This is in relation to JDK-8059234. If we instrument C1 NCE and run Nashorn/Octane benchmark suite, and then sort the methods by compile time, we can then observe this:
 http://cr.openjdk.java.net/~shade/8059235/nashorn-c1-nce.log.gz

The last lines there are most interesting:
 34662 1541.935 ms, 11911 blocks: Done with null check elimination for method jdk/nashorn/internal/scripts/Script$Recompilation$2839$768791A$mandreel::_ZN16btCollisionWorld18addCollisionObjectEP17btCollisionObjectss(Ljdk/nashorn/internal/runtime/ScriptFunction;Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;

That is, a sample method took 1.5 seconds to do the a single phase in C1; and that is because it is a mammoth method with almost 12K basic blocks. While we can address the issue partially within JDK-8059234, Nashorn really needs to split the work into smaller methods to play nicely with JITs.