Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!husc6!cs.utexas.edu!uunet!microsoft!brianw From: brianw@microsoft.UUCP (Brian Willoughby) Newsgroups: comp.sys.apple Subject: Re: UDC problems (long) Summary: Getting closer to a solution Message-ID: <7527@microsoft.UUCP> Date: 27 Aug 89 22:01:17 GMT References: <8908220011.aa16711@SMOKE.BRL.MIL> <7500@microsoft.UUCP> Reply-To: brianw@microsoft.UUCP (Brian Willoughby) Organization: Microsoft Corp., Redmond WA Lines: 73 In article jac@paul.rutgers.edu (Jonathan A. Chandross) writes: >I don't have a UDC, but from your description I think I know what >might be causing the problem. If I understand correctly, the UDC >disables the on-board 6502 using RDY and uses DMA to store the >result of a disk operation directly into memory. > Small correction: the DMA signal is not used - I have monitored the bus and it doesn't go active (low). Of course, it isn't incorrect to refer to the use of RDY as "DMA", because Direct Memory Access is what is happening, but I just didn't want any confusion about the actual DMA signal on the Apple Bus. Besides, I have a TransWarp, and I can't seem to get Applied Engineering to say whether or not it supports DMA accesses to its on-board RAM. I also haven't gotten any response on whether the TransWarp uses DMA when it is writing to motherboard video RAM. But, from observation, none of my peripherals use DMA. >The problem might be slow motherboard DRAM. I've run into this >problem on other computers. As the DRAM warms up, it's access >speed slows down. If the speed was marginal before (ie the DRAM >was just fast enough to work) you will start seeing some flakiness. > Thanks, that's a good idea. But, for me, I'll have to check on the speed of the TransWarp card RAM, because my problems occur whenever the TransWarp is active. Again (and I'm only re-iterating this in the hopes that someone else might have a few ideas), the W65C802 *ignores* RDY during a write when it is executing in Emulation mode. This "feature" causes the UDC to crash in a very repeatable way when I place the C802 on my II Plus motherboard (without the TransWarp active), because the UDC makes sure that the C802 is in Emulation mode. I've basically accepted that it is impossible to use a 6502 or W65C802 on the II Plus motherboard with the UDC (although reports from this net seem to indicate that some versions of the UDC function with this configuration), because this behavior matches the documentation on the processors. The TransWarp, however, throws a monkey wrench into my theories. Somehow, the UDC works with the W65C802 on my TransWarp card, but only until the computer heats up (I know I'm repeating myself, but bear with me :-). For this combination to function, the UDC *must* be switching the C802 into Native mode, where RDY *is* functional during a write. The only difference I can find between the C802 crashing (on the II Plus motherboard) and executing properly (on the TransWarp) is that the TransWarp adds a few //e features to my II Plus that aren't available without the TransWarp. This includes 128K of RAM (main and aux) and the alternate zero page and stack features. I *assume* that the UDC ROM code is recognising the differences with the TransWarp active, and deciding to put the C802 in Native mode before using the RDY signal. To make my point (finally, you're saying :-) I HAVE NOT FOUND THE LOCATION IN THE UDC ROM CODE WHICH ALLOWS THE C802 TO FUNCTION WITH RDY. I also have not figured out why this magical feature disappears after extended use. Any ideas at all are welcome, and if any UDC owner is fatally curious enough to read through all the ROM code, then let me know what you find. >Don't forget that you lose some memory bandwidth to the refresh >circuitry. This couple with slow chips might just be enough to >cause the strange behavior you cite. > We're talking about a II Plus here: motherboard RAM is not slowed by refresh, because video accesses accomplish refresh, and DMA is only allowed during CPU cycles (can't conflict with video). Refresh may be a problem with the TransWarp card RAM, however. Too bad AE or Video Technology doesn't have anyone listening to this group (and responding, that is). > >Jonathan A. Chandross Brian Willoughby UUCP: ...!{tikal, sun, uunet, elwood}!microsoft!brianw InterNet: microsoft!brianw@uunet.UU.NET or: microsoft!brianw@Sun.COM Bitnet brianw@microsoft.UUCP