Case Fan Recommendations?

Then make sure you consider the impact to your wallet too, Barry.

Here's some nice gadgets that I use to improve my temps:

TT Smart Case Fan II http://phamcomputer.safeshopper.com/35/540.htm?878
Cost: $11

TT Ducting Mod
http://phamcomputer.safeshopper.com/35/671.htm?878
Cost: $6.99

That fan comes with two ways to modulate the RPMs and resulting airflow and noise:

1. Temp Sensor - forget it, it doesn't work well at all.

2. A variable resistor (turnable pot) that lets you control the RPMs. VERY SWEET!

The fan is one of the best options for hard-core cooling geeks (I kind of fall into that category).

The ducting mod reduces noise because the fan is not mounted directly to the heatsink and also airflow is improved because you don't have the dead spot over the middle of the heatsink. Both are excellent products.

The duct can be used with typical four-hole 80mm heatsinks but not the tricked-out Thermalright SK6, SK7, or SLK series.

Barry, there are no answers to your questions about ideal operating temps or and it would be impossible to tell you what fan would be "best" for your rig. The thread has alot of good general advice. You'll have to experiment with your own rig to determine the balance of temperature, noise and cost that's right for you.

I don't have your cpu but your temps seem a bit high. I'm not sure if you mentioned what HSF you're using? That might be the place to start. I use the older but excellent Thermalright AX-7 with the above two gadgets and get excellent results.

 

I've listed all my info, including CPU and fans. Check my signature. I already have 2 of the Thermaltake Smart Fans, using the pot not the sensor. I only paid $9 each for them, delivered. 45 degree temp for CPU doesn't sound unreasonable for the AMD XP 2500+ w/ Barton, considering my ambient is 30. Sure, I'd like to go lower, but at what cost (money & noise) and how much lower?

 

It's hard to say whether your 45C temp is good or bad considering that you really don't know if that temp is "real" or not. I've been building PCs for about six years and because you never get any consistency in temps between mainboards, I've learned to take the temps with a grain of salt. In other words, I simply use the temp to see if changes I make have any effect.

As I'm sure you know:
a. some boards use the cpu's on-die diode to measure temps resulting in ~10C higher temps than in 'b'.
b. some boards use a sensor on the board which greatly under-reports the cpu's temps.
c. the cpu is most likely running hotter than temps reported in either a or b.

I would suggest you get a utility to stress-test your cpu and run it for hours. If your board shows no stability problems than it supports the conclusion that your temps are probably okay.

Whenever I put together a new rig, I loop 3DMark2001SE for hours though there are utilities whose main function is simply to stress the cpu.

 

I use SpeedFan for my temps. What would be my reason for stressing my CPU beyond my normal usage? As long as it is stable with how I use it, wouldn't doing this stress test only shorten the life of the CPU? Also, there are so many stress tests out there. How does one know which ones are legitimate and give usable information that can help diagnose any particular problem and offer solutions? How long should they be run, and what damage can be done if they fail the test? Does it need to be monitored the entire test?

 

Every rig you build should be stress-tested. And no, stress-testing isn't going to harm your cpu. Very few utilities out there will stress a cpu anywhere near its limits anyway. And a cpu's maximum safe operating temperature is way beyond the heat we generate using the PC in our everyday work.

Just keep in mind that, IMHO, you should never stress-test your hard-drive. Your HDD is a mechanical device so wear-and-tear builds up on this device. It would be like saying let's take your engine to 6000rpm and keep it there for a while and see if it can handle it - not a good thing.

The subsystems I try to stress-test are the processor, memory, and video card. None of which are going to suffer from being stress-tested.

At the very least, and I mean the VERY, VERY LEAST, run Memtest86 on your machine to see if your memory is happy and in good working order. It will write billions of patterns to your memory to see if it's defective or unhappy in your mainboard. Something like this could probably be run indefinitely as I wouldn't consider this to be real stress. I run this for at least 3 hours straight.

Prime95 is a popular cpu stressor. I have never used it though.

I usually just run 3DMark2001SE and Memtest86. To different extents, these apps will stress your cpu, video subsystem, and memory. These tests can reveal problems like faulty memory, incompatible memory, an insufficient power supply, or even poor case ventilation or a video card with poor cooling. The bottom line is that if your PC can't pass these tests, it's really not ready for "prime time ".

If you don't run these tests, your first indication of a problem may be a blue screen of death.

 

Prime95 seems to be geared toward Pentium, not Athlon. Is there a better utility for the AMD Athlon XP 2500+ w/ Barton? If not, would you recommend using the standard or the beta version? I ran 3 passes (~43min each) of the memtest86. No errors were found. I didn't see any reason to continue the looped tests.
I've been operating Prime95 v 2212. I couldn't figure out how to operate the beta version (235). It's been running since yesterday morning. I'm surprised that it doesn't even seem to affect my computer usage. I am operating at 100% CPU usage and 17 degrees C above my case temp, which continues to hover about 2 (maybe pushing 3 on occasion when the room heats up) degrees above ambient. I'm not sure exactly how long I am to run this program or what I am to look for. It's working on Stage 2 now (20% complete). Please let me know what I am to do now.
Prime95 says it "is a good stress test for the CPU, memory, caches, CPU cooling, and case cooling." It continues to say, "You'll need other programs to stress video cards, PCI bus, disk access, networking and other important components." It recommends running the program 6-24 hours. I guess my computer must be OK with good hardware and stability. I guess I didn't do too bad, considering I never used a computer till 2 years ago, and this is the first time I've totally rebuilt a computer from scratch. (Actually, I did take a computer class in 1970 but got turned off to all the cards I had to produce to input one line of data, so I wrote off computers for 3 decades.)