Subject: Re: [boost] Flow-based programming library for Boost?
From: Julian Gonggrijp (j.gonggrijp_at_[hidden])
Date: 2012-12-17 14:17:36
Dave Abrahams wrote:
> on Sun Dec 16 2012, Julian Gonggrijp wrote:
>
>>> [...]
>>> concurrency might be completely wasted unless you're able to take
>>> advantage of pipelining.
>>
>> Isn't execution of linearly dependent operations always a pipeline?
>
> Yes, but pipelining doesn't help if the input period exceeds the
> pipeline latency.
I think I see your point and I agree.
>> Assuming that, isn't every [edge] in a dependency graph a pipeline by
>> itself?
>
> I suppose it's a degenerate pipeline, but I don't know what you're
> getting at.
I was meaning to hint that pipelining is a *necessity* whenever
there's such a thing as a dependency graph, so there would really be
no question of whether you can take advantage of it or not (you /have/
to). In retrospect I understand you meant that there's not always a
speed benefit in doing the pipelining concurrently. Now your statement
turns out to be a tautology. ;-)
>> Assuming that, isn't pipelining the entire purpose of any form of
>> computational composition?
>
> IMO the purpose of computational composition is to manage complexity
> through re-use.
Well spoken.
(This is a hidden agreement.)
>>> [...]
>>
>> [...]
>
> Yes.
I'll assume that means "no dispute here".
-Julian
PS: could you be so kind to mangle email addresses in your quote
attribution lines, or perhaps leave them out entirely?