abair@turbinia.sps.mot.com (Alan Bair) (11/30/90)
I have a 3000 with just the 2MB of memory, 1MB chip + 1MB fast. Since the OS is loaded into fastmem, I only have about 520K of fastmem that is available for use. With a minimal WorkBench setup there is only about 900K of chipmem left. I plan on upgrading by adding 4MB of fastmem and moving the 1MB of current fastmem to the chipmem section. However, I need a larger continuous section of memory NOW. I am trying to use GNU emacs and GCC, either of which needs LARGE chunks of memory. So I have some questions about how to solve my current problem until I get the extra 4MB. 1. I can run nofastmem, which allows emacs to run by going to the chipmem, but I appear to be loosing the .5MB of fastmem. The workbench titlebar and avail indicate that after running nofastmem, that the fastmem is just turned off and no longer available. Is this what is and should be happening or am I misinterpreting the displays? This does work, but not very optimal, since I loose the .5MB. 2. I could move all the memory to the chipmem area, but will the 3000 still boot and run in this configuration? I know it will be slower since there will be no fastmem, but I will at least have 2MB of continuous memory now, though it will be short by .5MB for the OS. This will give me a single larger chunk of memory as compared to option #1. 3. Is there some other software method to join the chip and fast memory areas into a single logical chunk of memory? I looked at SetCPU, but I don't think it will do what I want. Any suggestions or hints would be appreciated. (I will even accept information about good sources for 1Mx4 80ns static column DIPs) -- Alan Bair SSDT (formerly SPS CAD) Motorola, Inc. Logic Simulation & Test Austin, Texas abair@turbinia.sps.mot.com
lphillips@lpami.wimsey.bc.ca (Larry Phillips) (12/01/90)
In <ABAIR.90Nov29224959@turbinia.sps.mot.com>, abair@turbinia.sps.mot.com (Alan Bair) writes: >However, I need a larger continuous section of memory NOW. I am trying to use >GNU emacs and GCC, either of which needs LARGE chunks of memory. So I have >some questions about how to solve my current problem until I get the extra 4MB. > >1. I can run nofastmem, which allows emacs to run by going to the chipmem, but > I appear to be loosing the .5MB of fastmem. The workbench titlebar and > avail indicate that after running nofastmem, that the fastmem is just > turned off and no longer available. Is this what is and should be > happening or am I misinterpreting the displays? This does work, but > not very optimal, since I loose the .5MB. This is what NoFastMem does. It is not a directive to not use the fast mem, but rather a switch to turn it off and make it unavailable. >2. I could move all the memory to the chipmem area, but will the 3000 still > boot and run in this configuration? I know it will be slower since > there will be no fastmem, but I will at least have 2MB of continuous > memory now, though it will be short by .5MB for the OS. This will give > me a single larger chunk of memory as compared to option #1. The 3000 will not (currently) boot in an all CHIP configuration. The reason for this is that the KS code is hardwired to load into a high address in FAST memory. >3. Is there some other software method to join the chip and fast memory areas > into a single logical chunk of memory? I looked at SetCPU, but I don't > think it will do what I want. Not that I know of. >Any suggestions or hints would be appreciated. (I will even accept information >about good sources for 1Mx4 80ns static column DIPs) As a temporary measure, you might want to go with 2 megs of FAST memory in the 256*4 package. It's cheaper in the short run, but probbaly more expensive in the long run, when you sell them to go to the 4 meg parts. -larry -- The only things to survive a nuclear war will be cockroaches and IBM PCs. +-----------------------------------------------------------------------+ | // Larry Phillips | | \X/ lphillips@lpami.wimsey.bc.ca -or- uunet!van-bc!lpami!lphillips | | COMPUSERVE: 76703,4322 -or- 76703.4322@compuserve.com | +-----------------------------------------------------------------------+