johnk@grebyn.COM (John Kennedy) (02/08/89)
Microport talks a lot in their manuals about special consideration for the 286's special hardware; they talk about not making assumptions concerning contiguity of memory acquired by _malloc_. They talk about not using _sbreak_, etc. Does the pathalias source, as distributed, contain any calls that would upset the System V/AT way of doing things? When I run pathalias with more than a trivial number of files as input, I get the message: pathalias: Out of memory (-88473600k allocated) pathalias: cannot recover! Does this happen with other users? Originally I got this error when processing a much smaller number of input files, but after recompiling with the large model, I was able to add more input, but not much more. Please email any suggestions. Thanks in advance, John
karl@sugar.uu.net (Karl Lehenbauer) (02/09/89)
In article <11802@grebyn.COM>, johnk@grebyn.COM (John Kennedy) writes: > Microport talks a lot in their manuals about special consideration > for the 286's special hardware; they talk about not making assumptions > concerning contiguity of memory acquired by _malloc_. They talk about > not using _sbreak_, etc. > Does the pathalias source, as distributed, contain any calls that would > upset the System V/AT way of doing things? That's right. Pathalias needs a lot of memory and expects it to be contiguous. steve@nuchat was running it on a 286 with, I think, some hacks he did to implement quasi-virtual memory through the disk. It took like eight hours to run the maps for North America alone. A 386 can handle the job, no problem. 'sugar' can crunch the maps in about five minutes... -- -- uunet!sugar!karl | "Everyone has a purpose in life. Perhaps yours is -- karl@sugar.uu.net | watching television." -- David Letterman -- Usenet BBS (713) 438-5018
bob@obiwan.UUCP (Bob Willcox) (02/10/89)
In article <3404@sugar.uu.net> karl@sugar.uu.net (Karl Lehenbauer) writes: >In article <11802@grebyn.COM>, johnk@grebyn.COM (John Kennedy) writes: >> Does the pathalias source, as distributed, contain any calls that would >> upset the System V/AT way of doing things? > >That's right. Pathalias needs a lot of memory and expects it to be contiguous. >steve@nuchat was running it on a 286 with, I think, some hacks he did to >implement quasi-virtual memory through the disk. It took like eight hours >to run the maps for North America alone. > >A 386 can handle the job, no problem. 'sugar' can crunch the maps in about >five minutes... I am running a hacked-up pathalias on my 286 which I run regularly to build the entire world maps. It takes some-what more than 1 meg of memory to run, but runs in about 10 minutes (on a 10mhz AT clone). It is based upon a copy of version 9.1 of pathalias that was ported to Xenix by Greg Laskin. I, in turn, ported it to Microport System V/AT. It is kind of a hack, but if you have enough memory seems to run fine. -- Bob Willcox ...!{rutgers|ames}!cs.utexas.edu!romp!obiwan!bob Phone: 512 258-4224