Re: [boost] [compute] Review period starts today December 15, 2014, ends on December 24, 2014

Subject: Re: [boost] [compute] Review period starts today December 15, 2014, ends on December 24, 2014
From: Paul A. Bristow (pbristow_at_[hidden])
Date: 2014-12-28 07:48:57

> -----Original Message-----
> From: Boost [mailto:boost-bounces_at_[hidden]] On Behalf Of Antony
Polukhin
> Sent: 15 December 2014 06:58
> To: boost_at_[hidden] List; boost-announce_at_[hidden]; boost-
> users_at_[hidden]; Kyle Lutz
> Subject: [boost] [compute] Review period starts today December 15, 2014, ends
on
> December 24, 2014
> 
> Dear All,
> 
> Review of the Compute library starts today on Mon 25st of December 2014 and
will
> last for ten days.
> 
> 
> The Compute library provides a C++ interface to multi-core GPGPU and CPU
> computing platforms based on OpenCL.

> Please, answer the following questions in your review:
> 
> 1. What is your evaluation of the design? Un-qualified to judge.
> 
> 2. What is your evaluation of the implementation?  Un-qualified to judge.

(I note some worrying about floating-point - is this not over-expectation ?

 "Die ganzen Zahlen hat der liebe Gott gemacht, alles andere ist Menschenwerk."
Integers are made by the Good Lord, all others are man's work.
 -- Leopold Kronecker

We should never expect to get exactly the same results from hardware C++ as
software GPU FP?

> 3. What is your evaluation of the documentation?  Very good.
> 
> 4. What is your evaluation of the potential usefulness of the library?
Invaluable.
> 
> 5. Did you try to use the library?  No.
> 
> 6. How much effort did you put into your evaluation? A glance? 

> 7. Are you knowledgeable about the problem domain?  Barely.
> 
> And finally, every review should answer this question:
> 
> 8. Do you think the library should be accepted as a Boost library? 

Yes because

I think Kyle has shown the feasibility and utility of this library and has some
real-life users.
I trust him to maintain and develop the library in future.

The C++ and GPU is fast moving, so I don't think we should be too fixated on
finding the ideal solution.
There will be more attempts to add GPU support to Standard C++ itself.
Obviously, the more we can allow other platforms than OpenCL to be added later
the better.
Like Boost.Sort, I think we should use the library as a container for various
tools to use GPUs.

I get the impression that the existing setup does not close too many doors too
firmly.

So I feel it is time to let it hit the streets.

Paul

---
Paul A. Bristow
Prizet Farmhouse
Kendal UK LA8 8AB
[phone]