[Bf-cycles] Boltzmann Initiative and Cycles?

Brecht Van Lommel brechtvanlommel at pandora.be
Wed Nov 18 01:34:48 CET 2015

Yes, my comment about OpenCL dying was overly dramatic, I'm just
wondering what will happen long term. I'm still cautiously optimistic
AMD and Intel will have a good implementation of OpenCL C++ at some
point, but that will probably take another year or so.

On Tue, Nov 17, 2015 at 8:14 AM, Sergey Sharybin <sergey.vfx at gmail.com> wrote:
> While tis is an interesting concept, will it be same rock-solid as bare CUDA
> itself or will have same weird issues as we've got with OpenCL kernels?
> That's more a rithorical questions, but something we should keep in mind.
> As for OpenCL -- to my knowledge Intel is currently very interested in it.
> Also AMD improved OpenCL quite a lot in the past year, so don't think OpenCL
> really dying.
> On Tue, Nov 17, 2015 at 2:24 AM, Brecht Van Lommel
> <brechtvanlommel at pandora.be> wrote:
>> This seems to be AMD's equivalent to the NVidia CUDA toolkit. The main
>> interesting thing for Cycles is that this can compile to both AMD GPUs
>> and NVidia GPUs (still using the CUDA toolkit underneath though).
>> It remains to be seen if using this is actually a good idea for
>> Cycles, there's a bunch of questions:
>> * What are the advantages over OpenCL, why use this instead of an open
>> standard?
>> * What about Intel GPUs? Do we need OpenCL for those anyway, and if so
>> why not use OpenCL for AMD too?
>> * Or is this a sign that OpenCL is slowly dying, and eventually we'll
>> have no choice but to use this?
>> * Will all CUDA features that we use be available and will performance
>> be equally good?
>> * Will this work on OS X? NVidia has their own extra CUDA driver
>> there, will AMD provide the same?
>> * Does it allow dynamic loading, so that a single Blender executable
>> works on all computers regardless of the GPU?
>> IF this can replace CUDA and OpenCL backends in Cycles then that could
>> simplify things, but it would be unfortunate to give up on OpenCL and
>> go with something proprietary instead.
>> I still hope the eventually there will be proper OpenCL support from
>> all vendors, or even better, native support for GPUs in LLVM Clang.
>> But so far the fragmentation isn't getting any better, with Apple
>> choosing Metal over Vulkan, NVidia not supporting OpenCL properly, and
>> OpenCL not providing enough control for optimal CPU performance.
>> On Mon, Nov 16, 2015 at 5:52 PM, Zauber Paracelsus <zauber at gridmail.org>
>> wrote:
>> > So, AMD has launched what they call the Boltzmann Initiative, with the
>> > apparent aim to create a compiler that allows CUDA to work on AMD
>> > graphics hardware.
>> >
>> > And I just have to wonder: what affect would this have on
>> > Blender/Cycles, especially in light of the relatively recent addition of
>> > the OpenCL split-kernel work?
>> >
>> >
>> > PS: Sorry for not providing a link, but doing so caused my original
>> > message to be blocked by the spam filters.
>> > _______________________________________________
>> > Bf-cycles mailing list
>> > Bf-cycles at blender.org
>> > http://lists.blender.org/mailman/listinfo/bf-cycles
>> _______________________________________________
>> Bf-cycles mailing list
>> Bf-cycles at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-cycles
> --
> With best regards, Sergey Sharybin
> _______________________________________________
> Bf-cycles mailing list
> Bf-cycles at blender.org
> http://lists.blender.org/mailman/listinfo/bf-cycles

More information about the Bf-cycles mailing list