G
Guest
Guest
I posted this info on the old forum but here it is again since a lot of people have been expressing interest in this chip.
First, in Tom's article he says that the voltage for the K6III+ is 1.8 and the picture he has of a K63-500+ has 1.8v stamped on the chip. My K63450+ however, came with 2.0 volts stamped on the chip so I used that as my starting point. Hear are some of the results I have achieved so far. This is with a Tyan 1598 ATX board which I highly recommend for stability (although it doesnt do bus speeds over 100 unless there is an undocumented setting for it). While no board Ive seen officailly supports the K62+ or K63+, my tyan board recognizes this chip as an ordinary K63 which should be fine as far as I can tell. The Tyan board is also great in that it offers voltages from 1.3 all the way to 3.5 in .1v increments. All of the results below were with a standard but decent quality heatsink and fan. Nothing exotic.
4.5X100 = 450 2.0v
This is the spec setting - no problems. With case temp at 72F, cpu temp was 95F while sitting in the bios setup and 88F in Windows 2000.
5.0x100 = 500 2.0v
Once again, no problems. With case temp at 72F, cpu temp was 98F while sitting in the bios setup and 90F in Windows 2000.
5.5x100 = 550 2.0v
Once again, no problems. With case temp at 72F, cpu temp was 107F while sitting in the bios setup and 95F in Windows 2000.
6.x100 = 600 2.0v
The 2.0/6.0 multiplier is undocumented. The manual for the Tyan 1598s only lists multipliers from 2.5x to 5.5x. But, after some experimentaion, I discovered that On/off/off for jumpers jp1/jp2/jp3 will give you this setting. At 600 the system booted to win2K and ran OK but with intermittant errors. With case temp at 72F, cpu temp was 115F while sitting in the bios setup and 100F in Windows 2000. This setting was no where near stable enough to satisfy me.
6.x100 = 600 2.1v
After the previous attempt at 600 failed I decided to up the voltage from 2.0 to 2.1. Two things happened. First, temperatures went up significantly. With case temp at 72F, cpu temp was 135F while sitting in the bios setup and 123F in Windows 2000. Second, the system was much more stable but still not stable enough for me. It was ever so close though!
I believe that at 600 and 2.1v settings heat may have been comming into play and that the chip could probably be made to run at that speed with better cooling. However since the point of this was to do a cheap upgrade (the K63-450+ only cost me $42.95 plus shipping) spending more money on better cooling didnt make any sense. Besides, subjectively I cant tell the difference between 550 and 600 anyway. So, I backed off to 5.5x100 at 2.0 volts. I'm extremely pleased with that. A 100Mhz overclock is a bit of a first for a chip in the K6 line, and now this system will keep me going for a few more months till I go with either an Athlon or PIII with DDR.
First, in Tom's article he says that the voltage for the K6III+ is 1.8 and the picture he has of a K63-500+ has 1.8v stamped on the chip. My K63450+ however, came with 2.0 volts stamped on the chip so I used that as my starting point. Hear are some of the results I have achieved so far. This is with a Tyan 1598 ATX board which I highly recommend for stability (although it doesnt do bus speeds over 100 unless there is an undocumented setting for it). While no board Ive seen officailly supports the K62+ or K63+, my tyan board recognizes this chip as an ordinary K63 which should be fine as far as I can tell. The Tyan board is also great in that it offers voltages from 1.3 all the way to 3.5 in .1v increments. All of the results below were with a standard but decent quality heatsink and fan. Nothing exotic.
4.5X100 = 450 2.0v
This is the spec setting - no problems. With case temp at 72F, cpu temp was 95F while sitting in the bios setup and 88F in Windows 2000.
5.0x100 = 500 2.0v
Once again, no problems. With case temp at 72F, cpu temp was 98F while sitting in the bios setup and 90F in Windows 2000.
5.5x100 = 550 2.0v
Once again, no problems. With case temp at 72F, cpu temp was 107F while sitting in the bios setup and 95F in Windows 2000.
6.x100 = 600 2.0v
The 2.0/6.0 multiplier is undocumented. The manual for the Tyan 1598s only lists multipliers from 2.5x to 5.5x. But, after some experimentaion, I discovered that On/off/off for jumpers jp1/jp2/jp3 will give you this setting. At 600 the system booted to win2K and ran OK but with intermittant errors. With case temp at 72F, cpu temp was 115F while sitting in the bios setup and 100F in Windows 2000. This setting was no where near stable enough to satisfy me.
6.x100 = 600 2.1v
After the previous attempt at 600 failed I decided to up the voltage from 2.0 to 2.1. Two things happened. First, temperatures went up significantly. With case temp at 72F, cpu temp was 135F while sitting in the bios setup and 123F in Windows 2000. Second, the system was much more stable but still not stable enough for me. It was ever so close though!
I believe that at 600 and 2.1v settings heat may have been comming into play and that the chip could probably be made to run at that speed with better cooling. However since the point of this was to do a cheap upgrade (the K63-450+ only cost me $42.95 plus shipping) spending more money on better cooling didnt make any sense. Besides, subjectively I cant tell the difference between 550 and 600 anyway. So, I backed off to 5.5x100 at 2.0 volts. I'm extremely pleased with that. A 100Mhz overclock is a bit of a first for a chip in the K6 line, and now this system will keep me going for a few more months till I go with either an Athlon or PIII with DDR.