Adding support for PDP- Major States within the CPU.

[poetnerd]

• Context

• Oscar Vermeulen has produced kits that offer a 3/4 scale model of traditional DEC hardware with software emulated on a Raspberry Pi: PiDP-10, PiDP-11, and the first, and still very popular PiDP-8/i. The joy of these kits is two-fold: experiencing old computer hardware and software architectures, and playing with the lights and switches of very-well emulated front panels. See: PiDP-8/I: RECREATING THE PDP-8/I

• The PiDP-8/i looks like the old PDP-8/i with the major registers and major states on display.

• SIMH doesn't emulate down to the Major States. It's an instruction-level emulation. Indeed, the first foundational principle reported at [About the Open SIMH project(https://opensimh.org/about/) says, but not cycle-accurate, simulation.

• The PiDP-8/i software project has developed, tested and benchmarked a fork of the OPEN SIMH PDP-8 emulation that supports the major states. This is of value to the PiDP-8/i project because it means that the "SING STEP" button on the emulated PDP-8/i front panel actually works and steps through the Fetch/Defer/Execute Major States.

• Benchmarking says that performance with emulation of Major States is half the performance of the instruction-level emulation.

• We would really like to get out of the business of maintaining a fork of SIMH. But we recognize that there's a performance penalty, and the proposed change is a step down the road of "cycle-accurate simulation."

• QUESTION: Might the PDP8 SIMH maintainers be willing to accept a pull request to add Major State support?

• the output of "sim> SHOW VERSION" while running the simulator which is having the issue

PDP-8 simulator Open SIMH V4.1-0 Current Simulator Framework Capabilities: 32b data 32b addresses no Ethernet Idle/Throttling support is available Virtual Hard Disk (VHD) support RAW disk and CD/DVD ROM support Asynchronous I/O support (Lock free asynchronous event queue) Asynchronous Clock support FrontPanel API Version 12 Host Platform: Compiler: GCC Apple LLVM 15.0.0 (clang-1500.1.0.2.5) Simulator Compiled as C (Release Build) on May 19 2024 at 13:01:35 Build Tool: autosetup+gmake Memory Access: Little Endian Memory Pointer Size: 64 bits Large File (>2GB) support SDL Video support: No Video Support No RegEx support for EXPECT commands OS clock resolution: 1ms Time taken by msleep(1): 1ms OS: Darwin Mac-mini.lan 22.6.0 Darwin Kernel Version 22.6.0: Mon Feb 19 19:43:13 PST 2024; root:xnu-8796.141.3.704.6~1/RELEASE_ARM64_T6020 arm64 Processor Name: Apple M2 Pro tar tool: bsdtar 3.5.3 - libarchive 3.5.3 zlib/1.2.11 liblzma/5.0.5 bz2lib/1.0.8 curl tool: curl 8.5.0 (aarch64-apple-darwin22.6.0) libcurl/8.5.0 OpenSSL/3.2.1 zlib/1.3.1 brotli/1.1.0 zstd/1.5.5 libidn2/2.3.7 libpsl/0.21.2 (+libidn2/2.3.4) nghttp2/1.58.0 git commit id: https://github.com/open-simh/simh/commit/ffe537a621d36bcf2e4d63b8b9ac3163929b57a9 git commit time: 2024-02-20T18:20:13-08:00

• how you built the simulator or that you're using prebuilt binaries

We successfully build from source cloned from https://github.com/open-simh/simh.git

• the simulator configuration file (or commands) which were used when the problem occurred.

No problems encountered.

• the expected behavior and the actual behavior

We propose internal changes to PDP8/pdp_8cpu.c to allow emulation of the Fetch/Defer/Execute major states which the PiDP-8/i hardware would be able to display, and control through the SING STEP front panel switch.

• you may also need to provide specific pointers to data files that may be necessary to demonstrate the problem

[markpizz]

I will be glad to take your changes and include it in the https://github.com/simh/simh repo. In all likelyhood, the results could then also be migrated to the https://github.com/open-simh/simh repo.

[eschaton]

I think the best thing to do here would be to:

1. Ensure your fork is rebased atop the latest Open SIMH, to remove any merge noise from its history;
2. Adjust/consolidate the commits in your fork to represent logical steps that can be easily reviewed; and,
3. Create one or more PRs against Open SIMH here to bring your changes into the main line of SIMH development.

People with personal forks can then pull your changes into them from Open SIMH, even before any PRs are merged if they want.

It would also probably be worth considering whether and how it would be possible to make the Major State simulation functionality optional (whether at build time or run time), in case anyone really does want PDP-8 simulation to run as fast as possible and doesn't need Major State support.

[poetnerd]

Thanks for the guidance, Chris.

The dev team is discussing the various trade-offs. At issue is that, as currently conceived, the inner loop of instruction execution changes in a fundamental way. So the way you turn it on or off is to basically switch between two fundamentally incompatible interpreter loops.

The simple way through, at this point, is to switch to Major State accurate, and take the 50% performance hit.

That said, we're going round once more to think if we can find some clever way to do allow switching between both in such a way as to minimize code bloat, and maximize performance.