ACT available

[brouhaha]

(02-25-2015 04:30 AM)PANAMATIK Wrote:  

(02-25-2015 01:04 AM)Mark Hardman Wrote:  Would it be possible to post your source patches so that we can create a working HP-29C KLM? Or could you pass along the source changes so that Eric can update the source tar ball?

Thanks!

I need some time until the tests are complete, this will be not before next week. Eric knows about my patches.

Bernhard

I have patches that resolve the problem in a different way. I thought they were in the last Nonpareil source release, but perhaps I'm mistaken. I'll have to check.

I don't think either Bernhard's or my patches are a completely accurate model of how the ACT works, but either patch is adequate to solve the label search problem of the 19C/29C/67/97.

The HP engineer that wrote the label search code (likely William Egbert) cleverly took advantage of a boundary condition in the ACT behavior to optimize the loop. I've seen an early ACT specification which documented that there was unexpected behavior when the P register wraps around after an increment or decrement, but didn't give a precise explanation of the actual behavior. The engineer writing the ACT spec probably didn't anticipate that the actual behavior would be useful, and might not have even known exactly what that behavior was. After the behavior was found useful in the label search loop, it was probably documented in a later revision of the ACT spec, since the microcode became dependent upon it.