Intro-
It took me awhile, I spent lots of time reading forums and standing in the return line at Fry's, but I've done it. Started out with an Abit IS7, returned that piece of crap, bought an IC7- and got up to a 250Mhz FSB stable. The burning smell should have been alarming, but I figured it was probably just all the case fans (4 Tornados [tornadoes?] ). I was standing in my living room, looking at my desk, and my computer just shuts off! "Aww crap, I was just going to watch that monkey knife fight video," I thought. It wouldn't boot, it was as dead as Chris Farley. I thought it was either the mobo or power supply, so I went out and got a 500W PSU and an Asus P4C800 (my original PSU was 420W max). Well, I put the power supply in my case and the IC7 booted, but looking at the unopened Asus box got me all antsy in my pantsy and my overclocking instincts took over. I thought "maybe I can push THIS mobo a little further- and I was right. So I finally ended up with an Asus P4C800, a 3Ghz P4, and 2x512 Mushkin Lvl II pc3500 dual channel. I upped the FSB in small increments- finally coming to a stable 256FSB and 3845Mhz of pure power.
Cooling-
The heatsink is a Thermalright 947u, w/ Arctic Silver 5. I originally had an 80mm Vantec Tornado on the heatsink, but switched to a 92mm Tornado since I figured I could use the extra CFMs. After switching to the 92mm Tornado, load temps dropped a few degrees, and its not even that loud- the 4 80mm case fans were another story. They moved a TON of air, my computer was practically scooting across the room, but with all of them running, I could barely hear the phone ring, so I got a Nexus fan controller. With all 4 fans on the slowest speed setting, the noise has dropped considerably, and temps only rose by 0.5c. One would think that lowering the total air thoroughput through the case by at least 75% would increase CPU temps noticeably- but it didn't. My guess as to why is based upon the following- fast moving air is at a lower pressure, and less air molecules striking the heat sink would result in less total energy transfer from heatsink to air. The case fans were not pushing air over the heatsink, but through the case, result in a lower pressure. So as long as you exhaust the case air at a reasonable rate, the processor can stay cool. The 4 Tornados, even at a low setting, do push plenty of air.
I'm a Linux guy at heart, but originally had WinXP Pro on the machine so that I could monitor temps using Motherboard Monitor, and until I could get it stable running Prime95 & Seti together for 24 hrs. With Linux, you have to compile Lm_sensors, and I wanted to make sure the temps were accurate. I also ran Prime95 by itself, using small FFTs for 8 hrs and large FFTs for 8 hours. Solid as a rock!
Now for the Linux mojo. I first installed Slackware 9.1 w/ hyperthreading on in BIOS, and everything ran fine. But when I tried to recompile the kernel to add in hyperthreading (configure for SMP and ACPI w/ HT only) and High MEM support since I have a Gig of RAM, the kernel would build fine but when I rebooted, prime95 would segfault and give me kernel errors (it's actually called 'mprime' for Linux). Trying to compile anything else would give me kernel errors. So I reloaded the bare.i kernel from the CD and ran Prime95 for 24 hrs, and everything was stable, BUT I WANTED MY HYPERTHREADING!!! I was confused- so I recompiled the kernel with the default bare.i options- and once again everything was giving me segfaults. So I finally figured out that I needed to turn OFF hyperthreading while installing Slackware and building the kernel for the first time, but once SMP and HT support were in the kernel, I could turn HT on just fine. IT WORKS!!! Prime95 runs like a champ, and /proc/cpuinfo and top show two processors. I compiled Lm_sensors and installed GKrellM to monitor things, sensors reports the same temps that Mothergboard Manager did (33c idle, 43c load). So I have Slackware 9.1 w/ the 2.4.22 kernel running, and WinXP running on my fourth virtual desktop under VMware. I don't have a wicked fast Vid card, only a Geforce Ti4400, since I don't do any gaming, only number crunching w/ Matlab and Maple.
This thing is FAST!!! This processor compiles things like it's going out of style, Maple and Matlab run blazing fast. Thanks for all your help (although passively- since I never asked any questions, but read suggestions).
sandeezy- san diego fo' sheezy
It took me awhile, I spent lots of time reading forums and standing in the return line at Fry's, but I've done it. Started out with an Abit IS7, returned that piece of crap, bought an IC7- and got up to a 250Mhz FSB stable. The burning smell should have been alarming, but I figured it was probably just all the case fans (4 Tornados [tornadoes?] ). I was standing in my living room, looking at my desk, and my computer just shuts off! "Aww crap, I was just going to watch that monkey knife fight video," I thought. It wouldn't boot, it was as dead as Chris Farley. I thought it was either the mobo or power supply, so I went out and got a 500W PSU and an Asus P4C800 (my original PSU was 420W max). Well, I put the power supply in my case and the IC7 booted, but looking at the unopened Asus box got me all antsy in my pantsy and my overclocking instincts took over. I thought "maybe I can push THIS mobo a little further- and I was right. So I finally ended up with an Asus P4C800, a 3Ghz P4, and 2x512 Mushkin Lvl II pc3500 dual channel. I upped the FSB in small increments- finally coming to a stable 256FSB and 3845Mhz of pure power.
Cooling-
The heatsink is a Thermalright 947u, w/ Arctic Silver 5. I originally had an 80mm Vantec Tornado on the heatsink, but switched to a 92mm Tornado since I figured I could use the extra CFMs. After switching to the 92mm Tornado, load temps dropped a few degrees, and its not even that loud- the 4 80mm case fans were another story. They moved a TON of air, my computer was practically scooting across the room, but with all of them running, I could barely hear the phone ring, so I got a Nexus fan controller. With all 4 fans on the slowest speed setting, the noise has dropped considerably, and temps only rose by 0.5c. One would think that lowering the total air thoroughput through the case by at least 75% would increase CPU temps noticeably- but it didn't. My guess as to why is based upon the following- fast moving air is at a lower pressure, and less air molecules striking the heat sink would result in less total energy transfer from heatsink to air. The case fans were not pushing air over the heatsink, but through the case, result in a lower pressure. So as long as you exhaust the case air at a reasonable rate, the processor can stay cool. The 4 Tornados, even at a low setting, do push plenty of air.
I'm a Linux guy at heart, but originally had WinXP Pro on the machine so that I could monitor temps using Motherboard Monitor, and until I could get it stable running Prime95 & Seti together for 24 hrs. With Linux, you have to compile Lm_sensors, and I wanted to make sure the temps were accurate. I also ran Prime95 by itself, using small FFTs for 8 hrs and large FFTs for 8 hours. Solid as a rock!
Now for the Linux mojo. I first installed Slackware 9.1 w/ hyperthreading on in BIOS, and everything ran fine. But when I tried to recompile the kernel to add in hyperthreading (configure for SMP and ACPI w/ HT only) and High MEM support since I have a Gig of RAM, the kernel would build fine but when I rebooted, prime95 would segfault and give me kernel errors (it's actually called 'mprime' for Linux). Trying to compile anything else would give me kernel errors. So I reloaded the bare.i kernel from the CD and ran Prime95 for 24 hrs, and everything was stable, BUT I WANTED MY HYPERTHREADING!!! I was confused- so I recompiled the kernel with the default bare.i options- and once again everything was giving me segfaults. So I finally figured out that I needed to turn OFF hyperthreading while installing Slackware and building the kernel for the first time, but once SMP and HT support were in the kernel, I could turn HT on just fine. IT WORKS!!! Prime95 runs like a champ, and /proc/cpuinfo and top show two processors. I compiled Lm_sensors and installed GKrellM to monitor things, sensors reports the same temps that Mothergboard Manager did (33c idle, 43c load). So I have Slackware 9.1 w/ the 2.4.22 kernel running, and WinXP running on my fourth virtual desktop under VMware. I don't have a wicked fast Vid card, only a Geforce Ti4400, since I don't do any gaming, only number crunching w/ Matlab and Maple.
This thing is FAST!!! This processor compiles things like it's going out of style, Maple and Matlab run blazing fast. Thanks for all your help (although passively- since I never asked any questions, but read suggestions).
sandeezy- san diego fo' sheezy