VOGONS


First post, by lalakobe

User metadata
Rank Newbie
Rank
Newbie

Hello all!

I have in posession Investronica M TUB-002 8086 motherboard (https://theretroweb.com/motherboards/s/invest … onica-m-tub-002) which is 8 MHz one. It has two oscillators on board, one 14 Mhz, the other 48 Mhz. I think that 48 one is divided by 2, and then by 3, giving 8 Mhz bus speed.

I want to replace 48 Mhz oscillator with 60 Mhz in order to run 10 MHz CPU (NEC V30 or Intel D8086-1) at full speed. There are 5V and 3.3V ones in local store, I guess I need 5V one?

Reply 2 of 8, by lalakobe

User metadata
Rank Newbie
Rank
Newbie

I have replaced oscillator to 60 Mhz one. Checkit shows me that CPU is 12 Mhz. What are reliable CPU test software of that era that can show real CPU speeds?

My CPU is Intel D8086-1 (https://www.cpu-world.com/CPUs/8086/Intel-D8086-1.html)

Shouldn't Checkit show real 10 Mhz speed?

Last edited by lalakobe on 2022-09-30, 16:11. Edited 2 times in total.

Reply 5 of 8, by lalakobe

User metadata
Rank Newbie
Rank
Newbie

chkcpu requires 386+ cpu, aida16 hangs on start, drhard also shows 12 Mhz.

My idea was that 48Mhz is divided by 6 thus giving 8Mhz bus and CPU speed. By replacing to 60 we should get 10Mhz, not 12. How is that possible? Could it be that tests are somehow wrong?

Reply 7 of 8, by mkarcher

User metadata
Rank l33t
Rank
l33t
lalakobe wrote on 2022-10-01, 12:11:

My idea was that 48Mhz is divided by 6 thus giving 8Mhz bus and CPU speed. By replacing to 60 we should get 10Mhz, not 12. How is that possible? Could it be that tests are somehow wrong?

Your idea is very reasonable. I would have expected the same. The 8086 can reach its top performance only if the clock low period is twice as long as the clock high period. At 10MHz, this means 67ns clock low and 33 ns clock high. That's why the 10MHz clock is usually generated from a 30MHz oscillator. Before integrated chipsets, this job was usually performed by the Intel 8284 clock/ready generator. So a divide-by-6 operation makes much more sense than a divide-by-5 operation. The benchmarks you used all compare the CPU speed in reference to the 14.318MHz oscillator that is used to drive the timer circuit. As long as you didn't swap the 14.318MHz oscillator, the benchmark results are most likely acurate. I'm afraid that the ALi M1203 chipset on your board possibly doesn't cope well with the 60MHz clock and mis-divides it. If you have a scope with sufficient bandwidth, take a look at the frequency and the duty cycle of the 8088 CLK input pin.

Reply 8 of 8, by lalakobe

User metadata
Rank Newbie
Rank
Newbie
majestyk wrote on 2022-10-02, 13:17:

Do the settings of JP2, JP3, JP4 and JP5 have any impact on the CPU frequency?

JP1 enables IDE controller
JP2 and JP3 switches FDD A and B between 1.2/1.44 and must be set both

Will check JP4 and JP5 and report back.