Deanodarlo
Senior member
The beta bios (16th Jan) is excellent but I've noticed a strange issue with it.
I'm not sure if this effects the official 8th Jan bios also, but I guess it does as the 16th is the 8th with minor changes to allow 4-way interleave over 144 FSB.
Whenever I use a multiplier below 7, memory scores drop by over 60 on average in memtest
This utility is excellent for overclocking because it boots from disk and gives a good indication of stability without corrupting your hard drive.
NB/
Using a Duron 800, max o/c is 1000 stable. Lowest multiplier is 6, so the highest FSB I can use is 6*166.
For instance,
133*7 (933) = 474 MB/S
144*6.5 (936) = 404 MB/S
139*7 (973) = 496 MB/S
162*6 (972) = 453 MB/S
140*7 (980) = 499 MB/S
150*6.5 (975) = 420 MB/S
Therefore even though the new bios resolves my reboot issue at 166FSB, it kills my memory scores because I need to use a multiplier lower than 7 !!!!!.
At the moment, the highest I can run is 7*143 (1001) without killing memory bandwidth using sub 7 multipliers.
After further tests I conclude multipliers above 7 seem unaffected by this strange bug (I think, only tried 7, 7.5 and 8). Must go and talk to Epox Tech, this can?t be hard to fix???.
I'm not sure if this effects the official 8th Jan bios also, but I guess it does as the 16th is the 8th with minor changes to allow 4-way interleave over 144 FSB.
Whenever I use a multiplier below 7, memory scores drop by over 60 on average in memtest
This utility is excellent for overclocking because it boots from disk and gives a good indication of stability without corrupting your hard drive.
NB/
Using a Duron 800, max o/c is 1000 stable. Lowest multiplier is 6, so the highest FSB I can use is 6*166.
For instance,
133*7 (933) = 474 MB/S
144*6.5 (936) = 404 MB/S
139*7 (973) = 496 MB/S
162*6 (972) = 453 MB/S
140*7 (980) = 499 MB/S
150*6.5 (975) = 420 MB/S
Therefore even though the new bios resolves my reboot issue at 166FSB, it kills my memory scores because I need to use a multiplier lower than 7 !!!!!.
At the moment, the highest I can run is 7*143 (1001) without killing memory bandwidth using sub 7 multipliers.
After further tests I conclude multipliers above 7 seem unaffected by this strange bug (I think, only tried 7, 7.5 and 8). Must go and talk to Epox Tech, this can?t be hard to fix???.