Re: Interpreters for VLIW

Jim Granville <jim.granville@designtools.co.nz>
18 Jan 2001 00:53:31 -0500

          From comp.compilers

Related articles
Interpreters for VLIW bonzi@pc-amo3.elet.polimi.it (2001-01-11)
Re: Interpreters for VLIW anton@mips.complang.tuwien.ac.at (2001-01-18)
Re: Interpreters for VLIW jim.granville@designtools.co.nz (Jim Granville) (2001-01-18)
Re: Interpreters for VLIW Xavier.Leroy@inria.fr (Xavier Leroy) (2001-01-18)
Re: Interpreters for VLIW germana.ricci@inwind.it (2001-01-20)
| List of all articles for this month |

From: Jim Granville <jim.granville@designtools.co.nz>
Newsgroups: comp.compilers
Date: 18 Jan 2001 00:53:31 -0500
Organization: paradise.net.nz customer
References: 01-01-070
Keywords: VLIW, optimize
Posted-Date: 18 Jan 2001 00:53:29 EST

Paolo Bonzini wrote:
>
> I have compiled GNU Smalltalk on the Itanium and found it to be
> horribly slow --- Less than half the speed of an Intel chip with the
> same clock! The reason is, the basic blocks in the interpreter are
> too small for GCC to do nice instruction scheduling. Does anyone know
> of pointers to papers on optimizing interpreters for VLIW
> architectures?


  I thought Intel was offering a fancy 'instruction shuffling' compiler
for this chip - have you tried that ?


  Was the same level of optimise done on both referance platforms ?


  Cache Sizes ?
On these high end MPU, there are interesting effects with interpreters
and Cache size.


  If they have to start from < 50% on equivalent code, it does not look
good for Itanium.


  Can you try a similar benchmark on the Crusoe - that would be
interesting:-)


  - jg


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.