Hi Ludo', 2011/3/30 <l...@gnu.org>
> So, getting back to Laurent’s initial message, I wonder how using > NVIDIA’s proprietary software fits in the Compile Farm’s “mission”. > I would consider it an incentive for free software developers to depend > on NVIDIA’s proprietary software stack and strengthen its de facto > monopoly. > I'm not familiar with this kind of theological issues, but here's my point : 1) OpenCL is an open standard with many implementations, some of which are not stable yet : closed source ones such as ATI Stream (for any SSE2+ CPUs and/or ATI GPUs, on Win/Lin), NVidia drivers (for their GPUs, Win/Lin), Intel OpenCL SDK (for any SSE4+ CPU on Win), Apple (Mac only, CPU and GPU), and open-source implementation(s) such as Clover (GPU ? on any Xorg-capable platform ?) 2) NVidia does not really have a monopoly here (I'm talking about OpenCL and not CUDA which of course *is* a private technology I have no wish to strenghen or get trapped into). And even more implementors are expected in the OpenCL field, such as PowerVR... 3) I believe it's in the users best interest to have free and open source projects like JavaCL/ScalaCL/JOGL that can work efficiently on any underlying OpenCL/OpenGL implementation (free or not), with the best possible performance on each specific hardware (yet with no vendor lock-in). Right now I'm struggling with 3), as I only have access to CPU implementations and low- to mid-range GPUs. Getting my (ssh-tunnelled-)hands on a stronger NVidia card (or ATI, for that matter) I'd be able to evaluate the real benefit of my higher-level utilities on tomorrow's low-cost hardware. Does it sound free-minded enough ? ;-) Cheers -- zOlive http://ochafik.com
_______________________________________________ Gcc-cfarm-users mailing list Gcc-cfarm-users@gna.org https://mail.gna.org/listinfo/gcc-cfarm-users
