Are there any principles guiding versioning? Would such be useful? (I think it would help if/when multiple vendors provide different versions that the versioning would be consistent and that these versions would not be serial numbers but contain general information useful for ordering or investigating parts.) E.g., it might be useful to have a simple distinction between 32-bit and 64-bit implementations (say, 3rd least significant digit is even or odd, respectively), between having vector/FP registers and not having them, between supporting all instructions and only required instructions, between having only standard operations and having custom instructions. Reserving zero digits for undetermined/default/generic versions might be useful (removing the need to use 'x'). If one uses even/odd distinctions, it might also be useful to define one of each as generic. The presence of vendor/implementation-specific extensions might also be indicated by a human-friendly versioning method. (It is also not clear if OR1000 is a subset of a larger family of mostly binary compatible processors. [Earlier OR2000 was presented as a radically different architecture, but there could be advantages to having the more significant version numbers indicate signficant changes in pipeline, cache consistency model, etc.]) Paul A. Clayton -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.opencores.org/forums/openrisc/attachments/20040921/59864219/attachment.htm

Very simple, all this can be accomplished with VR register. VR[VER] = 0x12 for OR1200, other OR1000 implementations would have 0x11, 0x13, 0x14 etc. There is OR1500 which, I hope, uses VR[VER] = 0x15. VR[CFG] configuration template is basically replacement for all UPR and CFGR registers. Let say you don't want to have all those CFGRs and UPR implemented but you still want your software be able to run on OR1200 with caches and MMUs and on OR1200 configured without caches/MMUs. So you would have two different VR[CFG] templates defines. Currently this is left to a vendor to define, the appropaite would be that for every implementation there is a predefined list. Currently there is no such for OR1200. VR[REV] Same implementations can go through different revisions, for example if there are bug fixes. I hope thios is nice and easy and simple to use. regards, Damjan ----- Original Message ----- From: Dysthymicdolt@aol.com> To: openrisc@opencores.org> Sent: Tuesday, September 21, 2004 2:11 PM Subject: [openrisc] Versioning Principles?


--------------------------------------------------------------------------------

damjanl@opencores.org wrote: I was referring to the existence of any principles for generating version numbers from a marketing/conceptual perspective. ISTM that it would be convenient for human communication to have consistent, _systematic_ versioning. Paul A. Clayton -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.opencores.org/forums/openrisc/attachments/20040923/618971a4/attachment.htm