68030 not running at 25Mhz

[ChaosLord]

Is it just 1 crystal that controls both CPU and FPU?

Or are there 2 different crystals?

Maybe the FPU crystal speed is somehow interfering with the CPU crystal speed?

I don't really know about such things.  I am just throwing out an idea.



Are your CPU caches on?  If some evil program somewhere is disabling them for some reason this could explain slowness.

[ChaosLord]

@Lurch

Please type

Code: [Select]


CPU

into a shell window and paste us the results.

p.s. I would not try to use an 80Mhz crystal with a 25Mhz cpu.  That sounds very dangerous.  If you are going to do that then please let your CPU say goodbye to her friends and family first.

[ChaosLord]

According to TBBOAH ram on this card is not autoconfigured. No (active) 32 bit ram will result in terrible performance.

What you are saying is 100% true.

BUT....

None of this makes any sense at all.

His 32-bit ram must be active because he told us his instruction cache was enabled and using burst mode. :idea:

But then why did he have his Datacache Burst Mode disabled? :crazy:

None of this makes any sense.

Either both instruction cache and datacache should be in burst mode or neither one should be. :rtfm:

I hereby award this accelerator card the "Screwiest Accelerator Ever Award" :insane:

[ChaosLord]

Benchmarks as a 13.4Mhz CPU not at 25Mhz or higher. Sysinfo/Sysspeed/AIBB all the same.

Those programs measure your cpu speed by executing instructions from the instruction cache.

Maybe your instruction cache is getting errors by trying to burst from a memory area that does not support bursting.

Try disabling the burst mode on your instruction cache to see if that makes any difference.  Make sure not to disable the whole instruction cache.  Only the Burst Mode.

[ChaosLord]

Sysinfo has no idea what clock speed your CPU is running at. It just makes a guess based on some timed loops which run at different speed on different CPUs... It probably isn't accurate on anything other than a 68000...

If sysinfo is properly coded then it should be extremely accurate on all CPUs.

I am not saying it is properly coded.

I am just saying that if it was then using the "timed loops" method is very reliable.  It would only be guaranteed unreliable on new chips released after the program was written.

[ChaosLord]

If your card is getting its Mhz from the motherboard then you could plug in a Commodore Genlock and dial up the knob and see if it increases the speed of your CPU.  If your cpu speed goes up then its getting its speed from the Amiga.

Its a trick someone told me many many years ago.