Re: [boost] [compute] Review

Subject: Re: [boost] [compute] Review
From: Kyle Lutz (kyle.r.lutz_at_[hidden])
Date: 2014-12-31 01:08:14

On Tue, Dec 30, 2014 at 9:35 PM, Gruenke,Matt <mgruenke_at_[hidden]> wrote:
> -----Original Message-----
> From: Boost [mailto:boost-bounces_at_[hidden]] On Behalf Of John Bytheway
> Sent: Tuesday, December 30, 2014 23:14
> To: boost_at_[hidden]
> Subject: Re: [boost] [compute] Review
>
>> On 2014-12-30 21:46, Gruenke,Matt wrote:
>
>>> Async host <-> device copies only exist at the low level
>>> (command_queue).  They sidestep the issue by using raw pointers and
>>> size.
>>
>> I dispute that.  There's a copy_async to which you need not supply a
>> command_queue.  It seems high-level to me.
>
> Good catch.  I looked for it, but somehow missed it.  I fully agree that the constraint should be specified & enforced at compile-time.

I've added a static_assert() for this, see pull-request #404 [1].

-kyle

[1] https://github.com/kylelutz/compute/pull/404