I am not kidding. SC2 will fry your v-card...

[F1shF4t]

I'm using the 10.7 release drivers. One thing that can help is by turning up your fan speed to around 75% rather than letting the drivers manage it. Unfortunately, this has been broken for me since the 10.5 drivers .

I thought ati cards did the fan control in vbios?

For my 4870s I set up a custom fan profile which increases the fan speed in small jumps over a short period of time as the temps rise. (No ennoying fan speed up/down noise). Its easy to set up too.

At about 90 - 95C I set it to just go to 100% so the card doesn't overheat. Hasn't ever hit those temps yet, never ran furmark but left sc2 menu on for over an hour before.

 

[Seero]

I left the house while the game is on its menu screen and there was no problem. Stock fan without any programs controlling the speed other than its driver, which is on default values. Temps never exceed 82c. However, whenever it exceeded 85c, I will take the all thing apart and put it back together.

When I play WoW, the temp is usually under 75c, which means SC2 is more demanding in comparison. However, EQ2 is as demanding as SC2 if you mass around its settings.

At times I hope a game kills my card so I get buy a new one, but every time it reaches 85c I will clean it because I don't really have extra $$ for a new card. In fact, the only card that actually fried was an old ATI card that was partly killed by a faulty PSU 10 years ago. After that I have 2 ATI cards and 1 Nvidia card spare that just wont die and are now sitting somewhere in the "computer parts box" and 4 working PCs, the oldest one barely runs window 95, but it just won't die...

 

[AdamK47]

It's all the casuals who don't push their hardware at all that are experiencing problems. The rig in my sig plays the game for several hours without issue. Gets mighty hot in my computer room if the AC hasn't been running for a while though.

 

[Washoe]

It's all the casuals who don't push their hardware at all that are experiencing problems. The rig in my sig plays the game for several hours without issue. Gets mighty hot in my computer room if the AC hasn't been running for a while though.

AdamK47 I'm just curious - how is that LG 32LD450 working for you? Is that your main monitor? (Just checked Amazon... under $450 seems pretty good, no?)

 

[AdamK47]

AdamK47 I'm just curious - how is that LG 32LD450 working for you? Is that your main monitor? (Just checked Amazon... under $450 seems pretty good, no?)

It's fantastic for use as a monitor. I also use it as a main monitor. I've tried a few Samsung and a ViewSonic 27" TN panel. Returned them before settling on this one (I'm very picky). It has an S-IPS panel, low pixel response time, low input lag, great non-dynamic color contrast, and great viewing angles.

 

[kinobas]

just a bump to make sure people see it at 1st page

 

[Modelworks]

Any software destroying a system from heat is a hardware cooling design problem not software. I routinely run setups with 4 quadro cards in a cage running cuda renderers 24/7 and they produce way more heat than SC2 ever could. These cages are about 1 inch larger than the cards on all sides. So not much space for cooling. I have yet to see a card fail. If SC2 is making your system crash or fail you really need to look at the way the air flow is done.

 

[Seero]

Any software destroying a system from heat is a hardware cooling design problem not software. I routinely run setups with 4 quadro cards in a cage running cuda renderers 24/7 and they produce way more heat than SC2 ever could. These cages are about 1 inch larger than the cards on all sides. So not much space for cooling. I have yet to see a card fail. If SC2 is making your system crash or fail you really need to look at the way the air flow is done.

The thread about this in Blizzard tech support forum
Unfortunately, it was closed, but you will be able to find joy out of it.

 

[llee]

If the game is going to fry your v-card, you should probably get a chastity belt.

 

[Modelworks]

The thread about this in Blizzard tech support forum
Unfortunately, it was closed, but you will be able to find joy out of it.

Engineers design a product they design it knowing how much heat it can generate and they design the cooling to fit that. There is no way that any software can create more heat than what the product is designed to produce. If the part is 85W and the cooling is designed for 85W then nothing software will change that.

What people are seeing here is hardware related. Either the card maker designed the product wrong or the end user isn't using it correctly.

People stress cpu with programs every day and they do not fail as long as they are cooled according to the spec.


People often see the need to replace thermal compound because they think it is dried out. As long as it was liquid when it was applied that is all that matters. It filled the void and whatever else was in it was just used to make the real conductor of heat easy to apply. Doesn't matter if it has dried on there for 50 years as long as it dried in place. I have worked on radios from the 1960's that still use the original compound and work fine.

 

[konakona]

what modelworks said.
if my understanding is correct, 100&#37; load on GPU is 100% load regardless of what kind of fps you get out of it. my gut feeling is sc2 just happens to be a bit more mainstream than games like crysis, so people that happen to have inadequate cooling to begin with got to put their hardware to real stress test for the first time ever. I bet furmark would have had a similar effect on their systems, only faster. Am I neglecting something here? <- more of question really, I admit I am no authority when it comes to highly technical detail of gpu hardware o_o

 

[Seero]

what modelworks said.
if my understanding is correct, 100&#37; load on GPU is 100&#37; load regardless of what kind of fps you get out of it. my gut feeling is sc2 just happens to be a bit more mainstream than games like crysis, so people that happen to have inadequate cooling to begin with got to put their hardware to real stress test for the first time ever. I bet furmark would have had a similar effect on their systems, only faster. Am I neglecting something here? <- more of question really, I admit I am no authority when it comes to highly technical detail of gpu hardware o_o

You have the authority. This is not a lecture room, it is a chat room. As long as no personal attacks, then this very thread will have value.

The number that you see which indicate the load of CPU/GPU is inaccurate. Video cards use shader cores (multi-cores) to compute, there are no sensors or algorithm that determine how much of them are being used. What people do is to see if the video card is available or not at a fix interval. Based on the availability of the interval it approximate its percentage of usage. We it shows 73%, it doesn't mean 73% of the shader cores are use, but 73% of the time the GPU is not available.

Just because it is unavailable doesn't mean it is at full load, it simply means it is unavailable. Let say 2 programs both run GPU at 100%, but one can generate much more heat than other because it utilize more shader cores. If you have used prime95, then you will see there are actually different tests. Look at the temperature of the CPU and you will see, although all tests puts the usage of CPU to 100%, some actually burns the CPU far more than others.

 

[konakona]

ok, that makes sense and the cpu example is convincing. I kinda wish someone did a very technical, in-depth analysis of how and exactly what happens in those burnout accidents.

 

[evolucion8]

That's right, specially with AMD's superscalar architecture, there will be scenarios where the TMU's are taxed more, or scenarios that the ALU's will be taxed more, and both scenarios might reach 99%. Crysis Warhead taxes the GPU to 99%, and yet it doesn't heat the GPU that much compared to Furmark and even SC2. (I haven't played SC2)

 

[Seero]

ok, that makes sense and the cpu example is convincing. I kinda wish someone did a very technical, in-depth analysis of how and exactly what happens in those burnout accidents.

When video card cease to work, laymen will simply say "it fried." Well, if something fried, there will be a smoke smell and smoke itself. With the video card HS, I can place a lighter under a GPU that is connected to its HS (wihtout the board theoretically) and nothing will melt for hours. If there is a fan blowing through the HS, then the lighter will run out of gas before it melts. That means, under normal cases, if the vent is blocked or not working properly, the plastic housing of the video card will start to melt before the actual die gets fried.

Having start that, what is likely to happen is some of those aged capacitors popped, or aged soldier connection due to bad air-flow or bad PSU.