Have fun with this one:
(defun do-leap-years (start end)
(do ((s start (+ s 1)))
((> s end))
(format t "~A: ~A~%" s (leap-year s))))
Name:
Anonymous2011-12-23 7:50
>>8
OpenCL and DirectCompute merely call CUDA functions. Both architectures mirrors CUDA to allow Nvidia to minimize their work.
I had SO much more trouble using OpenCL than CUDA. It has a million states that you need to set and things that can go wrong. CUDA is much more straightforward to use. Of course the tradeoff is that OpenCL code will be cross-platform for years to come, whereas CUDA code will only run on Nvidia cards.
>>12
Cross-platform compatibility is worth bothering a bit more with your programs.
Name:
Anonymous2011-12-23 17:35
>>12
They only call CUDA functions on NVidia cards, just because NVidia is full of lazy Jews. It's like saying "Intel's chip only call CPU functions" when you're talking about OpenGL 2.1 on their older chips
Did you mean "crawl"? ATI5970 beats shit out of GTX590, and it's so old that it's not even produced anymore.
We need more comptetitors on GPGPU market, IMHO. Because both of NVIDIA and AMD are lazy bastards. Also new opencl standard kind-of useless, but hey, it's khronos.
just curious, do you mean run 100 times on one machine, and it'll work sometimes, or run on 100 machines, and it works perfectly on only a few of them?
Name:
Anonymous2011-12-24 2:29
New GPUs will be integrated, even Intell started to make CPUs with on-die GPU:
In 2013, Intel’s Haswell architecture will have on-die GPUs that will rival today’s discrete offerings. Haswell is scheduled after Ivy Bridge and promises some hefty performance improvements. This isn’t the first time Intel has tried to rival modern discrete GPUs from AMD and Nvidia. In the past, Intel has proposed a discrete destop GPU codenamed Larrabee. Intel’s Larrabee was supposed to deliver an innovative GPU architecture and new development features. However, Larrabee has been pushed back and will only be for high performance server clusters.
Now Intel wants to bring serious performance to the table in the form of on-die GPUs. Intel’s 22nm Haswell architecture will utilize the same power-saving tri-gate 3D transistor technology as Ivy Bridge and will have dedicated vector coprocessor to assist the on-die GPU.
But, will Haswell be capable of delivering high end performance comparable to that of AMD’s HD6970 or Nvidia’s GTX 580? It will be interesting to see just how far Intel will be able to take Hawell’s on-die GPU performance and if it can keep up with future releases of AMD’s Llano on-die GPU solutions.