Am 11.05.2011 15:48, schrieb dott.Piergiorgio:
Il 11/05/2011 11:34, Holger Veit ha scritto: o.SAGE, as in www.sageandstride.orgAn (interesting) case of same naming, indeed...Interesting because this can be a good starting point for the sorely lacking emulation of a non-multimedia, non-graphic machine based on the 68K (there are many 68K CPU emulator code, but I guess that having a plain 68000 system emulated is useful....)I' still working (on and off) on the Occam treatment of a pair of simh_$CPU prior of doing an actual CPU implementation (I'm toying around an idea of an hypotethical 6-bit CPU (think if Intel produced as intermediate step, a 6006 between the 4040 and 8008) then, fool around 8008 emulation ;) )The most dear thing on SIMH code is that is relatively easy to "plug in" diverse hardware emulation (IMVHO is not by chance that the main fork is the altairZ80 project, and I guess you known well how diverse was the S-100 hardware...) and with a graphical hardware emulation (not limited to screen graphic, but also LED and, aptly for the "trailing edge" SIMH emulates, the blinkenlighten always associated with the large 1950s to 1970s machines) the usefulness of SIMH trascend the field of emulation, and can be really a core package in didactics (I don't hide that will be really interesting the emulation of the mid-1970s microprocessors/controllers SBC trainers...) research and like fields (last but not least the digital archeology, as pointed by Kossow's efforts in running 1960s software under emulator environments)heh... now I realize that this discussion ought not to be confined to both of us, so I let you decide if this ought or not to be reposted on the ML and if so, please feel free to do so :)Best regards from Italy, dott. Piergiorgio.
Hi,To begin with the last section: I have no problems with reposting to the list, as it is beginning to be a general interest thing, maybe it is also a matter of the sage_dev list (provided your are a member). The whole discussion thread being started is a bit about the future of simh.
I already mentioned MAME/MESS as another emulation project which is better known, and which provides more emulators, also and mainly for the microcomputer era which is so far only scratched by the AltairZ80 and the upcoming m6800 and m68k emulators. I don't like the MAME/MESS horde, though, for some reasons, so I rather spend my time for simh instead to make it better than just the
frequently used PDP and VAX simulation platforms.simh, IMHO, was built around the mini-computer paradigm of having separate boards in a cabinet, each being one or more entirely isolated device(s), that exclusively communicate through a general bus, but with the exception of a general DMA concept, not between each other. This makes it difficult for emulating more closely coupled microcomputer systems, because a strict separation of devices does no longer exist. The FDC is a separate device, but it is connected not to daisy-chained interrupts on something like a QBUS, but communicates with a more or less programmable interrupt controller (e.g. an i8259A PIC). The difficulty to simulate this is that the PIC itself behaves as a device to the CPU, but it glues together other devices. Similarly this happens to a programmable DMA controller. The FDC of the mentioned SAGE does not, in contrast to the reference FDC in the AltairZ80, use DMA, but polled I/O, so that lego block turned out to be non-fitting. The hard disk controller of the SAGE machine is not monolithic, but consists of various TTL and LSI circuits, some of which (like 8253 counters) are programmable and also triggers interrupt to an own slave PIC connected to a cascading pin of the primary PIC.
The fundamental architecture of such microprocessor systems is that there is a CPU and a number of programmable circuits more or less connected by glue logic. Despite of the CPU address, data and control buses, there is no clearly separatable bus concept that allows for a "a board is a device" model. In other words: having the Altair emulator won't help writing another Z80 emulator unless it also is an S100 one; you might be able to lift the sim_instr code, but essentially rewrite a completely new Z80 simulator with its specific glue logic which may or may not be devices or units. It is not a solution to cut the system in a way that DMA or IRQ logic is regarded as a part of the CPU or a part of a device that needs such services, because this even further specializes the component set rather than generalizing it. It is already a mess in the existing families, like PDP or HP2100, to deal with variants of different instruction sets, address busses, memory sizes and layouts, and model-specific quirks. This does not really contribute to good maintainability by others than the original code writer.
Provided, users and developers aren't just satisfied with basically enjoying VMS in a box on a Windows PC, the interesting (research) question to me is what extensions to simh are useful to assist a developer in supporting the microcomputer era. Some weeks ago, GIANO was mentioned as derivation into such a direction, and maybe there is something to learn there. What I want to avoid, though, is that simh will split into as many variants as BSD or Linux distributions because the simh developer base is by magnitudes smaller than the former ones. This might however happen as the set of now suitable computer systems with sufficient documentation and software to try is unfortunately limited, and worse, the developer base who still knows and understands valve-based SAGE computers, for instance, tends to age.
-- Holger
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ Simh mailing list Simh@trailing-edge.com http://mailman.trailing-edge.com/mailman/listinfo/simh
