SuperComputer Archives

I’ve ordered a Samsung Galaxy S3. I don’t know if it is the right thing to do. Part of me says moving to a new phone means that you should be moving to a significantly better device – I have an S2. Every new phone seems to be “just an iteration” of the incumbent spec. Certainly that’s how it seems with the iPhone and also with other devices that Samsung has brought out since the S2. I’m still on the original iPad!

On the other hand we live in a very fast moving world where even the smallest competitive edge can make a real difference. Can we ascribe this to our use of smartphones? I don’t know, perhaps. I think the S3 probably has just enough on the S2 to make it worth the upgrade. I need to do it because otherwise before I know it everyone else will have moved on to the S7 or S8 and I’ll be so far behind the times I will really have to think of packing it all in.

Some time ago I wrote that the laptop of today is the SuperComputer of 12 years ago. The SuperComputing community is constantly pushing forth the boundaries. Of course these computers need to talk to each other – that data has to go somewhere.

Because of this that research community is also having to push the envelope on data connectivity speeds and a team from Caltech and the University of Victoria has just demonstrated 186Gbps data transfer over a 100Gig connection (the sum of data speeds – both directions obviously).

OK this is all good stuff but so what you say? Well just like the SuperComputer of today is the laptop of 12 years hence there will come a time when 100Gigs is going to be mainstream for home and business connectivity. There is no point in hitting me with arguments suggesting otherwise. You are wrong 🙂

I don’t know what these data rates are going to be used for but used they will be. I am going to bookmark this day in 2023 to write a blog post reviewing progress towards this goal. By that time my laptop will be able to do 10.51 petaFLOPS (the current fasted SuperComputer is The K computer – it consists of 88,000 SPARC64 VIIIfx CPUs, and spans 864 server racks).

This is why I occasionally mention that really the only sensible long term investment for broadband speed infrastructure is fibre because in 12 years time I will need something that can handle the output of all those petaFLOPS. Who can argue with that?