Problem with vmargs.exit
skumars at earthlink.net
Wed Oct 21 15:18:19 PDT 1998
Godmar Back wrote:
> > seems to just go away some of the times. Is it possible that
> > they have been GC'ed?
OK, thanks. What you say is true. I was doing something
silly in my function, and as a result was not getting
I have some interesting results that I would like
to share. I can run my version of Kaffe in 4 modes.
1. pure JIT (default)
2. pure interpreter (-disableJIT)
3. Some user specified interpreted methods, rest JIT (-mixSomeJIT)
4. Mixture on internal heuristics. (-mixJIT)
The timing results are as follows, where in;
case 3: I interpret main and all <clinit>()V and all <init>()V
case 4: I JIT any method that is called the second time.
I know of a much better internal heuristic, which I will
be implementing very soon. I expect the run time
to reduce dramatically then.
> time Kaffe pizza.compiler.Main hello.java
10.49user 0.55system 0:11.08elapsed 99%CPU (0avgtext+0avgdata
0inputs+0outputs (167major+3610minor)pagefaults 0swaps
> time Kaffe -disableJIT pizza.compiler.Main hello.java
11.70user 0.62system 0:12.32elapsed 99%CPU (0avgtext+0avgdata
0inputs+0outputs (161major+3500minor)pagefaults 0swaps
> time Kaffe -mixSomeJIT pizza.compiler.Main hello.java
10.12user 0.70system 0:10.83elapsed 99%CPU (0avgtext+0avgdata
0inputs+0outputs (175major+3613minor)pagefaults 0swaps
> time Kaffe -mixJIT pizza.compiler.Main hello.java
8.85user 0.63system 0:09.48elapsed 100%CPU (0avgtext+0avgdata
0inputs+0outputs (175major+3588minor)pagefaults 0swaps
More information about the kaffe