By: etudiant (etudiant.delete@this.msn.com), February 2, 2017 9:39 am
Room: Moderated Discussions
Ireland (boh.delete@this.outlook.ie) on January 31, 2017 4:53 pm wrote:
> juanrga (nospam.delete@this.juanrga.com) on January 29, 2017 4:42 am wrote:
> >
> > It seems you also missed the announcement of Isambard,
> > an ARM-based HPC system that will be installed at GW4.
> >
> >
> >
>
> Juan, if I may, I'll just try and add to that.
>
> Well, let's look a little bit at the 'nationalism' side of the issue, for a second. I have had a bit of discourse
> with the people in computing and engineering at Bath University, back when I was sharing some of my own research
> work (part of the consortium of four joined institutes of technology in the south west of Britain, referred
> to as 'GW4'). I found them, to be a really creative community of engineers and thinkers - sort of off the
> beaten track - away from the noise, hustle and bustle places like inner London suburbia.
>
> The thing that 'the public', or the people of the nation of Britain, or whatever you want to call - really,
> really want now out of public research - are ideas about what to do about Britain, and the matter that
> half the country floods now on a regular basis. It's the same here in Ireland, because the geography tends
> to be the same - large, flat, low areas with a lot of rivers and precipitation at times of the year like
> this one - Jan, Feb, March. Places just get washed away. And while, that is bad enough, the thing that
> really sticks in the throat of Joe Public in a place like Britain - is the lack of an advance warning.
> I.e. The lack of advanced warning, followed up by stuff, like inter-agency planning and emergency strategy.
> Honestly guys, some of the flooding events that we've seen in Britain in recent times, are just diabolical.
> The cost to business in town centers etc, that get gutted out by swollen rivers - is many multiples of
> what this super computer research investment is going to cost. The folk at the 'GW4' institutes, know
> all about it - because the reality of it is surrounding them all the time.
>
> So, to propose an alternative, to the idea that this 'Isambard' supercomputing project,
> is some kind of foolish, mis-guided nationalism (as in, the fools, don't they know
> that they ought to use Xeon instead?) - I'd propose an alternative.
>
> The weather that we get in Britain, starts a way, way far out in the Atlantic. And some of the really,
> really bad weather than we do get comes from places above north. Sure, we've got satellites that look down
> from the sky. That spelled the end of things that were called 'weather boats', that we used to use here
> in Ireland and in Britain (back in the 'Isambard' old days, when instruments were mechanical etc). What
> we're seeing really, with the 'GW4' project in Britain, is not really some mis-guided nationalism. Private
> companies like 'British Petroleum', BP, learned some really hard lessons lately, in places like the gulf
> of Mexico. They learned some very hard lessons in particular, about using senior electricians to try and
> debug, off-the-shelf hardware/software. Britain is a country, that knows a lot about navigation, about
> building things like nuclear submarines that can remain outside in the oceans for a long time. There's
> a mixture of private sector, public sector, domestic environmental crisis prevention policy - there's a
> mixture in short, of a lot of things happening with this Isambard supercomputer effort.
>
> It's legitimate, and fully legitimate and it's not silly. So basically, a country such as Britain has a
> lot of the basic components - the submarines, the weather forecasting satellites and the human personnel
> to make a lot of it tie up. However, they're a bit short on the computing aspect. That was what was shown
> up in recent disasters. This is what's really going on with the Cray supercomputer though, and one has to
> analyze it in that way - not to try to dismiss it as something silly to do with nationalism. The 'national'
> problem, that it's aimed at, is early warnings and sampling weather data at source, in order to manage these
> flooding events that happen on mainland Britain now, a lot better. And it's the same for other lowland mainland
> European countries now too, that all have have large towns, cities etc. Over and out.
The flooding problems in the UK reflect decades of neglect of a well established drainage system, driven by budget pressures and possibly by the desire of farmers to maximize crop surface.
Fixing the problem requires ditch and drainage maintenance, not supercomputers.
Incidentally, there is no evidence that forecasting would be greatly improved by a 10x faster computer. The models are still too deficient to be cured by simply using a tighter grid.
> juanrga (nospam.delete@this.juanrga.com) on January 29, 2017 4:42 am wrote:
> >
> > It seems you also missed the announcement of Isambard,
> > an ARM-based HPC system that will be installed at GW4.
> >
It's one of, if not the first serious, large(ish)-scale ARMv8 64-bit production machines. And it's the
> > first time Cray has explicitly announced an ARMv8 product meant for more than just prototyping.
> >
> >
>
> Juan, if I may, I'll just try and add to that.
>
> Well, let's look a little bit at the 'nationalism' side of the issue, for a second. I have had a bit of discourse
> with the people in computing and engineering at Bath University, back when I was sharing some of my own research
> work (part of the consortium of four joined institutes of technology in the south west of Britain, referred
> to as 'GW4'). I found them, to be a really creative community of engineers and thinkers - sort of off the
> beaten track - away from the noise, hustle and bustle places like inner London suburbia.
>
> The thing that 'the public', or the people of the nation of Britain, or whatever you want to call - really,
> really want now out of public research - are ideas about what to do about Britain, and the matter that
> half the country floods now on a regular basis. It's the same here in Ireland, because the geography tends
> to be the same - large, flat, low areas with a lot of rivers and precipitation at times of the year like
> this one - Jan, Feb, March. Places just get washed away. And while, that is bad enough, the thing that
> really sticks in the throat of Joe Public in a place like Britain - is the lack of an advance warning.
> I.e. The lack of advanced warning, followed up by stuff, like inter-agency planning and emergency strategy.
> Honestly guys, some of the flooding events that we've seen in Britain in recent times, are just diabolical.
> The cost to business in town centers etc, that get gutted out by swollen rivers - is many multiples of
> what this super computer research investment is going to cost. The folk at the 'GW4' institutes, know
> all about it - because the reality of it is surrounding them all the time.
>
> So, to propose an alternative, to the idea that this 'Isambard' supercomputing project,
> is some kind of foolish, mis-guided nationalism (as in, the fools, don't they know
> that they ought to use Xeon instead?) - I'd propose an alternative.
>
> The weather that we get in Britain, starts a way, way far out in the Atlantic. And some of the really,
> really bad weather than we do get comes from places above north. Sure, we've got satellites that look down
> from the sky. That spelled the end of things that were called 'weather boats', that we used to use here
> in Ireland and in Britain (back in the 'Isambard' old days, when instruments were mechanical etc). What
> we're seeing really, with the 'GW4' project in Britain, is not really some mis-guided nationalism. Private
> companies like 'British Petroleum', BP, learned some really hard lessons lately, in places like the gulf
> of Mexico. They learned some very hard lessons in particular, about using senior electricians to try and
> debug, off-the-shelf hardware/software. Britain is a country, that knows a lot about navigation, about
> building things like nuclear submarines that can remain outside in the oceans for a long time. There's
> a mixture of private sector, public sector, domestic environmental crisis prevention policy - there's a
> mixture in short, of a lot of things happening with this Isambard supercomputer effort.
>
> It's legitimate, and fully legitimate and it's not silly. So basically, a country such as Britain has a
> lot of the basic components - the submarines, the weather forecasting satellites and the human personnel
> to make a lot of it tie up. However, they're a bit short on the computing aspect. That was what was shown
> up in recent disasters. This is what's really going on with the Cray supercomputer though, and one has to
> analyze it in that way - not to try to dismiss it as something silly to do with nationalism. The 'national'
> problem, that it's aimed at, is early warnings and sampling weather data at source, in order to manage these
> flooding events that happen on mainland Britain now, a lot better. And it's the same for other lowland mainland
> European countries now too, that all have have large towns, cities etc. Over and out.
The flooding problems in the UK reflect decades of neglect of a well established drainage system, driven by budget pressures and possibly by the desire of farmers to maximize crop surface.
Fixing the problem requires ditch and drainage maintenance, not supercomputers.
Incidentally, there is no evidence that forecasting would be greatly improved by a 10x faster computer. The models are still too deficient to be cured by simply using a tighter grid.
