> I am looking into PICmicro chips (integrated cpu, some memory and lots of > read/write pins) and was thinking of building a `joynet accelarator' with > it: it should just buffer all reads and writes until they are used. This > way you can increase the speed of a joynet link (or a different type, if > you like), because the PIC will always immediately send the ack, as > opposed to the msx that always has to poll for data. Since joynet can also > be used to connect to other computers than msx, this could be an > alternative. Obviously, being a 1-bit protocol, it is quite slow compared > to 4 or 8 bit parallel transfers.
It won't be JoyNet anymore, then. The advantages of JoyNet are simplicity and cheapness. If you want more, it's better to just use an RS232-interface which offers higher rates and easier control (interrupts, bytes, etc), and also is easier to link with PC computers. If you want a network, that would be very nice. That can indeed not be done with RS232 unless you use some expensive additional hardware. Just don't call it JoyNet, because it isn't. By the way, I don't think it would speed up the transfer process. In general, joynet transfers work like this: - one MSX wants to contact another, and is put on halt until the other replies or a timeout occurs. - the peer polls from time to time (every interrupt is acceptable for a data network and the poll itself doesn't take a lot of time at all, so you can put it on faster interrupts like line or music interrupts) until the transfer request bit is set. - an entire package of data is sent in one go at max speed (this can be several kb's in size). So the PIC would have to be able to buffer the entire package. The fifo has to be quite large if you want it to be useful. I don't think PIC chips (at least the cheap ones) have much storage capacity. Unless you plan some streaming application, ofcourse. > That reminds me I should still write my ideas of the protocol down. Is > anyone still interested in them, anyway? Is anyone using joynet in a > program, or planning to use it? The existence of this accelerator-thing > makes the algoritm for reading/writing a bit different as well. As far as JoyNet is concerned, I don't think anyone is planning to use it at the moment. But, it is a standard, and if people want to use it, the hardware design is there, ready to use. I myself still want to program a JoyNet application someday. I have good ideas on how to program it. However, I never get around to actually *do* it... And ofcourse you should share your ideas. At the least it will trigger some interesting discussion, and maybe some nice network adapter thingie will grow out of it. Just fool around with it, I suggest. I for example would like to know what exactly you had planned, and if you have any more ideas. ~Grauw -- For info, see http://www.stack.nl/~wynke/MSX/listinfo.html
