Ben,
this is probably gcc 2.95.3 generating bad code for x86 targets.
I've seen it happen myself.
For an explanation and remedy see the bottom of the CHANGES file:
Warning:
The function KneserNey::lowerOrderWeight() seems to trigger a compiler
bug in gcc 2.95.3 with optimization, on the i686-pc-linux-gnu,
i386-pc-solaris2, and i686-pc-cygwin platforms. Compile
lm/src/Discount.cc with OPTIMIZE_FLAGS=-O1 if the "ngram-count-kn-int"
test in the test/ directory does not terminate.
--Andreas
In message <000f01c1ba6d$38786360$e100a8c0@BENJ0EJR796N2M>you wrote:
> After installing SRILM v1.3, the ngram-count-kn-int test is hanging. Any
> idea why? All prior tests are finishing successfully (class-ngram disambig
> hidden-ngram hidden-ngram-nbest lattice-expansion multi-ngram
> ngram-count-abs ngram-count-gt ngram-count-kn) and then ngram-count-kn-int
> fails. I haven't yet run all tests.
>
> Environment is as follows:
> OS: Cygwin, running under Win2000 Professional
> Memory: 512kb
> Hardware: MicronPC TransPort XT2
>
> ______________________________________
> Ben Reaves b.reaves at ieee dot org
>
>
Click here to go to the SRILM home page.