6502.org

Mostly just to post this somewhere for myself so I don't lose it again; there is another suite of 65816 tests (amongst many others) over here:

https://github.com/SingleStepTests/Proc ... main/65816

Very useful for emulators that can arbitrarily set memory values and the program counter because it's just a bunch of JSON for the start state and the expected end state.

All recommendations are de facto opinions, Ed, whether offered individually or by a group.  It should be obvious to others reading this topic that my recommendation is an opinion, one based upon extensive experience in programming the 65C816.  I have never suggested that my recommendations are consensus.

Interesting find Andrew, thanks! (Single instructions, full bus activity, randomly generated, verified against hardware)

Well, I finally got my SBC-4 reconfigured and ran the full test suite and all tests passed, including the BCD math. This confirms the suite's tests are accurate.

If you haven't dug into the code, both binary and BCD math tests all the status bits - Carry, Zero, Negative, and Overflow. All are accurate in both modes, including math on invalid BCD numbers. Don't ask me how many hours I worked on that feature.

I am much more confident now that the 65816 Simulator is mostly working. I still have to work on getting the cycle counting verified, as I was focused on instruction accuracy. I'm sure there will be many exceptions to the basic cycle times listed in the datasheet.

Daryl

Splendid news, well done! I get the feeling this test suite will lead to a great leap forward in several ways.

Seems that there is no uart.asm in the zip file. I commented out the include and it assembled and ran without issue though.

Oops - uart.asm was for the sbc-4. The Kowalski version just uses the io.asm for terminal IO.

Good catch! I'll edit the post with the update.

thanks!
Daryl

The zip file has been updated in the head post. It also includes a fix for a few 16-bit immediate values that were not declared with the !# operand specifier. This would have resulted in errors for the pending 1.4.0.3 version release coming very soon.

Daryl