Could I have damaged a CPU?

Question

I built a PC around a Q9550 1 1/2 years ago (motherboard is an Asus P5Q deluxe). Specs give 2.83GHz, I overclocked it to 3.40 GHz without any problem (or so I thought) until two months ago. Cooling is provided by a stock Intel fan.

Two months ago, I began to see random crashes, the BIOS saying the CPU had overheated. The PC would reboot at the overclocked speed without any problem. Since last Saturday and a few more crashes, the PC won't reboot at 3.40 GHz, and even at stock speed (2.83 GHz), I got core temperatures of (60 C idle, 95 C under load) on the first two cores. This is the 4 core temperatures I am talking about, not the T-CPU which obviously is lower. The fan is running at a steady 2000 RPM.

Questions:

1. Is 2000 RPM the normal speed of the Intel fan or is my fan somehow broken (which could explain the overheating). In this case, any recommendation for a good fan for overclocking?

2. The hypothesis I fear is the right one: could the CPU have been slowly damaged over time by this overclocking, meaning there is nothing much to do except waiting for it to die? (As a side note, I am surprised that the 9550 is still around 300 $CDN here... Thought it would have been cheaper with all those i3/i5/i7 around).

Any help or advice would be more than welcome.

Answer 1

First, what TIM (thermal paste) are you using? Have you checked it lately? If it is old and dried out, you need to clean it out and refresh it. If it is the stock interface material, you should never have been using it in the first place. Arctic Silver 5 is the standard for overclockers, there may be better (and more expensive) but you won't go wrong with AS5.

Second, what verification software are you using? Does your system pass Prime95 (for example) before crashing?

Third, if you are using the stock Intel fan and cooler, you are doing it wrong. Nothing besides very modest overclocks should be attempted with stock coolers.

Is 2000 RPM the normal speed of the Intel fan or is my fan somehow broken (which could explain the overheating). In this case, any recommendation for a good fan for OCing ?

Yes, 2000 RPM is a common speed, but it more up to your Motherboard what speed will be maintained than the cooler. Megahelams is the current standard for overclockers, but it is very large and a little expensive. The Corsair H50 self-contained liquid cooler is another option.

Hypothesis I fear is the right one: can the CPU have been slowly damaged over time by this OCing, meaning there is nothing much to do except waiting for it to die ?

Spontaneous reboot/crash may be the motherboard saving itself from heat. CPU 95C under load is doable but really pushing it, especially for long term operation. You may have damaged something, or lowered the OC thresholds. The damage may be the Northbridge or RAM instead of the CPU. But, I'm thinking running at 95C is just too hot for your cooling solution. What are your Motherboard temps?

(As a side note, I am surprised that the 9550 is still around 300 $CDN here... Thought it would have been cheaper with all those i3/i5/i7 around).

The price of the 9550 is a product of supply and demand. Intel is phasing out Core 2 era CPU's, the 9550 is rather high-end and still in production, so high demand from (a large number of) upgraders and a limited supply means vendors can charge what the market will bear. MicroCenter was charging $169 recently, current $179 for Q9550 proves that Intel is not the price gouger here. Unfortunately, you gotta pick it up in store to get this price.

The final answer is this, you may have a damaged CPU, north bridge, ram, or maybe just dried up TIM or a crooked fan mount or just too much dust built up past a heat threshold. How much extra air are you getting into the case? How many input fans are you running? How many extra output fans have you added? Size and RPM? Healthy overclocks require Big Air.

Answer 2

I can't comment on the specific numbers, but by overclocking the CPU you would have made it run hotter. The stock fan would be designed to cope with the normal heat output of the CPU (but there will be some margin for error), so you have been making the CPU produce more heat than the cooling system can cope with.

While you might get away this for a while, continued operation at above normal temperatures will damage the components. So my guess is that you have been gradually damaging the CPU over time and it has now tipped over the edge into failure.

Answer 3

heh, 95C under load? Thats like 200F. Way too hot for a CPU to stay for even more than a minute. Most older CPUs(read: heat resistant) could only handle like 85C and most newer CPUs only handle 75C by spec. So according the specifications, it's probably fried.

Oh, actually the thermal max is only 71.4C

Answer 4

IMHO it's unlikely that overclocking will radically alter the behaviour over time. Parts will last for a while, but when they break, they tend break completely.

I'd echo kmarsh's recommendation of AS5 - there are a number of very good thermal pastes/greases on the market, but no matter who you ask, AS5 will be in the top 3, and it is also very affordable.

I'd also echo his suggestion of an aftermarket cooler. The Cooler Master one you have chosen is better than the stock Core 2 cooler, but you could have done better still. I'd have recommeded a Be Quiet Dark Rock Advanced (or Pro), a Titan Fenrir or perhaps a Thermaltake Frio. Caveat: I haven't checked that all these offer S775 mounts.

The next step is to check/improve case airflow. Even if you have fresh TIM and a good cooler, it isn't much use if you don't have a clear cool intake and decent exhaust. You can go to a lot of trouble to put numbers to this, but the easiest check is to feel the exhaust air from your PC. Then open your case and quickly sample the air inside. You should be able to get a feel as to whether the case layout is dragging enough cool air in and expelling enough hot air out. You might find that if key motherboard components (such as the Northbridge, or the RAM) aren't cooled sufficiently, the PC can hang (or break) - you have more to worry about than just the CPU.

Tidy cables (and other air-flow obstructions) out of the way. Use an air duster to clear vents, fans and heat-sinks - in fact do this regularly.

The final step to consider is lapping the CPU; by making sure that both the CPU surface and the HSF surface are both flat and smooth, you'll minimise the amount of air trapped in the join which acts insulation. It's a simple enough technique if you are careful. It invalidates your CPU warranty, but I think you are beyond that stage now!

Before you attempt lapping, it might be worth checking your CPU first. You can use this approach if you like, but often you can tell by the TIM distribution if you have problems.

Hope this helps.