How not to build a PC

[c_hegge]

1. Don't put a P4 Prescott CPU in an SFF case

2. Don't leave cables loose, especially with IDE cables

3. If possible, use a case fan or 2

4. NEVER do all of the above. As I type this, there's a little PC sitting next to me which I'm stress testing. It is a P4 Prescott 3.0GHz, with an Antec NSK1380 case, NO case fans whatsoever, and the cables are all in a birds nest. The outside of the case is currently too hot to touch and the CPU is hitting 85°C with the fan at max speed (3200RPM).

I actually put this together. It's my new way of stress testing re-capped motherboards. Not only do I hammer the CPU for a few days, but I now do it in an SFF case with no ventilation

[acstech]

Re: How not to build a PC

That poor little processor. I almost feel sorry for it. Well, almost.

[bigbeark]

Re: How not to build a PC

Pity the nearby components. Prescotts seem to thrive on their own heat!
Too bad they end up pan-frying Northbridge chips and any caps silly enough to get near them.

[Wizard]

Re: How not to build a PC

Check your Prescott CPU thermal throttling. There should be a utility out there that will tell you if this thermal throttling is happening.

Cheers, Wizard

[acstech]

Re: How not to build a PC

Pity the nearby components. Prescotts seem to thrive on their own heat!
Too bad they end up pan-frying Northbridge chips and any caps silly enough to get near them.

I think there's a term for that...

http://en.wikipedia.org/wiki/Nuclear_meltdown

That said, these are good suggestions. I generally consider it a problem when I see processor temperature go above about 60c, since I know things are only going to get worse when in the hands of the customer. Heatsinks fill up with dust, fans fail etc.

What program are you using to stress the processor? May I suggest Intel Burn Test?

http://majorgeeks.com/IntelBurnTest_d5987.html

[c_hegge]

Re: How not to build a PC

It will only be there for the stress testing. Once that's over it will get put in a standard case with some better ventilation.

To stress the CPU, I use prime95.

[bgavin]

Re: How not to build a PC

This is about the same as stress testing an egg on a railroad track.
The egg isn't much good after the test.

Why do you expose your gear to such detrimental conditions?
Do you think the processor 'heals itself' after coming back down from severe temps?

If you want to stress test your hardware, give it optimum running conditions (read: cooling) and drive it to maximum performance. Prime95, etc will do the job for you.

[c_hegge]

Re: How not to build a PC

I don't normally do this. I was just curious this time around to see how it would handle the heat, but it's the first time I've done this and will probably be the last.

[Spode]

Re: How not to build a PC

I'm sure I remember a particular memory manufacturer stress testing their memory in an oven at extreme temperatures before shipping them out, to make sure that the overclocking was stable.

And no, a Prescott CPU never deserves sympathy

[TheLaw]

Re: How not to build a PC

1. Don't put a P4 Prescott CPU in an SFF case

2. Don't leave cables loose, especially with IDE cables

3. If possible, use a case fan or 2

4. NEVER do all of the above. As I type this, there's a little PC sitting next to me which I'm stress testing. It is a P4 Prescott 3.0GHz, with an Antec NSK1380 case, NO case fans whatsoever, and the cables are all in a birds nest. The outside of the case is currently too hot to touch and the CPU is hitting 85°C with the fan at max speed (3200RPM).

I actually put this together. It's my new way of stress testing re-capped motherboards. Not only do I hammer the CPU for a few days, but I now do it in an SFF case with no ventilation

I have a 3.06GHz Northwood in a SFF case. I had to buy a copper heatsink and a Delta fan to keep it cool. I kid you not.

[c_hegge]

Re: How not to build a PC

I was just using the stock cooler here.