🚀 go-pugleaf

In article <1992Aug3.183428.22522@sbcs.sunysb.edu>, kskapin@csws8.ic.sunysb.edu (Keith T Skapinski) writes:

| Buy the 386DX-40 rather than the 486SX-25. The 386DX-40 is faster than the
| 486SX-25 as documented in many trade magazines. Since the 486SX doesn't
| come with a built in math coprocessor like the 486DX, the 486SX has
| no real advantage over the 386DX-40 other than it is cheaper.

  The 386DX40 is faster than a 486SX20, but I would question the 25. In
16 bit mode (MS-DOS) the 486 is roughtly twice as fast as the 386 at the
same speed. Some of this is due to cache in chip, some to burst mode
memory access. Adding external cache reduces the advantage.

| I believe the norton SI rating (v.6) for a 386 DX 40 is about 47.
| I dont know what it is for a 486SX off hand but I know it is slower
| according to magazines which published the ratings for the CPUS.

  I believe you're talking about the 486SX20, not 25, based on some
benchmarks I've run. These were real apps, not SI, so if you run SI a
lot you may be better off with the 386 ;-)

  Note that 32 bit apps, such as UNIX, windows-NT, etc, see roughly a
2.6 improvement at the same speed. This is because the compilers make
use of the full memory bandwidth.

  All that said, I'd buy the 386, because (a) it's a lot more cost
effective, and (b) because it doesn't reward Intel for making the 486SX
have a different pinout than the DX. Also consider getting the new
486DLC (part #??) which goes in a 386 socket. Buy a bare board and put
in the CPU yourself. The baby 486 only has 1k cahce instead of 8, no
FPU, and no burst mode (because 386 has no burst mode). In spite of that
it is quite a step up from the 386 at the same speed.

  Should be shipping this month, there's also a part which goes in an SX
socket.
-- 
bill davidsen, GE Corp. R&D Center; Box 8; Schenectady NY 12345
    I admit that when I was in school I wrote COBOL. But I didn't compile.