[coyotos-dev] Accept GCJ into the tool chain?
Jonathan S. Shapiro
shap at eros-os.com
Mon Nov 27 19:43:06 CST 2006
On Mon, 2006-11-27 at 14:45 -0800, Tracy R Reed wrote:
> Jonathan S. Shapiro wrote:
> > In short, gcc/g++ win against Haskell on both memory consumption and
> > performance.
> I would rather have slightly slower code that works than no code at all
> or incorrect code.
I tend to agree, but "slightly slower" should certainly mean "within a
factor of two". Even on these fairly contrived benchmarks, we aren't
seeing a factor of two in far too many of the cases.
It will be very interesting to see how BitC does when it stabilizes.
More information about the coyotos-dev