On Mon, Jan 29, 2001 at 12:27:08PM -0500, Jeremy Hylton wrote:
Depends on your defintion of "compiler problem" <wink>. If you mean, it compiles our code so it runs slower, then, yes, we've got one :-).
Compiling with gcc and -g, with no optimization, 2.0 and 2.1cvs seem to be very close, with 2.1 slightly slower: 2.0: Pystone(1.1) time for 10000 passes = 1.04 This machine benchmarks at 9615.38 pystones/second This machine benchmarks at 9345.79 pystones/second This machine benchmarks at 9433.96 pystones/second This machine benchmarks at 9433.96 pystones/second This machine benchmarks at 9523.81 pystones/second 2.1cvs: Pystone(1.1) time for 10000 passes = 1.09 This machine benchmarks at 9174.31 pystones/second This machine benchmarks at 9090.91 pystones/second This machine benchmarks at 9259.26 pystones/second This machine benchmarks at 9174.31 pystones/second This machine benchmarks at 9090.91 pystones/second Would it be worth experimenting with platform-specific compiler options to try to squeeze out the last bit of performance (can wait for the betas, probably). --amk