R5K@PSUVMA.BITNET (01/22/87)
No, I haven't tried it yet, but would like to know if anyone has, and if it is compatable with all of the current software. I have all of the information on it and the autoconfig mods to make it autoconfig under 1.2 along with pictures of the setup. But I'm a bit shaky about trying it until I hear that there are no problems. Randy Katz PSUVMB
carolyn@cbmvax.UUCP (01/24/87)
In article <9797R5K@PSUVMA> R5K@PSUVMA.BITNET writes: >No, I haven't tried it yet, but would like to know if anyone has, >and if it is compatable with all of the current software. I have >all of the information on it and the autoconfig mods to make it >autoconfig under 1.2 along with pictures of the setup. But I'm >a bit shaky about trying it until I hear that there are no problems. If you're talking about one of the internal piggyback ram kits, I hear they are inexpensive and work pretty well. But here are a few warnings: 1. You'll void your warranty. 2. Many 1.1 programs don't work properly if your system has ANY non-CHIP ram because they have bit images defined in their code which by default will be loaded into FAST ram if FAST ram is available. (many of these programs can be fixed with "fixhunk" which is available on a Fish disk) 3. Piggyback ram is not as fast as real external FAST ram. 4. I don't know what type of "autoconfig" your setup has. Several developers have called me with problems caused by the software that unlinks the piggyback ram from the CHIP memory list. (at bootup the system sees it as CHIP mem). A program called "Make512K" is run to take the memory OUT of the Chip memory list. Then Addmem is used to add it to the FAST mem list. The version of Make512K that my developers had only adjusted the system's idea of TOTAL chip ram. It did not search for the big chunk of chip mem which was still tagged as being 512K bytes larger than it actually is. So any system function that is MemChunk oriented (AllocMem, LoadSeg, etc.), still think there is a chunk of chip mem >512K bytes in size. And keeps allocating from it after the real chip mem is used up (-->GURU). Simple test for this problem is to run Avail after configuring the piggyback memory. If your "Available" chip and fast are correct but your "Largest" chip chunk is way too large (like 512K too large) then you've got the problem. The following changes to Make512K should fix the problem. (This is theoretical of course, since I do not have piggyback ram) ---------------------------------------------------------------------- ... struct MemChunk *chunk; ... (add these lines after "this_Mem_Hdr->mh_Free -= diff;") /* Try to find the chunk that's too big */ for (chunk=this_Mem_Hdr->mh_First; (chunk)&&(chunk->mc_Bytes < 0x80000); chunk = chunk->mc_Next); /* If we found it, fix it */ if(chunk) chunk->mc_Bytes -= 0x80000; ... ---------------------------------------------------------------- -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Carolyn Scheppner -- CBM >>Amiga Technical Support<< UUCP ...{allegra,caip,ihnp4,seismo}!cbmvax!carolyn PHONE 215-431-9180 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=