/sys/doc/ Documentation archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Lucent Technologies & Sun Microsystems

> As for the claimed performance penalty for Java, implementing the VM in
> silicon will negate all but the intrinsic language overhead.

Neither bytecode-VMs nor portable languages are new ideas.  They go back to
the 60s, if not further.  One thing that has remained true, however, is
that they have never performed adequately.  If the solution you propose was
easy, we'd already have it.  

Put another way, making fast Java-specific silicon seems easy when you're
comparing to 1st generation VMs running on platforms that suffer all sorts
of Java related penalties (e.g., le x86).  We have yet to see anyone,
including Sun, ship Java silicon that has performance comparable to a fast
mainstream CPU running that CPU's native code.

RSR