es1@cunixb.cc.columbia.edu (Ethan Solomita) (11/11/90)
I ran Dhrystones 2.1 on a 3000/25 with 68030 specific options (and 881 for the printf's %f) using non-register args and the optimizer. I got 5263 dhrystones. Does this sound right? I thought people here were claiming 7000? Anything I should be doing differently? Also, I had a hard time making heads or tails out of what should be done with the header file. I defined TIME and assumed 60Hz. -- Ethan Ethan Solomita: es1@cunixb.cc.columbia.edu GorbachevAwards++; free (SovietUnion); IndependentRepublics += 15;
eachus@linus.mitre.org (Robert I. Eachus) (11/16/90)
Are you using SetCPU to turn on the data cache in the 68030? (I don't think you need to use FASTROM.) This sounds about like the right difference. (Under 1.3 the data cache is off by default, since with some DMA controllers it could cause problems, but I think I remember reading that the 3000s disk controller and hddisk.device do the "right" things. (I have a 2090 and a 2630 card and I have never had a problem caused by the data cache. SetCPU leaves both data and instruction caching off for the CHIP memory. -- Thanks Dave :-) -- Robert I. Eachus with STANDARD_DISCLAIMER; use STANDARD_DISCLAIMER; function MESSAGE (TEXT: in CLEVER_IDEAS) return BETTER_IDEAS is...
es1@cunixb.cc.columbia.edu (Ethan Solomita) (11/16/90)
In article <EACHUS.90Nov15140452@aries.linus.mitre.org> eachus@linus.mitre.org (Robert I. Eachus) writes: > > Are you using SetCPU to turn on the data cache in the 68030? (I >don't think you need to use FASTROM.) This sounds about like the >right difference. (Under 1.3 the data cache is off by default, since >with some DMA controllers it could cause problems, but I think I >remember reading that the 3000s disk controller and hddisk.device do >the "right" things. (I have a 2090 and a 2630 card and I have never >had a problem caused by the data cache. SetCPU leaves both data and >instruction caching off for the CHIP memory. -- Thanks Dave :-) > I have the cache on, but burst off because I don't have static column ram, just burst. I've gotten the numbers up just above 6,000 (6,060 to be exact) by changing a few more parameters and by using changetaskpri to give it priority. Two things. When I add forbid() it won't compile and when I try compiling with registerized parameters it won't compile, I get unknown symbols when it links, including things like _strcpy. Does anyone know how to make either of those changes work? >-- > > Robert I. Eachus > >with STANDARD_DISCLAIMER; >use STANDARD_DISCLAIMER; >function MESSAGE (TEXT: in CLEVER_IDEAS) return BETTER_IDEAS is... -- Ethan If God were suddenly condemned to live the life which he has inflicted upon men, He would kill himself. -- Alexandre Dumas Which is it: is man one of God's blunders, or is God one of man's blunders? -- Nietzsche They're JOKES! Lighten up! 8) Flames will be ignored.