system: Gigabyte ga-k8n ultra-9 AMD 4000+ 939 single core 2.4 Ghz 1 GB Ocz low latency 2-3-2-5 RAM Asus Nvidia GeForce EN-6800 pci-x video WD 2500JD Sata drive (JBOD on SiI 3114) SuSE 9.3 x86_64, 22.214.171.124-21.9-default kernel
I ran for almost 2 months with no serious mishaps. Within the past week I had 2 times where scores of journal replays were triggered: lost power during lightning and a failed overclock setting wherein SuSE froze during bootup. SuSE managed to recover in the first case, but the bad overclock was too much -- the reiserfs was corrupted. No effort to repair the install helped. I went ahead and reinstalled with a safer, more careful overclock.
Normally when I try a new overclock setting, I'm very careful and boot into memtest86 (v. 1.65) before I get to the Grub loader. Is this good enough, or is overclocking asking for trouble? If memtest86 shows errors I start over with cntrl-alt-delete and go back into the bios.
My current overclock CPU freq.=216 -> DDR 433 RAM freqz=200 ( 1/1 ratio) Dram relaxed as 2.5-3-3-5 CPU voltage 1.4->1.425 volts this gives 2.59 Ghz, 5.25 Ghz/s bandwidth
With the above oc, Apps load noticably faster. memtest86 predicts the numbers quite well before I get to WinXP and test with Sandra.
Except for my collasol error, this is a dream machine. I restored from a BootitNG image.
__________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com
On Sunday 09 October 2005 08:03, mark pryor wrote:
Is this good enough, or is overclocking asking for trouble?
Overclocking is asking for trouble.
With the above oc, Apps load noticably faster.
It's most likely psychological. The rule of thumb is that an unaided human cannot perceive short performance differences of less than 25%. And program loading is usually disk bound, not CPU bound. But when you put that much effort into tweaking things the results just *have* to be faster, haven't they?