I got my 75MHz oscillator today and so far it's been rock solid in my CSMK3. Here's my bustest...
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
fast $08AB0000 readw 35.7 ns normal 56.0 * 10^6 byte/s
fast $08AB0000 readl 58.6 ns normal 68.3 * 10^6 byte/s
fast $08AB0000 readm 59.7 ns normal 67.0 * 10^6 byte/s
fast $08AB0000 writew 39.6 ns normal 50.5 * 10^6 byte/s
fast $08AB0000 writel 79.3 ns normal 50.5 * 10^6 byte/s
fast $08AB0000 writem 82.8 ns normal 48.3 * 10^6 byte/s
chip $00020000 readw 859.6 ns normal 2.3 * 10^6 byte/s
chip $00020000 readl 860.1 ns normal 4.7 * 10^6 byte/s
chip $00020000 readm 859.8 ns normal 4.7 * 10^6 byte/s
chip $00020000 writew 565.3 ns normal 3.5 * 10^6 byte/s
chip $00020000 writel 565.6 ns normal 7.1 * 10^6 byte/s
chip $00020000 writem 565.8 ns normal 7.1 * 10^6 byte/s
Nice memory speedup from overclocking. Processor speed is still more important because of
the caches though. I have heard of a guy with an Apollo '060 that claimed 100 MHz. Bad
thing about the Apollo is the bad/broken SCSI. The Quickpack '060 with EDO SIMMs would
have faster memory access than my CSMK3. I wonder how well they overclock. One last note
about SIMMs. They should be as least as fast as they are marked but some that are not
marked fast are. For example, I have some 16Mb SIMMs that are marked 60ns but tested 47ns
on a SIMM tester. Some SIMMs may work but cause problems or produce errors later when they
heat up. Some accellerators just don't like certain SIMMs. The best thing is to try
different SIMMs. Also, fit as few of SIMMs as possible during testing.