving my computer too much credit, and it REALLY ISNT able to run programs at these settings? Maybe 2gb video ram doesnt cut it? Maybe 12gb ram is TOO much?

please help!
 
my psu is the corsair hx850.

what program can i run to check my temps in game?

as per devices, I have 2 30W fan controlls, 2 dvd drives, and 1 western digital 7200 rpm, and a dual cathode.

again, per MB overheating, I dont think that is an issue, as my case is the 800d and I have a 140mm fan RIGHT next to the northbridge.
 
Humor me and remove the video card and then remove it's heatsink. Note the heatsink is probably far from flat. Note the garbage thermal paste that is probably used.

I only say these things because I had the same problem with two of the newest video cards I received. Cost cutting has hit the industry hard. You better break out the lap kit and lap the heatsink flat and replace the compound with something decent.

It doesn't matter what the temp says in CCC. It's too hot. Both of the cards I have read 70c max and that was still crashing them. So I'm pretty sure that temp is not accurate.

I actually used one for months playing TF2 and many other games flawlessly. The problems came when I tried to run Crysis for MWLL. I did those things and it fixed my problems.

Furmark is the program that I used to determine that the problem was indeed the video card. I would DL that and see if doesn't crash the same as the games you are having problems with. I would bark up that tree if I were you.
 
are you reccomending that i run 3dmark? and see if that crashes... as i have run the free version of 3dmark2006 without any problems... scoring like 24000 marks...

and what do you mean by lapping the heatsink... how do i do this? Seems pretty scary as its a 650 dollar videocard...
 
Did I say 3dmark? I'm pretty sure I said Furmark. Trust me. DL it and turn all of the settings up as high as it will go. It should run like 10FPS and get the card really hot. Let it run for a while until it either reaches max temp and you have no problems or it starts to get way hotter than it should. If it goes past about 80c then I would stop it.

Lapping the heat sink means to use sandpaper and make it perfectly flat. There are many guides on the net on how to do it. Here is a good one.
http://www.overclockersclub.com/guides/lapping/
You don't need any sandpaper past 800grit thought. Going any higher isn't going to do you any good.

So follow the guide but, you only really need 400, 600, and 800grit.

I do realize that it is a $650 video card but, it probably has a 30 cent heatsink on it. I have had to either replace the thermal paste or the heatsink completely for every video card I have owned in the past 10 years. They all overheat, be it Nvidia or AMD.
 
Last edited:
so get this... i just played a lengthy session of bc2 with full Anti-aliasing, AFTER installing my northbridge fan that asus packed with my mobo. And it didnt crash. I'm not saying that it wont crash, and i need to do more tests, but what would cause my north bridge to heat up so much!?!? is it because my ram? its not overclocked... is it because there is 12gb, or because i use all 6 slots??
 
Just ran furmark, and my GPU got to about 87 celcius in 2 and a half minutes...

not sure what i should be looking for here.
 
CherryonioN said:
Just ran furmark, and my GPU got to about 87 celcius in 2 and a half minutes...

not sure what i should be looking for here.
Click to expand...

It didn't crash is surprising, perhaps cooling your chipset better may have fixed it. However 87 degrees is fairly hot and ballerX offers solid advice. I'd do what he suggests and you'd probably shave 15 degrees off your temperature. Keep in mind he noted that the temps are reading lower than they are really. So your 87 degrees is more like 95 which is too hot. Lap your heatsink and apply some good thermal compound, project cost $10 or less along with 20 minutes of your time.

You may be able to then overclock everything, also once this is done re-do the furmark and other stability tests and note the improvement and post back.
 
CherryonioN said:
Just ran furmark, and my GPU got to about 87 celcius in 2 and a half minutes...

not sure what i should be looking for here.
Click to expand...

What happened to the thread I was replying to ?

Anyway, a few things:

1) Furmark does that, so don't look too much into those numbers
2) Did you try reverting to an older set of drivers and see what happens ?
3) If you don't want to try 2) or already did, do you have a spare graphics card, to test it out and see if the problem still remains ? If it does, then it's most likely not a problem with the graphics card. If it doesn't, then you may need to RMA the card, since something is wrong.
 
well the 87 degrees Celsius reading was taken directly from the furmark stress test.

how should i run that test in the future? should I just run it until it crashes? or get to a specific temp and stop the program?

but what about the northbridge?? i just installed the north bridge fan, and I seemed to have a more stable experience when running aa. does anyone know what could be the cause of this?
 
CherryonioN said:
well the 87 degrees Celsius reading was taken directly from the furmark stress test.

how should i run that test in the future? should I just run it until it crashes? or get to a specific temp and stop the program?

but what about the northbridge?? i just installed the north bridge fan, and I seemed to have a more stable experience when running aa. does anyone know what could be the cause of this?
Click to expand...

Some people leave the Furmark running for 24 hours to stress test, I do not recommend that in this case. That is a big risk as those temps are high and as mentioned before may actually be higher than they appear.
 
Silus said:
3) If you don't want to try 2) or already did, do you have a spare graphics card, to test it out and see if the problem still remains ? If it does, then it's most likely not a problem with the graphics card. If it doesn't, then you may need to RMA the card, since something is wrong.
Click to expand...

99.9% of the time I would agree with advice like that, however he's running a Hemlock 5970. trying another graphics card is not going to be enough for him to know if his card is an RMA candidate as few cards would stress his chipset or system like a Hemlock. If that other card is a GTX 470/480 or a GTX 465 then perhaps yes.

If he put in a Geforce FX 5900 and his system didn't crash I dont think that indicates his monster 5970 is bad as there could be other variables involved, PSU load, chipset stability etc...
 
OP, contact the manufacturer if you're still under warranty. Removing the heatsink often voids the warranty.

BallerX said:
I have had to either replace the thermal paste or the heatsink completely for every video card I have owned in the past 10 years. They all overheat, be it Nvidia or AMD.
Click to expand...

If you are running at stock clocks in a well ventilated case you should not have to lap the heatsink or change the paste. I've never done this to any video card I've owned in over 10 years.
 
i just ran furmark for a 20 min test, and it stayed at a pretty steady 87 degrees Celsius. i ran it at 8x msaa and it did not crash, so i dont really know where to go. i can try lapping the heatsink later.

oh and i ran it at 1920x1080 res. but still no crash. so do you guys still think ints my vid card? or should I test for longer.

also, what program can i use to check my temps (cpu, northbridge, and gpu) WHILE playing a game? maybe i can narrow it down if its a NB crash.

also, what do you guys think about the north bridge issue?? i installed the fan, and i didnt crash earlier due to high AA values...

thanks for the help thus far.
 
CherryonioN said:
i just ran furmark for a 20 min test, and it stayed at a pretty steady 87 degrees Celsius. i ran it at 8x msaa and it did not crash, so i dont really know where to go. i can try lapping the heatsink later.

oh and i ran it at 1920x1080 res. but still no crash. so do you guys still think ints my vid card? or should I test for longer.

also, what program can i use to check my temps (cpu, northbridge, and gpu) WHILE playing a game? maybe i can narrow it down if its a NB crash.

also, what do you guys think about the north bridge issue?? i installed the fan, and i didnt crash earlier due to high AA values...

thanks for the help thus far.
Click to expand...

I think you solved your issue, do nothing unless you have another crash. Then you may want to consider lapping your northbridge heatsink and carefully reapplying thermal paste there (a very small amount) so that it doesn't ooze out. Also at that point I'd also lap and reapply thermal compound on the 5970 as many owners of that card have had success after doing this. I've seen this on many forums.
 
so you think that was it? just installing the northbridge fan? Do you guys have any idea what was making my northbridge heat up more when using antialiasing? Because when i use no antialiasing, i suffer no crashing...

is it possibly the fact that i have all 6 dimm slots filled up? does that put additional stress on the north bridge?

i agree that lapping my vid card heatsink would be worth it
 
Okay, so i just ran prime 95 for 1 hour, with no system crash.

my load temps ran up to 73 degrees celsius with 100 percent load on each core.
something strange though... my i7 930 was running at 2.9x ghz... not sure why, i havent overclocked..
and the test was utilizing around 3.2 gb ram.

What do you guys think would happen if i ran furmark AND prime 95 at the same time? is that suicide?

so im now going to run a game and see if i still crash

im going to be very upset if it was just a fan i needed to install...

also here is a pic of what was going on.
vhVeT.jpg
 
You must log in or register to reply here.
Back
Top