Just a little bit. I registered a company, submitted a bid to a
government portal for research funding and... I work in a caravan shop
too. Hummmmmm...

Anyway

On 05/03/2010 09:13 PM, Bill Hart wrote:
> Hi Antony,
> 
> sounds like you've been busy!
> 
> When you are ready, just send me a link to what you have working
> (off-list if you like, until it's secure enough to make public).
> 
> I think everyone will wait until we get a website set up before
> working on bsdnt. We also need a devel list, which I'll set up around
> the same time.
> 

No pressure then. Ok, won't be long.

> We also have to sort out git on Windows. But Brian has some other
> things to do for the time being, so that won't happen immediately.
>

Sounds fair enough. My other project involving git on windows has died a
death because both of us are too busy (not related to MP Arithmetic at
all either, sadly).

> I did write a long paper setting out a "vision" for bsdnt. But I
> decided I didn't like it, so threw it away. In particular, me setting
> out a vision is like me saying I want to be "in charge". But I really
> don't want to take on that role. I'd like to be one of a small group
> of people responsible for setting out a vision for the project. For
> now, lots of people are either on holidays, busy with other things or
> unavailable for one reason or another. So let's focus on setting up
> the infrastructure first, then we'll use that to begin setting out a
> plan for 6 and 12 month targets.

Anything we specifically need on the website front for this? I write the
site in terms of static pages and modelled items, a model is just
something django turns into a row in a table for me (so I never have to
such SQL!!!). E.g:

class Contributor(models.Model):
    Name = models.CharField(max_length=200)
    Organisation = models.CharField(max_length=200, blank=True)
    RoleOrContribution = models.TextField()
    Homepage = models.URLField(blank=True)
    Repository = models.CharField(max_length=1000, blank=True)
    About = models.TextField(blank=True)

So if there is something that can help, I'm listening, however, it might
add to the time to live.

> 
> Oh, one other thing. I had a nice idea for getting MP arithmetic
> working on cuda. But I need some more time to think through the
> specifics of how it would work in practice. I probably need to write
> some code to try it out and see how feasible it is. More when I know
> what I am doing....

Really? Sounds interesting...! I started playing with cuda months ago
and that has died slightly. I'm told cudaMalloc, cudaMemcpy operations
etc aren't too fast, although I think the developer in question expected
parallelisation to produce near-lightspeed computing, which simply isn't
true. You need enough numbers to justify the overhead of going parallel
in the first place and when you get to that size, everything is going to
be slow!

Anyway, I'm interested! I was trying to figure out in my head how you'd
write a "parallel interface" to the various different setups you might
encounter - like, if you have say four nehalem cores and two tesla
cards, what is the most efficient way of splitting operations across them?

More to the point, can you produce an efficient algorithm for roughly
allocating resources that doesn't add masses to the run time!

I've been getting quite familiar with Knuth's book Seminumerical
algorithms too. I take it to work and get funny looks when I read a bit
of it on my lunch break.

Antony

> 
> Bill.
> 
> On 3 May 2010 19:57, Antony Vennard <[email protected]> wrote:
>> Hi All,
>>
>> Quick update on the bsdnt website, mostly for Bill but keeping on list
>> anyway.
>>
>> I've begun uploading bits and pieces of the source. I won't give away
>> urls here as they're web facing, but, a nearly bug free authentication
>> system (including permissions, groups etc) is ready and up (I had to
>> write it for another project anyway - gotta fund things like this
>> somehow!).
>>
>> Anyway, the really important piece is I've uploaded the base template
>> which includes, most importantly, a navigation structure as a ul list
>> (some fancy css makes these into menus when done but we'll leave off
>> that for a while). So the question is, roughly, what am I missing?
>> Anything in particular?
>>
>> I'll probably dream something up - I had something in my head but it
>> kinda disappeared. I've spent the past week writing a bid so I've been
>> in that mode. Anyway, keep an eye on those folders Bill, they're going
>> to start filling up hopefully, quite quickly.
>>
>> A quick couple of questions:
>>
>> * Any geniuses (or appropriate plural of genius) in CSS/web design on
>> this list or known to members of this list want to take my html and make
>> it look really good? I can get so far, but these web devs do pretty
>> impressive stuff.
>> * What's the status of the bsdnt project?
>>
>> Antony
>>
>> p.s. I will get to mpir, I promise! Unfortunately I also have to earn a
>> living as well...
>>
>> --
>> You received this message because you are subscribed to the Google Groups 
>> "mpir-devel" group.
>> To post to this group, send email to [email protected].
>> To unsubscribe from this group, send email to 
>> [email protected].
>> For more options, visit this group at 
>> http://groups.google.com/group/mpir-devel?hl=en.
>>
>>
> 

-- 
You received this message because you are subscribed to the Google Groups 
"mpir-devel" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/mpir-devel?hl=en.

Reply via email to