-   Linux - Hardware (
-   -   case airflow, and fan control, HD cable question, and stress testing help a NEWBIE (

zetsui 03-14-2004 10:05 AM

case airflow, and fan control, HD cable question, and stress testing help a NEWBIE
1. I have a switch for my volcano 9 that allows me to control the fan speed, the only problem is that I do not know where I can put this little switch. Its size is like a screw driver and it has one vertical slash in it, it's white and made out of plastic. I am thining of drilling a hole in the top of my case, but would that do any damage, or somehow mess up the airflow?

2. Also how do you think my airflow is so far? Picture a rectangle (that's my case) on the top right is the PSU and it's fan blowing OUT. Then there is the fan in the heatsink blowing onto the CPU.

3. The HD cable that shows the status of the HD (red blinking light). I have it setup as the front of the case connected to the mobo. The mobo has wires connected to the HD. Is that correct.

4.What is the best way I can stress test my computer's overclocking?

PsychosisNode 03-15-2004 02:03 PM

1. The switch you describe is a "set and forget" switch - you should NOT have to keep altering this. Therefore, you should not need a hole to fit this switch into. If you wish to adjust fan speed in accordance with the CPU temperature then you need a different type of control box that has a temperture sensor that you (usually) glue or tape to the underside of your CPU.

2. The required airflow depends on how much heat your system is producing. For a "normal" system, say an Athlon XP2000 or equivalent plus a fast video card then I reccomend an aditional extraction fan - many cases have further fan mounting holes situated below the PSU. Even faster systems with two or more hard drives need an intake fan fitted lower down in the front of the case.

3. I'm assuming you are refering to the HDD indicator built into the case. This should have a corresponding two-pin connector (often labeld "HDD") that fits into the HDD indicator socket on your motherboard. The manual for the motherboard will show you which socket this is.

4. Before you start testing, make sure that you, from the BIOS, set the CPU overheat alarm to a safely low setting (say 70 degrees Celcius). If the overheat alarm goes off during testing, you're overstressing the CPU. However, if you don't expirience any thermal issues, you can write a shell script (or just google for some) that compiles the same kernel several times, then compares the binary output. This may sound weird but compilation is amongst the most stressful tasks you can give your CPU. If the binaries are different, or you hit bizarre errors during the compilation then you're most likely overstressing your CPU.

zetsui 03-15-2004 03:07 PM

i definately agree about multiple compiles

Thx everybody for the help I really appreciate it

michael@actrix 03-17-2004 03:36 AM

Use lm_sensors to monitor you CPU and Mainboard temperatures - installed on most modern distros:

And use an external thermometer to measure the room temperature. hddtemp can get the IDE drive temperatures.

If you want to monitor your readings over time look at rrdtool - you can set yourself up with time series data sets and graph them - scripting skills will be required. But pen and paper notes are probably good enough for tuning.

Do some stress tests, such as kernel builds, and see how the temperatures fluctuate allowing for the room temperature. Google should help in finding what good temperatures to expect for a given CPU - beware that some overcloaker web pages tend to want things really cool - so you can aim for a few degrees above where they aim.

I have an Athlon XP 1700+ - people have reported that Athlons can start to get unstable somewhere above 55C - but some chips are quite stable above this - luck of the draw - mine seems 100% stable at 58C, a friends Athlon survived 100C when his CPU fan failed. My two hard drives seem to run in the range 32-35C. My CPU idles at about 46-48C. My mainboard idles at about 36-38C.

AMD recommends a vent fan below the power supply moving air out of the case - and also recommends against inlet fans at the bottom of the case - they claim this can actually reduce cooling to the CPU - but they're probably ONLY worried about the CPU.

I wanted a quiet machine - so high reving fans are not on my shopping list. I found that adding an extra low rev cooling fan below the power supply venting out did improve all the temperatures. I eventually switched to a low noise (low rev) CPU cooler. When a really hot summer kicked in, I removed the entire plastic front panel and took base measurements with front to back airflow at the maximum. I then mod'ed the front panel by enlarging some downward pointing inlets (nobody can see them - so it looks OK). I then replaced one 5.25" bay cover with a 5.25" cover designed to accommodate a 3.25" peripheral, but instead of a peripheral I added a grill made from an old PC-speaker grill. With the extra air inlets the system now runs within 1C of the base measurements. If I ever buy another mid-tower I'll look for large speaker-like front grills with dust filters.

All times are GMT -5. The time now is 03:43 AM.