pawn@wpi.wpi.edu (Kevin Goroway) (03/25/89)
I finally got my 5.02 in the mail and installed it on my Hard drive...now comes the question(s)... When compiling an old program (worked w/ v. 4.0) I get a Blink Error (are these described anywhere in the new manuals?) Error 504: _custom symbol - Distance for Data Reloc16 > 32768 etc... humph...what did I do wrong that v 4.0 didn't find? Question 2... What is the ABSOLUTE BEST WAY to get SPEED out of a program that is very math intensive, and only needs to run on my 2500? I've tried compiling normally and then linking with the 68881 lib and the math lib, but this actually doesn't work whereas the ieee lib/math lib combo works ... also what about the compiler options? -f8? -m2? Thanks in advance!!! -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= | Pawn@wpi.wpi.edu Worcester Polytechnic Institute, Mass, U.S.A. | | Pawn@wpi.Bitnet main() { printf("Hello World!\n); } | =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
kevin@uts.amdahl.com (Kevin Clague) (03/27/89)
In article <1512@wpi.wpi.edu> pawn@wpi.wpi.edu (Kevin Goroway) writes: >I finally got my 5.02 in the mail and installed it on my Hard drive...now >comes the question(s)... > >Question 2... >What is the ABSOLUTE BEST WAY to get SPEED out of a program that is very >math intensive, and only needs to run on my 2500? > >I've tried compiling normally and then linking with the 68881 lib and the math lib, but this actually doesn't work whereas the ieee lib/math lib combo works >... >also what about the compiler options? -f8? -m2? Yes. Using these options will get you inline 68020/28881 code. You must also link with the lcm881.lib. In theory this should work. . . but I was up 'till midnight last night with Tom Rokicki shooting a bug we found when using sscanf. The stack frame was getting trashed and a5 was ending up with zeros in it after returning from a subroutine. It got too late for us to be sure whether it was my code, but my code works fine with FFP, and IEEE. We were using 5.0 (not 5.02) Using Manx 5.0, a 2000 with a 2620, and the same program (3D HAM fractals) got me an 8X performance increase over my 2000 using FFP. > >Thanks in advance!!! > > >-- >=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= >| Pawn@wpi.wpi.edu Worcester Polytechnic Institute, Mass, U.S.A. | >| Pawn@wpi.Bitnet main() { printf("Hello World!\n); } | >=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Good luck. Kevin -- UUCP: kevin@uts.amdahl.com or: {sun,decwrl,hplabs,pyramid,seismo,oliveb}!amdahl!kevin DDD: 408-737-5481 USPS: Amdahl Corp. M/S 249, 1250 E. Arques Av, Sunnyvale, CA 94086 [ Any thoughts or opinions which may or may not have been expressed ] [ herein are my own. They are not necessarily those of my employer. ]