I was revisiting some of Europe's recent supercomputing centers' achievements. It turns very interesting to see how some of them started to claim the achievements of their users -that is to say external users from another institution, that apply for computing power.
This is indeed a very dangerous problem for newly formed centers: it does not reflect any of their own research! of none of its researchers. To be very clear: these centers have huge costs but research function is not working as it should. Scientific representants in the government or science policy managers which are really pressed by IMF will have to justify this.
Since the begining of this blog I have claimed that this is not the way to go. Today, unfortunately these centers know that they may become target of bailouts.
What makes it even more exceptional is: why when we had money, we though that competing for the TOP500 would give us scientific competitive advantage ? Just for having an F1 chassis, we can run in the F1? Buying computing parts is something ANY government can do.
I believe that in the forthcoming years, we may see some of these baby places merging or disappearing in favour of places that DO real science or engineering. The rest only need one thing: extremely efficient management with large efficiency ratios for their systems. IMF will shutdown the rest.
This is indeed a very dangerous problem for newly formed centers: it does not reflect any of their own research! of none of its researchers. To be very clear: these centers have huge costs but research function is not working as it should. Scientific representants in the government or science policy managers which are really pressed by IMF will have to justify this.
Since the begining of this blog I have claimed that this is not the way to go. Today, unfortunately these centers know that they may become target of bailouts.
What makes it even more exceptional is: why when we had money, we though that competing for the TOP500 would give us scientific competitive advantage ? Just for having an F1 chassis, we can run in the F1? Buying computing parts is something ANY government can do.
I believe that in the forthcoming years, we may see some of these baby places merging or disappearing in favour of places that DO real science or engineering. The rest only need one thing: extremely efficient management with large efficiency ratios for their systems. IMF will shutdown the rest.
No comments:
Post a Comment