This talk ...

• This talk ...

• takes a QCD practitioner's point of view: (Let's build a QCD machine)

• concentrates on cluster computers

• can certainly not consider all the activities carried out worldwide

• can only touch in short APE and QCDOC

• will not discuss commercial systems

• reflects the speaker's personal taste !

• I thank all the people who have sent contributions. I apologize for not being able to present all the important material.