Marcel, As I am working on accelerated boards, would you be willing to outline the tuning/timing issues and what code needs editing to achieve correct performance? How do you test? By successive approximation, or using a calculation?
EG: Issue 8 runs the logic at 8MHz and the 8049 has a standard 11.0592 xtal (1/8th the price of an 11.0000 xtal and still works) for baud rate generation. This allows the Issue 8 to eke out about 10% extra CPU performance by running the CPU a little faster. Sidebar: I've been learning about the dirty tricks the QL pulls by asserting /DTACK long before the data is actually ready in the hope it will be, and how this limits CPU speed. It's surprising what you can do with much faster RAM *and* a slight delay on the 8301's assertion of /DTACK will do. Passing /DTACK through all four inversions of a 78LS00 bought me a 33ns delay, and got me a bit of overclockability. That was a night, I can tell you! Catch is, how does the CPU tell where /DTACK is being generated from? (Nasta is a good teacher.) PS: "Overclockability" IS SO A WORD! :P Dave On Sun, Mar 4, 2018 at 9:59 AM, Marcel Kilgus via Ql-Users < ql-users@lists.q-v-d.com> wrote: > I have released a new (or actually two new) versions of TK2. > > One versions only includes the timing critical parts of the QL-NET > code and the rest can be LRESPRed afterwards. > > The other versions included the complete QL-NET code and misses the > "ED" SuperBasic editor instead, which can be LRESPRed afterwards. > > Depending on your QL-NET affinity you can chose which version you like > more, in the end both version have the same capabilities when the > missing parts are loaded. > > https://www.kilgus.net/2017/03/19/toolkit-ii-the-sequel/ > > Marcel > > _______________________________________________ > QL-Users Mailing List > -- Dave Park d...@sinclairql.com _______________________________________________ QL-Users Mailing List
